Health appraisal models in multiple sclerosis

11
HEALTH APPRAISAL MODELS IN MULTIPLE SCLEROSIS GREG ROBERTS 1 * and ALEXA K. STUIFBERGEN 2 1 Department of Educational Psychology, The University of Texas at Austin, Austin, TX, U.S.A. and 2 School of Nursing, The University of Texas at Austin, Austin, TX, U.S.A. Abstract—This study used multi-group path analysis to test the hypothesis that disability’s total eect on self-rated health diers for dierent-aged persons with multiple sclerosis (MS). Data (n = 806) col- lected as part of a larger study examining quality of life for persons with MS were used to construct and compare models of health appraisal for three age groups, 18 to 45 (n = 347), 46 to 60 (n = 339), and 61 to 78 (n = 120). The results suggest that while the same general model describes health appraisal in each age group, disability’s total eect on self-rated health is less in the older and middle-aged groups than in the youngest group. Further, disability’s direct eect, rather than the hypothesized indir- ect eects, accounts for the group dierences. The findings support the other work in this area suggesting that age may moderate the eect of ‘‘objective’’ health status on self-rated health. The gener- ally surprising finding was the ‘‘break point’’ for this eect — contrary to other research that suggests older persons may dier from other-aged individuals, in this sample of persons with MS, the youngest group diered from the others. These results are discussed in terms of MS specifically and health appraisal more generally. # 1998 Elsevier Science Ltd. All rights reserved Key words—chronic illness, multiple sclerosis, self-rated health, health appraisal, social comparison INTRODUCTION The way an individual appraises his or her health is related to a number of important health outcomes, including use of health care services (Weinberger et al., 1986) and recovery from a major medical event (Wilcox et al., 1996). Most notable, though, is the relationship between health appraisal and sub- sequent mortality. More than 15 large-scale studies over the past 20 years suggest a robust relationship between self-reported health level in older samples and the mortality trend in that group, even when other known and suspected correlates are controlled (see Mossey, 1995). While a causal role for self-rated health can not be inferred from these findings, the results do suggest a need for greater understanding of the fac- tors influencing the appraisal process (Idler and Kasl, 1991). Of particular interest is how the experi- ence of chronic illness and disability may influence self-rated health status across the life span (Blaum et al., 1994; Feinburg et al., 1985; Stewart et al., 1989; Tarlov et al., 1989). Previous research has focused almost exclusively on older samples, due, largely, to the fact that chronic illness becomes more common with age. The limited number of stu- dies that have included middle-aged and younger persons have used samples drawn from general populations, and comparisons of individuals with chronic illness across the life span have not been possible. There is some evidence that the factors aecting the appraisal process may dier across dierent chronic conditions, though very little research to date has taken a disease-specific approach (Blaum et al., 1994). The few chronic illnesses that have been a target of investigations of this type have relied on single-equation methods of analysis to iso- late single variables or sets of variables that relate in meaningful ways to health appraisal (Hays et al., 1996; Johnson and Wolinsky, 1993). Only recently have the analysis techniques necessary for integrat- ing these discrete variables into comprehensive models (i.e. multi-equation modeling) been applied to health appraisal generally (Johnson and Wolinsky, 1993), and there is virtually no research of this sort addressing the eects of age and chroni- city. This study investigated the multi-factor structure of health appraisal in a sample of persons with Multiple Sclerosis (MS), a progressive, often debili- tating illness of the central nervous system generally diagnosed between 20 and 40 years of age. The pri- mary question was whether age moderates the eect of disability on perceived health for persons with this disabling condition. Assuming such an eect, a second purpose was to isolate the source(s) of age’s moderating role. Details on these purposes and a discussion of the conceptual framework on which Soc. Sci. Med. Vol. 47, No. 2, pp. 243–253, 1998 # 1998 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0277-9536/98 $19.00 + 0.00 PII: S0277-9536(98)00080-X *Author for correspondence. 243

Transcript of Health appraisal models in multiple sclerosis

HEALTH APPRAISAL MODELS IN MULTIPLE SCLEROSIS

GREG ROBERTS1* and ALEXA K. STUIFBERGEN2

1Department of Educational Psychology, The University of Texas at Austin, Austin, TX, U.S.A. and2School of Nursing, The University of Texas at Austin, Austin, TX, U.S.A.

AbstractÐThis study used multi-group path analysis to test the hypothesis that disability's total e�ecton self-rated health di�ers for di�erent-aged persons with multiple sclerosis (MS). Data (n= 806) col-lected as part of a larger study examining quality of life for persons with MS were used to constructand compare models of health appraisal for three age groups, 18 to 45 (n= 347), 46 to 60 (n= 339),and 61 to 78 (n= 120). The results suggest that while the same general model describes health appraisalin each age group, disability's total e�ect on self-rated health is less in the older and middle-agedgroups than in the youngest group. Further, disability's direct e�ect, rather than the hypothesized indir-ect e�ects, accounts for the group di�erences. The ®ndings support the other work in this areasuggesting that age may moderate the e�ect of ``objective'' health status on self-rated health. The gener-ally surprising ®nding was the ``break point'' for this e�ect Ð contrary to other research that suggestsolder persons may di�er from other-aged individuals, in this sample of persons with MS, the youngestgroup di�ered from the others. These results are discussed in terms of MS speci®cally and healthappraisal more generally. # 1998 Elsevier Science Ltd. All rights reserved

Key wordsÐchronic illness, multiple sclerosis, self-rated health, health appraisal, social comparison

INTRODUCTION

The way an individual appraises his or her health is

related to a number of important health outcomes,

including use of health care services (Weinberger et

al., 1986) and recovery from a major medical event

(Wilcox et al., 1996). Most notable, though, is the

relationship between health appraisal and sub-

sequent mortality. More than 15 large-scale studies

over the past 20 years suggest a robust relationship

between self-reported health level in older samples

and the mortality trend in that group, even when

other known and suspected correlates are controlled

(see Mossey, 1995).

While a causal role for self-rated health can not

be inferred from these ®ndings, the results do

suggest a need for greater understanding of the fac-

tors in¯uencing the appraisal process (Idler and

Kasl, 1991). Of particular interest is how the experi-

ence of chronic illness and disability may in¯uence

self-rated health status across the life span (Blaum

et al., 1994; Feinburg et al., 1985; Stewart et al.,

1989; Tarlov et al., 1989). Previous research has

focused almost exclusively on older samples, due,

largely, to the fact that chronic illness becomes

more common with age. The limited number of stu-

dies that have included middle-aged and younger

persons have used samples drawn from general

populations, and comparisons of individuals with

chronic illness across the life span have not been

possible.

