Kiecolt-Glaser et al.[1]

8/7/2019 Kiecolt-Glaser et al.[1]

1/13

Chronic Stress and Immunity in Family Caregiversof Alzheimer's Disease VictimsJANICE K. KIECOLT-GLASER, PH D , R O N A L D GLASER, PHD,E D W I N C. S H U T T L E W O R T H , MD, C A R O L S. DYER , BA, P A U L A O G R O C K I , BS, ANDCARL E. SPEICHER, MD

Although acute stress has been associated with transient immunosuppression, little is knownabout the immunologic consequences of chronic stress in humans. In order to investigatepossible health-related consequences of a long-term stressor, we obtained blood samples forimmunologic and nutritional analyses and psychologic data from 34 family caregivers of Alz-heimer's disease (AD) victims and 34 sociodemographically matched comparison subjects.Family caregivers for AO victims were more distressed than comparison subjects without similarresponsibilities. Greater impairment in the AD victim was associated with greater distress andloneliness in caregivers. Caregivers had significantly lower percentages of total T lymphocytesand helper T lymphocytes than did comparison subjects , as well as significantly lowerhelper-suppressor cell ratios; caregivers also had significantly higher antibody liters to Ep-stein-Barr virus than did comparison subjects, presumably reflecting poorer cellular immunesystem control of the latent virus in caregivers. The percentages of natural killer cells andsuppressor T lymphocytes did not differ significantly. These data suggest that chronicallystressed AD family caregivers do not show immunologic or psychologic adaptation to the levelof their well-matched age peers.

There is good evidencethat acute stress-ful events are associated with adverse im-munologic changes in humans (1,2). Evenrelatively commonplace events like aca-demic examinations have been linked totransient changes in immunity; data fromblood samples taken during examinationsshow poorer cellular immunity than thosetaken 1 month to 6 weeks earlier, when

Received for publication August 25,1986; revisionreceived December 10,1986.

From the Departments of Psychiatry (J.K.K.-G.J,Medical Microbiology and Immunology (R.G.; C.S.D.;P.O.), Neurology (E.C.S.), and Pathology (C.E.S.), TheOhio State University College of Medicine, Colum-bus, Ohio.

Address reprint requests to : Janice K. Kiecolt-Glaser,Ph.D., Department of Psychiat ry, The Ohio State Uni-versity College of Medicine, 473 West 12th Avenue,Columbus, Ohio 43210, USA

students were less distressed (3-7). In ad-dition, individuals undergoing major novelstressful life changes such as marital sep-aration and divorce have a poorer prolif-erative response than well-matched com-parison subjects (8); bereaved spouses showpoorer mitogen responsiveness after thedeath of a spouse than before (9).There is some evidence from rodentstudies suggesting that the chronicity ofthe stressor mediates both immunologicresponses and tumor development. Datafrom one rodent study by Monjan and Col-lector (10) suggested that chronic stressmight lead to an enhancement of immunefunction. Using daily high-intensity inter-mittent noise, they found that the acute orshort-term consequence of the auditorystressor was immunosuppression, whilemore chronic stress appeared to result inenhanced mitogen responsiveness. Simi-

Psychosomatic Medicine 49:523-535 (1987)Copyright 1987 by the American Psychosomatic Society, Inc.Published by Elsevier Science Publishing Co., Inc.52 Vanderbill Avenue, New York, NY 1001 7

5230033-3174/87/S3.50


2/13

J. K. KIECOLT-CLASER et a l .

larly, Sklar and Anisman (11) found thattumor size and survival were adversely af-fected by a single session of inescapableshock in mice injected w ith a tumo r. How-ever, mice that underw ent ten daily shocksessions had tumor areas that were signif-icantly less than tho se of controls, and sur-vival times approximated those of the con-trols.Although these data suggest that theremay be adaptation or even enhancementof immunity in response to a more chronicstressor, there are problems in extrapolat-ing from research with rodents to hum ans.In particular, adaptation may be differentbecause the nature of the stressor is quitedifferent; physical stressors like intermit-tent loud noise, rotational stress, or elec-tric shock are used in roden t research, andadaptation to physical stressors could bequite different than adjustment to the cog-nitive stressors that are of primary interestin research with humans. Use of familycaregivers for Alzheimer's disease (AD)victims provides an opportunity to ex-amine the immunologic and psychologicconsequences of a more chronic psycho-social stressor.