There is some evidence that the factors a�ecting

the appraisal process may di�er across di�erent

chronic conditions, though very little research to

date has taken a disease-speci®c approach (Blaum

et al., 1994). The few chronic illnesses that have

been a target of investigations of this type have

relied on single-equation methods of analysis to iso-

late single variables or sets of variables that relate

in meaningful ways to health appraisal (Hays et al.,

1996; Johnson and Wolinsky, 1993). Only recently

have the analysis techniques necessary for integrat-

ing these discrete variables into comprehensive

models (i.e. multi-equation modeling) been applied

to health appraisal generally (Johnson and

Wolinsky, 1993), and there is virtually no research

of this sort addressing the e�ects of age and chroni-

city.

This study investigated the multi-factor structure

of health appraisal in a sample of persons with

Multiple Sclerosis (MS), a progressive, often debili-

tating illness of the central nervous system generally

diagnosed between 20 and 40 years of age. The pri-

mary question was whether age moderates the e�ect

of disability on perceived health for persons with

this disabling condition. Assuming such an e�ect, a

second purpose was to isolate the source(s) of age's

moderating role. Details on these purposes and a

discussion of the conceptual framework on which

Soc. Sci. Med. Vol. 47, No. 2, pp. 243±253, 1998# 1998 Elsevier Science Ltd. All rights reserved

Printed in Great Britain0277-9536/98 $19.00+0.00

PII: S0277-9536(98)00080-X

*Author for correspondence.

243

this study was based are in the next sections of thispaper.

Health appraisal, social comparison, and chronicity

One of the more enduring ®ndings of the researchon self-rated health is the lack of substantial corre-lation between an individual's self-rating and

``more-objective'' health measures (by clinicians,etc.). The considerable research that has consideredthis discrepancy suggests several key mediating vari-

ables, and these can generally be grouped as eitherphysical, psychological, or social in nature (Angeland Cleary, 1984; Idler and Angel, 1990). Withinthe ``physical domain'', actual physical health pre-

dominates. Though less implicated in the appraisalprocess than one might expect (thus the ``gap''described above), actual health status is nonetheless

an important factor in predicting self-ratings ofhealth (Rakowski and Cryan, 1990). Depression isprominent among the psychological factors a�ecting

health appraisal. Considerable evidence suggeststhat depressive symptoms may mediate the e�ect ofphysical health status on subjective health (seeMossey, 1995 for a review), and at least one study

suggests that depression may be directly associatedwith the comparison process that presumablyunderlies health appraisal (VanderZee et al., 1995),

a ®nding that may be more prominent for older in-dividuals when compared to the middle-aged(Levko� et al., 1987). Within the social domain,

economic status has been a consistent predictor ofself-rated health status. Better educated, employed,urban dwellers generally view health more optimisti-

cally than others when similar levels of disabilityare considered (Blaum et al., 1994; Liang, 1986;Wan, 1976).A second major ®nding characterizing this litera-

ture concerns the moderating e�ect of age on healthappraisal. Although older age is generally con-sidered to be a time of decline and failing health, a

number of early investigators found that elderly in-dividuals tend to be more optimistic in their healthevaluations than younger persons (Maddox and

Douglas, 1973; Murray et al., 1982; Stoller, 1984).Research comparing older persons to middle-agedsamples has been less conclusive (Cockerham et al.,1983; Levko� et al., 1987).

The apparent salience of ``age'' in this area ofresearch may be due to the social comparative pro-cess individuals often engage in when asked to

evaluate a personal attribute (Festinger, 1954; Sulsand Miller, 1977). In the case of self-evaluatedhealth, same-aged persons appear to be a prominent

reference group (Maddox, 1962), and older individ-uals may adjust their health expectations accord-ingly. Age may not be the only dimension along

which comparisons are made, however. Individualswith serious illness may use ``more ill'' others as acomparison group to insure favorable assessmentsof health (A�eck and Tennen, 1991; Suls et al.,

1991; Wood et al., 1985; VanderZee et al., 1995).

Indeed, the use of downward comparison (Wills,1981) to maintain positive appraisals of health,despite the presence of serious illness, appears to be

a widely-employed coping strategy and may havean e�ect on subjective health assessments indepen-dent of the e�ects of objective physical health and

self-reported psychological distress (VanderZee etal., 1995).

Much of the research on downward comparisonhas been done in controlled settings or in situationswhere participants have access to groups that make

such comparisons possible (cancer wards in hospi-tals is an example). Under such circumstances, the®ndings in favor of its use are considerable. When

chronically ill individuals are considered as mem-bers of their communities, however, social factors

that are not a consideration in more controlled set-tings may predominate, making downward com-parison less likely. For instance, in cases where one

of several salient groups has a numerical advantage,the more prevalent group is likely to be most instru-mental even if such means a less favorable compari-

son (Singer, 1981; Tornblom et al., 1993). For theperson with a chronic illness who is living in the

community, it's likely that the group of relativelyhealthy individuals will constitute the majority, ifnot the overwhelming majority, and, as a result, be

the key comparison group. A subset of similar-agedindividuals may also be selected from this largergroup for the reasons discussed previously.

A social comparative process may also shape thevariables thought to in¯uence levels of self-rated

health in individuals with chronic illness (i.e. social-economic and psychological factors). For instance,individuals may appraise their economic standing

based on the relative status (or perceived status) ofproximate, similar-aged persons (Sheeran et al.,1995). Because individuals with a chronic illness are

likely to be less fully employed (Minden et al.,1993) and more seriously in debt due to medical ex-penses (Minden et al., 1993), comparisons with non-

disabled and more ®nancially advantaged othersmay be less than productive (Sheeran et al., 1995).

Cognitive theories of depression also featuresocial comparison as a chief mechanism in under-standing depressive symptoms (e.g. Allbright et al.,

1993; DeVillis et al., 1990; Flett et al., 1987;Heidrich and Ry�, 1993; Swallow and Kuiper,

1990, Swallow and Kuiper, 1992, Swallow andKuiper, 1993; Weary and Edwards, 1994). Thiswork suggests that comparison with others operates

along a number of continua, including health status(Lewisohn et al., 1985) and economic status(Sheeran et al., 1995) and that an individual's level

of depression may stem partly from perceptions inthese respects.

Social comparison, then, may operate at pointsthroughout the appraisal process (physical, social-economic, psychological). The comparative process

G. Roberts and A. K. Stuifbergen244

may also di�er across the life span. Indeed, if agemoderates the relationship of disability and self-

rated health status, as has been suggested, agedi�erences may also be present in the social com-parative process thought to underlie this phenom-

enon.

Multiple sclerosis and self-rated health across the lifespan

A ®rst step in addressing the ``age-e�ects ques-tion'' is to assemble a general or baseline model of

health appraisal for persons with MS regardless ofage (Byrne, 1995). Figure 1 represents the startingpoint for this process. This model suggests thathealth appraisal in persons with MS is in¯uenced

largely by the same factors that operate in moregeneral populations (i.e. the psychological andsocial-economic domains suggested earlier).

Accordingly, direct e�ects are anticipated for thephysical aspects of health appraisal, as well as forthe social-economic and the psychological factors.

As suggested earlier, though the impact of physicalhealth factors (conceptualized here as disability dueto MS and comorbidity) is less than one might

expect, it nonetheless explains a considerableamount of the variance in self-rated health status.For persons with MS, health ratings are likely to beless optimistic than those of more-abled individuals.

This group is generally more disabled than the com-munity in general, and such is likely to be a salientfactor in the health appraisal process.

The direct e�ect for the social-economic factorre¯ects the considerable relationship between econ-

omic adequacy and actual health status (Campbell

et al., 1976). Persons in lower economic strata tend

have more health problems and less access to health

care, often resulting in lower actual health status as

well as lower self-rated health. Because MS-related

disability can reduce a family income by as much as

90% (Inman, 1984), the economic impact, in ``real

terms'', can be considerable, suggesting that this

group may be especially vulnerable. For persons

with MS, a second consideration in this respect is

the debilitating e�ect of the underlying comparative

process. As suggested earlier, persons with MS are

not likely to fare well when comparing the ade-

quacy of their resources to the resources of others

in their community, and the results of this compara-

tive process, based on a cognitivist view, may pre-

dispose this group to greater levels of depression,

and, presumably, lower self-ratings of health.

Social-economic factors, then, may a�ect self-

rated health status in two qualitatively distinct

ways. First, a direct e�ect may be evident due pri-

marily to the less adequate health and less available

care experienced by persons who are lower in this

respect. The indirect e�ect re¯ects the often disap-

pointing outcome of the comparative processes of

persons with MS in the economic sphere and the

increased levels of depression and decreased levels

of self-rated health that are likely to result.

A direct e�ect of depression on health appraisal

is also speci®ed. As indicated earlier, this relation-

ship has been well researched in general popu-

lations. Individuals who are depressed

psychologically tend to be less optimistic when eval-

uating a number of personal attributes, including

Fig. 1. Model of health appraisal for persons with MS.

Health appraisal models in MS 245

health status (VanderZee et al., 1995). Althoughseveral aspects of depression in persons with MS

are yet to be adequately sorted out (Boyle et al.,1991; MvIvor et al., 1984), the bivariate relationshipof depressive symptoms and self-rated health is

likely to be as evident in this group as in more gen-eral populations.Finally, age is also treated as a continuous, ex-

ogenous variable in Fig. 1. Although the primaryquestion posed by this study relates to the moderat-ing e�ects of age on the relationship of disability

and self-rated health, age's e�ect on the variables inFig. 1 may di�er at points across the life span. Thatis, the direct and indirect e�ects of age within di�er-ent age groups may vary across the cohorts. Thus,

the paths from age to the other variables in themodel are included.

Age-group di�erences in health appraisal for personswith MS

Figure 1 is a ``general'' model in that it representsthe health appraisal process across the life span forpersons with MS. However, if age moderates thee�ect of disability on self-rated health, as suggested,

this model may be more appropriate for some agegroups than for others. The presence of an age-e�ect may also suggest group di�erences in the

social comparative process that is (presumably) fun-damental to health appraisal. There are reasons tobelieve such di�erences may exist. The likely ``gap''

between the general group of persons with MS andtheir comparison group has already been discussed.The argument here is that the ``width'' of this gap

may di�er at points across the life span, dependingon the degree to which di�erent-aged groups withMS diverge from their respective comparisongroups. If one age group of persons with MS, for

example, is more similar (within a given domain) tosame-aged, proximate peers than a second agegroup with MS is to theirs, the comparisons of the

former group are likely to be more favorable.In the case of self-rated health, it has already

been pointed out that older individuals may enjoy

an advantage over younger persons with MS due tothe relatively more disabled nature of their same-aged peers. That is, older persons with MS are lessdissimilar, health-wise, from their comparison

group than are younger individuals with and with-out MS. What has been less fully addressed is thepossibility of age di�erences in the social compari-

sons associated with the other areas re¯ected inFig. 1 (i.e. social-economic and psychological).Several possibilities warrant mention.

Age group di�erences in disability's indirect e�ectthrough social-economic and psychological factors.Path 3 in Fig. 1 represents the indirect e�ect of dis-

ability through social-economic and psychologicalfactors. As suggested earlier, disability's e�ect inthis respect may be largely the result of the dis-couraging comparisons with same-aged, proximate

peers. Also, from a cognitivist view, the more dis-

couraging the results of such a comparison, the

greater the resulting level of depression and the

lower the level of self-rated health. Because middle-

aged persons with MS and their comparison group

are likely to be more economically dissimilar than

either older or younger individuals with MS and

their proximate same-aged peers (middle-aged per-

sons without MS are likely to be in their peak earn-

ing years), they may be more impacted by this

indirect e�ect of disability.

While the person with MS of any age may judge

himself or herself as ``lower'' in this respect (i.e.

economically) than his or her same-aged and more

healthy peer, and, accordingly, experience greater

levels of depression, the degree of such a discre-

pancy, and presumably the depth of the associated

depression, is likely to be less for the younger and

older groups than for the group of middle-aged in-

dividuals with MS. However, this may be the case

only when economic factors are considered in the

appraisal process. A di�erent pattern of age-di�er-

ences may be evident when disability's indirect

e�ect through depression alone (path 4 in Fig. 1) is

considered.

Age group di�erences in disability's indirect e�ect

through depression. Because MS is a progressive ill-

ness and disability is likely to worsen with age, one

would expect older individuals to be more disabled,

more depressed, and to report lower levels of self-

rated health. However, when a multi-equation

approach is taken, and speci®cally when the poten-

tially moderating e�ect of age is considered, the

assumptions suggested by a bivariate approach are

less tenable. The earlier-cited study by VanderZee

et al. (1995) suggests that increasing levels of dis-

ability may be associated directly with a need for

social comparison in the health domain (see also

Swallow and Kuiper, 1987; Swallow and Kuiper,

1990; Swallow and Kuiper, 1992). As an individual

becomes increasingly ill or disabled, he or she is

more likely to socially compare in an e�ort to man-

age and/or reduce levels of depression.