AD affects two million older adults inthis country. The progressive cognitiveimpairments that are characteristic of ADlead to increasing needs for supportive careof afflicted individuals. Although mildmemory impairm ents may be the only ob-vious problem in the early stages, the ir-reversible deterioration of brain tissueeventually culminates in profound cog-nitive and behavioral changes, includingdisorientation, incontinence, and an in-ability to provide any self-care (12, 13).Since the modal survival time after onsetis approximately 8 years (14), long-termcare of these patients by family membersmay be conceptualized as a chronic stressor

(15). The majority of AD patients live inthe community, under the care of their rel-atives (13).Cross-sectional data from several labo-ratories suggest that the s tresses of AD car-egiving leave family members at high riskfor depression (16, 17). Moreover, theremay be some progressive deterioration incaregivers' well-being over time related tothe increasing impairment in the AD fam-ily member. George and Gwyther (18) foundsubstantial deterioration in caregivers' well-being when measures w ere taken at 1-yearintervals, including perceived decremen tsin health, decreased satisfaction with theamount of time available for social partic-ipation, decreased life satisfaction, and in-creased levels of stress-related psychiatricsymptoms. These changes were particu-larly noteworthy because the baseline (timeone) levels of well-being in these caregiv-ers were already quite low in absoluteterms.

Although limited evidence suggests thatcaregiving responsibilities may be asso-ciated with self-reported health impair-men ts (19, 20), the h ealth of AD caregivershas not been studied using objective p hys-iologic measures. The relatively high in-cidence of depression among caregiverscould have implications for their health,particularly in regard to imm une function.Convergent data have linked depressionwith impaired immune function (9, 21 , 22).In order to better unders tand the hea lth-related consequences of caregiving, weobtained psychologic data and bloodsamples for immunologic and nutritionalanalyses from AD caregivers and socio-demographically matched comparisonsubjects. We expected that caregivers w ouldbe more depressed and w ould hav e poorerimmune status than comparison subjects.Moreover, greater impairment in AD pa-

524 Psychosomatic Medicine 49:523-535 (1987)


3/13

CHR ON IC STRESS AN D IMM UN ITY

tients was expected to be correlated withgreater dysfunction in caregivers.

METHODSThe subjects were 34 caregivers and 34 matched

comparison subjects. All subjects were interviewedand completed questionnaires at the time blood wasdrawn for the immunologic assays. All blood sam-ples were drawn at the same time of day to controlfor diurnal variation.The caregivers were recruited from the practicesof an Ohio State neurologist and an internist (n = 20),and from local support groups affiliated with the Alz-heimer's Disease and Related Disorders Association(n = 14). The comparison subjects were recruitedfrom newspaper advertisements, notes placed oncommunity bulletin boards, and personal contacts.In our laboratory's behavioral immunology re-search with younger populations we normally limitparticipation to those individuals who are not takingany medication and who have no health problemswith possible immunologic components. However,caregivers are primarily middle-aged or older, andthe majority of older adults are likely to be takingsome medications. Limiting the sample to those whoare unmedicated would probably result in biased data,because only the very healthiest of the populationwould be eligible (23). An alternative strategy thatwe used successfully in other research with olderadults (24) involved selecting as subjects those in-dividuals who were not taking any imm unosuppres-sive medication, and whose health problems did nothave an immunologic component (e.g., excluding thoseindividuals with cancer, recent surgeries, strokes,hormonal disorders, etc.). There is evidence that thisapproach results in reliable data: Goodwin and co-workers (25) examined various immunologic param-eters in 279 healthy and 24 chronically ill individ-uals over 65, and in young controls. Although therewere the expected differences between the young andold subjects on certain aspects of immune function,there were not differences between the two groupsof older adults. They noted that the data supportedage-related relationships in immune function, ratherthan age-associated diseases.

In addition, we matched subjects on the presenceor absence of two kinds of medicationuse of betablockers for hypertension and estrogen supplem ents.Beta blockers are widely prescribed among the el-

derly, and data from in vitro studies suggest that theymay have some consequences for blastogenesis (26).Although we are unaware of any comparable im-munologic data for estrogen, there is good evidencefor interactions between the immune system and theendocrine system (2). Rather than exclude all care-givers who might use these medications, we selec-tively recruited matched comparison subjects (de-scribed below) who were similarly medicated.Subjects were also matched on three sociodemo-graphic dimensions: age, sex, and education. Edu-cation was used as the socioeconomic status variablefor matching purposes because occupation is of lim-ited value in a sample that includes significant num-bers of older women who may or may not have workedoutside the home (27).