The VanderZee et al. (1995) study did not include

age as a variable, so it does not entirely address the

present discussion. However, if one considers that

disability due to MS increases with age, then older

persons with MS may be more likely to engage in a

social comparative process. Further, because older

individuals with MS appear to enjoy an advantage

in this respect for the reasons outlined earlier, the

results of their more frequent comparisons will be

more optimistic than the comparisons of other-aged

groups with MS. The moderating e�ect suggested

for age may be the result not only of the more

favorable conditions experienced by older folks, but

also the more frequent comparisons made by this

group.

G. Roberts and A. K. Stuifbergen246

Age group di�erences in the direct e�ect of disabil-ity. Finally, along with possible age-group di�er-

ences in the indirect e�ects of disability, groupdi�erences in its direct e�ect (path 1 in Fig. 1) mayalso be evident. Indeed, this path perhaps most

directly re¯ects the moderating e�ect of age thathas been discussed throughout this paper. In theabsence of the other e�ects that may mediate such

age di�erences, this relationship is essentially thebivariate correlation that has been described by anumber of researchers and serves as the bedrock of

much of the research in this area.

Research questions

In summary, two primary questions were

addressed by this study. First, ``Does age moderatedisability's e�ect on self-rated health status?'', and,second, ``Assuming such a moderator e�ect, at

what point(s) in the appraisal process do age di�er-ences emerge?''

METHOD

Sample

The data used in this study were collected as partof a larger study examining the impact of multiplesclerosis on quality of life (Stuifbergen, 1997).Participants were recruited through two Texas

chapters of the National Multiple Sclerosis Society.A total of 936 provided their name and address tothe investigators and indicated their willingness to

receive information about the study. Packets con-taining a questionnaire booklet, letter of infor-

mation (informed consent) and a postage paidreturn envelope were sent to all potential respon-dents. A reminder (postcard) was delivered to those

failing to return the survey within four weeks, anda second questionnaire was sent four weeks later incases where the postcard failed to produce a re-

sponse. A total of 834 questionnaires were returned(91%), of which 807 were usable, resulting in a ®nalresponse rate of 88%. From this total sample, the

806 participants between ages 18 and 78 wereselected for this study (mean of 48).Most participants were Anglo (93%) and female

(80%). The average length of diagnosis was

10.6 years. The average years of education was14.25, and 84% have at least a high school diploma(36% have a college degree or higher). Most

respondents were married (71%), and a total of 197(25%) reported living in a nonmetropolitan or ruralarea. About one-third (34%) reported being unem-

ployed due to their disability, while 30% wereworking full-time for pay. Sample characteristicsare summarized in Table 1. Population estimates on

these same variables for individuals with MS arealso included in this table (Minden et al., 1993).

Measures

Self-rated health. The four-item self-rated health(SRH) subindice of the Multilevel Assessment

Instrument (Lawton et al., 1982) was used tomeasure self-rated health. The four items compris-ing the SRH are items that have been widely used

elsewhere (Lawton et al., 1982), including the globalsubjective health item found on most health surveys(``How would you rate your overall health at thepresent time?''). The range of possible scores on the

SRH is four to 13. In the present sample, scoresranged from four to 13, and an alpha coe�cient of0.76 was obtained.

Functional disability due to MS. The incapacitystatus scale (Kurtzke, 1981) was used to measurethe level of disability due to MS. The incapacity

status scale (ISS) provides an objective measure ofthe degree to which an individual is unable to per-form his or her usual roles and activities. The ISShas been endorsed by an international group of

experts on MS. Evidence supporting its validity hasbeen presented by Kurtzke (1981).A single item from the ISS was used to measure

comorbidity: ``With regard to your physical healthÐ do you have any medical conditions (heart dis-ease, ulcers, diabetes) that require treatment: (0) no

signi®cant medical conditions (1) medical conditionsrequiring maintenance medication, but monitoringfrom physician is not required more often than

every 3 months (2) medical conditions requiring oc-casional monitoring by a physician or nurse, moreoften than every 3 months, but less often thanweekly (3) medical conditions requiring regular (at

Table 1. Comparison of study sample with characteristics of MSpopulation

Variable Study sample Population estimate

Age18±44 39% 35%45±60 46% 37%60±75 15% 23%75+ NA 5%

Gender 80% female 73% femaleDuration of Illness

Under 1 year 9% 12%1±5 years 25% 21%Over 5 years 66% 67%

Educational attainmentHigh school

graduate59% 60%

College graduate 36% 23%Employment status

In labor force 33% 29%Marital status

Married 74% 85%Separated 2% 2%Divorced 15% 4%Widowed 3% 4%Never married 6% 5%

EthnicityCaucasian 94% 94%African-American 4% 5%Hispanic 2% 1%

Self-rated health statusa

Excellent or good 50% 44%Fair or poor 50% 56%

aMeasured using the global-health item (``How would you rateyour overall health at the present time?'').

Health appraisal models in MS 247

least weekly) attention from a physician or nurse(4) medical conditions requiring daily attention by a

physician or nurse''. The sum of the remaining 15items were used as the measure of disability due toMS. The range of possible values on this modi®ed

scale was 0 to 60. In this sample, the total scoresranged from 0 to 48. An alpha coe�cient of 0.87was obtained.

Depressive symptomology. To measure levels ofdepressive symptomology, an abbreviated version ofthe Center for Epidemiological Studies Depression

Scale (Andersen et al., 1994) was used. Like thelonger edition, the ten-item version of the CESDuses a four-point scale ranging from zero (``none ofthe time'') to three (``most of the time''). Respon-

dents are asked to indicate how often in ``the pastweek'' they experienced each of the ten descriptionsof depressed mood (e.g. ``I was bothered by things

that usually don't bother me''). Andersen et al.(1994) report that the CESD-10 accurately predictsscores on the longer version of the scale, and it has

shown good test-retest reliability (Andersen et al.,1994). An alpha coe�cient of 0.86 was obtained forthis sample. The ten-item scale has a possible score

range of zero to 30, and the scores for the presentsample spanned this entire range.Perceived economic adequacy. The eight-item

Economic Adequacy Scale (Lobo, personal com-

munication, 1992) was used to assess the degree towhich participants felt their resources were ade-quate to meet their economic needs. Because this

self-report measure allows the respondent to deter-mine what ``adequate'' is, it re¯ects the social com-parative process that was described earlier as being

instrumental in this aspect of the health appraisalprocess for persons with MS. A sample item is``Does your family income allow you to meet yourdaily needs for living?'', to which participants are

asked to indicate ``not at all'', ``less than ade-quately'', ``adequately'', or ``more than adequately''.The possible range of scores is eight to 32. In this

sample, the scores ranged from eight to 32. Thealpha coe�cient was 0.94.

Analyses

The sample was divided into three groups. The``young'' group (n = 347) was aged 18 to 45.``Middle-aged'' individuals were 46±60 years of age

(n= 339), and the ``older'' group ranged from 61to 78 (n= 120). These categories re¯ect those used

by a number of the studies on self-rated health.Also, because age di�erences in social comparisonare thought to largely underlie health appraisal, the

age groups re¯ect stages in the life span tradition-ally associated with di�erences in these respects.Inter-correlations amongst the key variables were

computed for the total group. Means and standarddeviations for all of the measures in the study werecalculated for the three age groups. Analyses of var-

iance (ANOVA) were used, along with post-hoccomparisons, to determine if di�erences existedamong the observed means.The research questions were addressed using

multi-group structural equation modeling, followingprocedures described by Bollen (1989) and Byrne(1995). This approach involved comparing two

models at a time: one with the groups constrainedas equivalent on the parameters of interest and asecond, less restrictive model, in which these par-

ameters were free to take on any value (Byrne,1995). The competing models were tested using thedi�erence in their respective w2 values (Dw2). A sig-

ni®cant Dw2 indicated nonequivalence of groups,and provided a basis for evaluating the researchquestions addressed in this study. A ®rst step wasto ®t a general model of self-rated health to the

three age groups (Byrne, 1995). The modeldescribed in Fig. 1 was used as the starting pointfor this task. Path coe�cients and e�ect sizes were

estimated using LISREL 8.14 (Joreskog andSorbom, 1996).

Results

Intercorrelations of key variables are in Table 2.

A number of these coe�cients, when considered interms of their relative e�ects (Cohen, 1977), weremedium to large in size, ranging in value from

ÿ0.55 (disability and self-rated health) to ÿ0.31(comorbidity and self-rated health). Several rep-resented smaller-sized e�ects (Cohen, 1977), most

notably the correlation of age with depression(r=ÿ 0.11, p < 0.01), age with self-rated health(r=ÿ 0.09, p < 0.05), and age with comorbidity

(r= 0.14, p< 0.01).Table 3 includes the group means and standard

deviations for the instruments used in this study.

Table 2. Intercorrelations of key variables

Age CESD ECO SRH ISS Com

Age 1.0000CESD ÿ0.1135** 1.000ECO 0.0664 ÿ0.4125** 1.000SRH ÿ0.0869* ÿ0.4996** 0.2867** 1.0000ISS 0.2805* 0.3623** ÿ0.3319** ÿ0.5516** 1.0000Com 0.1372** 0.2005** ÿ0.1569** ÿ0.3052** 0.4300** 1.000

*p< 0.05; **p < 0.01.CESD = Depression; ECO = Economic adequacy; SRH = Self-rated health; ISS = Functional disability due to MS;

Com = Comorbidity level.

G. Roberts and A. K. Stuifbergen248

The three age groups di�ered signi®cantly from one

another in the level of disability due to MS

(F= 25.8, p< 0.01). Di�erences in levels of comor-

bidity were also present (F= 8.00, p< 0.01). While

the younger and middle-aged groups did not di�er

from one another, the older group reported a

greater prevalence of comorbid conditions. On the

CESD-10, the older group reported fewer depressive

symptoms than the younger (F= 4.40, p < 0.01),

while the middle-aged participants did not di�er

from the other two age groups in this respect.

There were marginal di�erences in the economic

adequacy of the three groups (F= 2.95, p < 0.05).

The older cohort indicated higher levels of ade-

quacy than the youngest group, while the middle-

age individuals did not di�er from the others. On

the self-rated health measure, there were no signi®-

cant di�erences between the age groups (2.60,

p = 0.07). This absence of a di�erence is in spite of

the signi®cantly higher levels of disability and

comorbidity in the older sample. It is evidence that

the ``age-e�ect'' observed in other samples may be

evident in a group of persons with MS, as well.

Further evidence of such was provided by the

comparison of the health appraisal models for the

di�erent groups. The proposed model of health

appraisal in Fig. 1 ®t the data well when run simul-

taneously within the three age groups (w2=7.68,

df = 12; n.s.; GFI = 0.96). The parameter esti-

mates (unstandardized) for this multi-group baseline

model are in Fig. 2. When analyzed in this way,

several age-group di�erences in the speci®ed paths

were suggested. For example, the middle-aged

group appears to di�er signi®cantly from the other

two in the e�ect of comorbidity on self-rated health.

While this path was not signi®cant for the middle-

aged cohort (b=ÿ 0.20), it was in the other two

groups (b=ÿ 0.298 and b=ÿ 0.364, respectively).

Table 3. Mean group di�erences on key variables

Variable Mean F p

young middle older

CESD 11.5a 10.5a,b 9.5b 4.40 <0.01ECO 2.9a 3.0a,b 3.1b 2.95 <0.05SRH 7.3 7.2 7.2 2.6 0.07ISS 15.2a 17.8b 22.6c 25.8 <0.01Com 0.77a 0.77a 1.02b 8.0 <0.01

Means with same letters in superscript do not signi®cantly di�er.CESD = Depression; ECO = Economic adequacy; SRH = Self-rated health; ISS = Functional disability due to MS;

Com = Comorbidity level.

Fig. 2. Group speci®c path coe�cients for baseline model. *zr1.96 (i.e. p< 0.05). NOTE: Coe�cientsare unstandardized. An estimate for each of the three groups is provided for each parameter; the ®rst

value represents the youngest group and the last value represents the oldest of the three groups.

Health appraisal models in MS 249

Similarly, the older and younger groups were simi-

lar in the e�ect of economic adequacy on self-ratedhealth (b =ÿ 0.158, n.s. and b =ÿ 0.167, n.s., re-spectively), while the middle-aged group appeared

to di�er signi®cantly (b= 0.391, p < 0.01).There also appears to be a downward linear

trend in the groups' coe�cient for both the direct

e�ect of disability on self-rated health (b =ÿ 0.104,b =ÿ 0.050, and b=ÿ 0.039, respectively) as well

as the total e�ect (b=ÿ 0.131, b=ÿ 0.083, andb =ÿ 0.056). To further explore this latter possi-bility, the total e�ect (all indirect and direct paths)

of disability due to MS on health appraisal wasconstrained to be equal across groups. This model(w2=37.5, df = 24, p= 0.03; GFI = 0.96) rep-

resents a signi®cantly less adequate ®t than themulti-group baseline model (Fig. 1). The Dw2 value

(Dw2=29.82, Ddf = 12, p < 0.01) suggests that thetotal e�ect of disability was not the same across thegroups, and serves as additional evidence that age

moderates the e�ect of disability on health apprai-sal for persons with MS (Byrne, 1995).To isolate the source or sources of these group

di�erences, and to answer the second research ques-tion, four models were run. The ®rst tested the

equivalence of the indirect path through economicadequacy along with the direct path from disabilityto health appraisal (paths 1 and 2 in Fig. 1). When

these were constrained equal across the groups, the®t was marginally acceptable (w2=31.2, df = 18,p = 0.03; GFI = 0.97), although when compared to

the baseline model, it was signi®cantly less adequate(Dw2=23.5, Ddf = 6, p< 0.01), suggesting that the

groups were not equivalent in this respect. Whenthe indirect path through depression along with thedirect e�ect of disability (paths 1 and 4) were con-

strained, a similar result was found. The model(w2=24.5, df = 18, p = 0.14; GFI = 0.98) was asigni®cantly less acceptable ®t than the baseline

model (Dw2=16.8, Ddf = 6; p< 0.01), suggestingagain the nonequivalence of the three groups.

The direct e�ect of disability as well as its indir-ect e�ect through economic adequacy and de-pression (paths 1 and 3 in Fig. 1) were also

considered. When these parameters were set equalacross groups, this model ®t the data reasonablywell (w2=27.7, df = 20, p = 0.12; GFI = 0.97),

although, as before, it represented a signi®cantlyless adequate ®t when compared to the multi-group

baseline model (Dw2=20.02, Ddf = 8, p < 0.01).Once again, this result suggests a signi®cant degreeof nonequivalence across the groups.

The ®nal model constrained only the direct path(path 1 in Fig. 1) from disability to health apprai-sal. This model also di�ered signi®cantly from the

baseline (Dw2=14.62, Ddf = 2, p< 0.01). Whenthis model was compared to the previous models

however, no di�erences were evident. The absenceof signi®cant Dw2 value indicates that the more con-strained models did not di�er signi®cantly from this

last model in their degree of ®t, suggesting that thesource of age-group di�erences in health appraisal

may be due only to disability's direct e�ect.Finally, because the source of group di�erences

in total e�ect of disability appeared to be due only

to di�erences in the direct e�ect (b=ÿ 0.104,b=ÿ 0.05, and b =ÿ 0.039), two additionalmodels were speci®ed and compared to the baseline

model to determine where the age groups di�eredfrom one another. When the young and middle-aged groups were constrained to have equal direct

e�ects, and the older group was allowed to freelyestimate on this parameter, the model (w2=18.85,df = 13, p = 0.13) di�ered signi®cantly from thebaseline (Dw2=11.17, Ddf = 1, p< 0.01) suggesting

that these two groups di�ered signi®cantly in theirdirect e�ect on self-rated health. When middle-agedand older individuals were constrained to have

equal coe�cients for the direct e�ect (w2=8.11,df = 13, p= 0.84), there was no di�erence in w2

values (Dw2=0.43, Ddf = 1, p>0.05). Thus, the pri-

mary di�erence in groups was in disability's directe�ect for the younger group when compared to themiddle-aged and older samples.

DISCUSSION

Considerable literature in recent years suggeststhat di�erent-aged individuals rate health in di�er-ent ways. Little of this research, however, has

focused on middle-aged and younger samples, andvirtually none of it comes from a disease-speci®cperspective. The life span model of health appraisal

developed and tested in this study suggests that thedi�erence in the direct e�ect of disability dis-tinguishes the health-appraisal models of di�erent-aged individuals with the chronic, disabling con-

dition of MS. Like members of other groups, olderpersons with MS appear to have an advantage inthis respect over younger individuals with this ill-

ness. Even though more disabled, these older per-sons appear to ``protect'' their self-perceptions ofhealth by relying less on their level of disability

when making a health appraisal.Further, it appears that this di�erence is due pri-

marily to disability's direct e�ect, rather than itsvarious indirect e�ects on self-rated health. This

suggests the health appraisal processes of di�erent-aged persons with MS di�er primarily in the e�ectthat illness- and disability-related variables have on

self-rated health. While economic and psychologicalfactors are clearly implicated in the health apprai-sals of all three age-groups, the age groups do not

di�er in the degree to which these two factors facili-tate disability's e�ect. This also suggests that thesocial comparison process, as it relates to health

appraisal, may di�er primarily along a health-re-lated dimension.What is unexpected is the absence of a signi®cant

di�erence between middle-aged and older individ-

G. Roberts and A. K. Stuifbergen250

uals, in this respect. The reasons for this are notclear, although one possibility relates to the di�cul-

ties associated with the early stages of diagnosis.The pervasive emotional e�ect of coping with MSearly on may make it a more salient issue in the

appraisal process, even though there may be littleevident disability. Perhaps, only after an individualhas the opportunity to adjust and develop some

coping strategies does he or she become less in¯u-enced by disability status when making healthappraisals. This topic warrants additional study.

Another possibility for further research suggestedby this study concerns age's within-groups-e�ect ondepression. The ®ndings suggest the younger groupof persons with MS may become less depressed as

they move towards middle-age, while middle-agedand older individuals experience little change in thisrespect. This is consistent with much of the litera-

ture addressing depression and MS, and it alsohighlights the importance of separating the respect-ive e�ects of age, disability level, and length of

diagnosis on depression and self-rated health. Morework on these interrelationships in this populationis needed.

Limitations

Several comments concerning the study's design

are necessary. First, while the sample generallyre¯ects the population of persons with MS on mostkey variables, participants were solicited usingmembership rosters of MS societies, making it poss-

ible that the sample overrepresents people with amore active and concerned approach to their dis-ease. Also, because MS is less prevalent in more

temperate climates, the comparison process thatpresumably underlies health appraisal may operatedi�erently in southwestern parts of the U.S. when

compared to regions where greater numbers of indi-viduals with MS reside.There are also some problems associated with the

measurement of disability (Blaum et al., 1994; Idler,1993). There is little consensus on the best way tomeasure this variable and it seems reasonable thatdi�erent settings and perhaps di�erent types of dis-

ability will require di�erent approaches (Idler,1993). The disease-speci®city and extensive use ofthe ISS made it a reasonable choice in this case.

Finally, because cohort e�ects may play a role indi�erent levels of subjective health for di�erent agegroups, cross-sectional data, as this is, may not ade-

quately capture the phenomenon (see Idler, 1993for a discussion of this issue).

Implications

There is a need for more knowledge on the healthappraisals of persons with lifelong chronic con-

ditions. The apparent relationship of self-ratedhealth and mortality patterns suggests that knowl-edge about the former may provide insights aboutthe latter. This may be especially important for a

group like the one studied here. Individuals withchronic physical illness with no known cure and lit-

tle available treatment may bene®t from interven-tions that are aimed at more psychosocial aspectsof the disease. While disability's direct e�ect on

self-rated health may be relatively di�cult to in¯u-ence, health-care providers and others may havemore success in addressing disability's indirect

e�ects. For instance, depression and the social com-parative process it implies are powerful mediatorsof disability's e�ect, regardless of age. By aiming

interventions in this direction, the individual withMS may experience improved levels of psychologi-cal functioning, resulting in increased perceptions ofhealth.

A focus on the indirect e�ects of disability mayalso may help to minimize the disability experiencedby an individual, or at least his or her ``perception''

of the disability. While the course of this illness ispresently beyond control, the way a person withMS responds to his or her illness is not. By increas-

ing the availability of psychological copingresources, the individual with MS may be less likelyto play the social role of ``sick person'' and, as a

result, experience less disability (Brooks andMatson, 1990; Hyman, 1975).Increasing resources in other respects may also be

important. The indirect e�ect of disability through

economic factors (alone as well as through de-pression) suggests that perceived and real inadequa-cies in this respect may minimize levels of self-rated

health. Support, ®nancial and otherwise, may be animportant factor in helping persons with MS ratetheir health more favorably, and, quite possibly, ex-

perience less severe levels of disability.

AcknowledgementsÐThis research was supported by grantR29NRO3195-03S1, National Institute of NursingResearch, National Institutes of Health andR29NRO3195-03S1, O�ce of Research on Women'sHealth, O�ce of the Director, National Institutes ofHealth. The authors would like to thank Da'LynnClayton, Ida Miller, and Sharon Rogers for their helpwith data collection and Dr. Heather Becker for her help-ful comments on an earlier version of this paper.

REFERENCES

A�eck, G. and Tennen, H. (1991) Social comparison andcoping with major medical problems. SocialComparison: Contemporary Theory and Research, ed. J.Suls and T. Willis, pp. 369±394. Lawrence ErlbaumAssociates, Hillsdale.

Allbright, J., Alloy, L., Barch, D. and Dykman, B. (1993)Social comparison by dysphoric and nondysphoric col-lege students: The grass isn't always greener on theother side. Cognitive Therapy and Research 17, 485±509.

Andersen, E., Malmgren, J., Carter, W. and Patrick,D. (1994) Screening for depression in well older adults:Evaluation of a short form of the CESD-10. AmericanJournal of Preventative Medicine 10, 77±84.

Health appraisal models in MS 251

Angel, R. and Cleary, P. (1984) The e�ects of social struc-ture and culture on reported health. Social ScienceQuarterly 65, 814±828.

Blaum, C., Liang, J. and Liu, X. (1994) The relationshipof chronic disease and health status to health servicesutilization of older Americans. Journal of the AmericanGeriatrics Society 42, 1087±1093.

Bollen, K. (1989) Structural equations with latent variables.John Wiley & Sons, New York.

Boyle, E., Clark, C., Klono�, H., Paty, D. and Oger,J. (1991) Empirical support for psychological typesobserved in multiple sclerosis. Archives of Neurology 48,1150±1154.

Brooks, N. and Matson, R. (1990) Social-psychologicaladjustment to multiple sclerosis. Social Science andMedicine 16, 2129±2135.

Byrne, B. (1995) One application of structural equationmodeling from two perspectives. Structural EquationModeling: Concepts, Issues, and Applications, ed. R.Hoyle, pp. 138±157. Sage Publications, Thousand Oaks.

Campbell, A., Converse, P. and Rogers, W. (1976) TheQuality of American Life. Russell Sage, New York.

Cockerham, W., Sharp, K. and Wilcox, J. (1983) Agingand perceived health status. Journal of Gerontology 38,349±355.

Cohen, J. (1977) Statistical Power Analysis for theBehavioral Sciences. Lawrence Erlbaum Associates,Hillsdale.

DeVillis, R., Holt, K., Renner, B. and Blalock, S. (1990)The relationship of social comparison to rheumatoidarthritis symptoms and a�ect. Basic and Applied SocialPsychology 11, 1±18.

Feinburg, S., Loftus, E. and Tanur, J. (1985) Cognitiveaspects of health survey methodology: An overview.Millbank Memorial Fund Quarterly 63, 547±566.

Festinger, L. (1954) A theory of social comparison pro-cesses. Human Relations 7, 117±140.

Flett, G., Vredenburg, K., Pliner, P. and Krames, L. (1987)Depression and social comparison information-seeking.Journal of Social Behavior and Personality 2, 473±484.

Hays, J., Schoenfeld, D. and Blazer, D. (1996)Determinants of poor self-rated health in late life.American Journal of Geriatric Psychiatry 4, 188±196.

Heidrich, S. and Ry�, C. (1993) Physical and mentalhealth in later life: The self system as mediator.Psychology and Aging 8, 327±338.

Hyman, M. (1975) Social-psychological factors a�ectingdisability among ambulatory patients. Journal ofChronic Disease 33, 199±204.

Idler, E. (1993) Age di�erences in self-assessments ofhealth: Age changes, cohort di�erences, or survivorship.Journal of Gerontology 48, S289±S300.

Idler, E. and Angel, R. (1990) Age, chronic health, andsubjective assessments of health. Advances in MedicalSociology 1, 131±152.

Idler, E. and Kasl, S. (1991) Health perceptions and survi-val: Do global health evaluations of health status reallypredict mortality. Journal of Gerontology 46, S55±S65.

Inman, R. (1984) Disability indices, the economic costs ofillness and social insurance: The case of multiple scler-osis. Acta Neurological Scandinavica 101, 46±55.

Johnson, R. and Wolinsky, F. (1993) The structure ofhealth status among older adults: Disease, disability,functional limitation, and perceived health. Journal ofHealth and Social Behavior 34, 105±121.

Joreskog, K. and Sorbom, D. (1996) LISREL 8: User'sReference Guide. Scienti®c Software International,Chicago.

Kurtzke, J. (1981) A proposal for a uniform minimalrecord of disability in multiple sclerosis. ActaNeurological Scandinavica 64, 110±129.

Lawton, M., Moss, M., Fulcomer, M. and Kleban,M. (1982) A research and service-oriented multilevel

assessment instrument. Journal of Gerontology 44, P61±P71.

Levko�, S., Cleary, P. and Wetle, T. (1987) Di�erences inthe appraisal of health between aged and middle-agedadults. Journal of Gerontology 42, 114±120.

Lewisohn, P., Hoberman, H., Teri, L. and Hautzinger, M.(1985) An integrative theory of depression. TheoreticalIssues in Behavior Therapy, ed. S. Reiss and R. Bootzin,pp. 331±359. Academic Press, San Diego.

Liang, J. (1986) Self-reported physical health among agedadults. Journal of Gerontology 41, 248±260.

Maddox, G. (1962) Some correlates of di�erences in self-assessment of health status among the elderly. Journalof Gerontology 17, 180±185.

Maddox, G. and Douglas, E. (1973) Self-assessment ofhealth: A longitudinal study of elderly subjects. Journalof Health and Social Behavior 14, 87±93.

MvIvor, G., Riklan, M. and Rezniko�, M. (1984)Depression in multiple sclerosis as a function of lengthand severity of illness, age, remissions, and perceivedsocial support. Journal of Clinical Psychology 40, 1028±1033.

Mille®orini, E., Padovani, A., Pozzilli, C., Loriedo, C.,Bastainello, S., Buttinelli, C., DiPiero, V. and Fieschi,C. (1992) Depression in the early phase of MS:In¯uence of functional disability, cognitive impairment,and brain abnormalities. Acta Neurologica Scandinavica86, 354±358.

Minden, S., Marder, W., Harrold, L. and Dor, A. (1993)Multiple Sclerosis: A Statistical Portrait. National MSSociety, New York.

Mossey, J. (1995) Importance of self-perceptions for healthstatus among older persons. Mental Health and Aging,ed. M. Goetz, pp. 124±162. American PsychologicalAssociation, Washington.

Murray, J., Dunn, G. and Tarnopolsky, A. (1982) Self-assessment of health: An exploration of the e�ects ofphysical and psychological symptoms. PsychologicalMedicine 12, 317±378.

Rakowski, W. and Cryan, C. (1990) Associations amongperceptions of health and health status within three agegroups. Journal of Aging and Health 2, 58±80.

Sheeran, P., Abrams, D. and Orbel, S. (1995)Unemployment, self-esteem, and depression: A socialcomparison theory approach. Basic and Applied SocialPsychology 17, 65±82.

Singer, E. (1981) Reference groups and social evaluations.Social Psychology: Sociological Perspectives, ed. M.Rosenberg and R. Turner, pp. 66±93. Basic BooksPublishers, New York.

Stewart, A., Green®eld, S. and Hays, R. (1989) Functionalstates and well-being of patients with chronic con-ditions: Results from the medical outcomes study.Journal of the American Medical Association 262, 907±913.

Stoller, E. (1984) Self-assessments of health by the elderly:The impact of informal; assistance. Journal of Healthand Social Behavior 25, 260±270.

Stuifbergen, A. (1997). Maximizing Health with MultipleSclerosis (NINR, Grant No. R29NRO3195). NationalInstitutes of Health, Washington, DC.

Suls, J., Marco, C. and Tobin, S. (1991) The role of tem-poral comparison, social comparison, and direct apprai-sal in the elderly's self-evaluations of health. Journal ofApplied Psychology 21, 1125±1144.

Suls, J. and Miller, R. (1977) Social Comparison Processes:Theoretical and Empirical Perspectives. Hemisphere,Washington.

Suls, J. and Wills, T. (1991) Social Comparison:Contemporary Theory and Research. Lawrence ErlbaumAssociates, Hillsdale.

Swallow, S. and Kuiper, N. (1987) The e�ects of de-pression and cognitive vulnerability to depression on

G. Roberts and A. K. Stuifbergen252

judgments of similarity between self and other.Motivation and Emotion 11, 157±167.

Swallow, S. and Kuiper, N. (1990) Mild depression, dys-functional cognitions, and interest in social comparisoninformation. Journal of Social and Clinical Psychology 9,289±302.

Swallow, S. and Kuiper, N. (1992) Mild depression andfrequency of social comparison behavior. Journal ofSocial and Clinical Psychology 11, 167±180.

Swallow, S. and Kuiper, N. (1993) Social comparison indysphoria and nondysphoria: Di�erences in target simi-larity and speci®city. Cognitive Therapy and Research17, 103±122.

Tarlov, A., Ware, J. and Green®eld, S. (1989) The medicaloutcomes study: An application of methods for moni-toring the results of medical care. Journal of theAmerican Medical Association 262, 925±930.

Tornblom, K., Stern, P., Pirak, K., Pudas, A. andTornlund, E. (1993) Type or resource and choice of tar-get. Resource Theory: Explorations and Applications. ed.G. Uriel, J. Converse, Y. Kjell and E. Foa. AcademicPress, New York.

VanderZee, K., Buunk, B. and Sanderman, R. (1995)Social comparison as a mediator between health pro-blems and subjective health evaluations. British Journalof Social Psychology 34, 53±65.

Wan, T. (1976) Predicting self-assessed health: A multi-variate approach. Health Services Research 1, 464±477.

Weary, G. and Edwards, J. (1994) Social Cognition andClinical Psychology: Anxiety, Depression, and theProcessing of Social Information. Lawrence ErlbaumAssociates, Hillsdale.

Weinberger, M., Darnell, J. and Tierney, W. (1986) Self-rated health as a predictor of hospital admission andnursing home placement in elderly public housingtenants. American Journal of Public Health 76, 457±459.

Wilcox, V., Kasl, S. and Idler, E. (1996) Self-rated healthand physical disability in elderly survivors of a majormedical event. Journals of Gerontology, PsychologicalSciences and Social Sciences 51, S96±S104.

Wills, T. (1981) Downward comparison principles in socialpsychology. Psychological Bulletin 90, 245±271.

Wood, J., Taylor, S. and Lichtman, R. (1985) Social com-parison in adjustment to breast cancer. Journal ofPersonality and Social Psychology 49, 1169±1183.

Health appraisal models in MS 253