Self-Report DataDepression, social contact, loneliness, self-re-ported physical health, and financial resources wereassessed. The self-report m easures we used have beenused previously in studies with older adults.Demographic and Health Data. Information onage, sex, years of education, current or former oc-cupation, and household income were obtained.Health status data included medication usage, caf-feine intake, and recent alcohol intake. Subjects wereasked how many hours they had slept in the last 3days compared to their usual needs, and whetherthey had experienced any recent weight changes,particularly during the previous week. Data on recentillnesses included the number of physician visitsduring the previous 6 months, and the number ofdays the subject was ill enough to be in bed duringthe same time period.Depression. The short form of the Beck Depres-sion Inventory (BDI) (28) was used because it appearsto be more sensitive to mild to moderate levels ofdepression than other scales that are more biologi-cally based (29); in addition, it has fewer somaticallybased items that may be characteristic of older non-depressed individuals. It has been used extensivelyin research, including several studies with caregivers(15). Population norms provide cutoffs for varyinglevels of depression.Social Contact. We collected social contact in-formation using selected questions from the OlderAmericans' Resources and Services Multidimen-

Psychosomatic Medicine 49:523-535 (1987) 525


4/13

J. K. KIECOLT-GLASER et al.

sional Functional Assessment Questionnaire (OARS),since there are good normative data for older adults,and its reliability and validity have been well-doc-umented (30). The questions included informationon the number of persons the subject knew wellenough to visit in their homes, the number of timesthe subject talked to someone else in the past week,and the number of times they spent time with some-one who did not live with them (and who was nottheir AD relative). They were also asked how oftenthey find themselves feeling lonely.

AD Patient History and Current Function-ing. Information on the estimated duration of theillness was provided by the family caregiver, as wellas the time since a physician had made a tentativediagnosis of AD. Caregivers were also asked to esti-mate the amount of time they spent in caregivingactivities during the previous week.The ratings of family members can provide onesource of reliable data on level of AD patient function(31). Family members responded to the Memory andBehavior Problem Checklist (MBPC) (32) by provid-ing ratings of the frequency of relevant behaviorsassociated with AD, as well as separate reaction rat-ings for the degree of associated bother or upset. Thesum of the cross products of these two ratings pro-vides a m easure of the impact of the behaviors (33).

Immunologic AssaysThe fiveimmunologic assays included assessmentof the percentages of total T lymphocytes, helper Tlymphocytes, suppressor T lymphocytes, and naturalkiller (NK) cells. We also measured antibody titersto the latent herpesvirus Epstein-Barr virus (EBV),the etiologic agent for infectious mononucleosis.Monoclonal antibodies w ere used to provide dataon certain quantitative aspects of immune function.

Natural killer cell percentages were assessed becauseNK cells are thought to be an important antiviral andantitumor defense. Distress-related changes in NKcell percentages have been found previously in med-ical students (7). Reliable stress-related changes in afunctional NK measure, NK cell lysis, have been shownwith two different target cells (3, 4, 7) .Data on the relative percentages of helper and sup-pressor T lymphocytes were also collected. Helper Tcells stimulate important imm unologic activities, in-cluding the production of antibody by B lympho-cytes, an important defense against bacterial infec-tions. Suppressor T cells act to shut off the activityof helper cells when sufficient antibody has been

produced. Low helper-to-suppressor cell ratios areassociated with immunodeficiency (34). Alterationsin the percentages of helper and suppressor cells havebeen associated with examination stress (4, 5).We measured antibody to EBV because antibodytiters to latent herpesviruses appear to provide anindirect measure of cellular immune system com-petency (6). For example, patients on immunosup-pressive therapies like chemotherapy or patients withimmunosuppressive diseases (e.g., AIDS) have char-acteristic elevated herpesvirus antibody titers; ces-sation of an immunosuppressive drug therapy is ul-timately followed by a drop in antibody titers to latentherpes-viruses. The increased herpesvirus antibodyproduction in immunosuppressive conditions isthought to reflect the humoral immune system's re-sponse to an increased load of viral antigens. We havepreviously shown large and reliable stress-relatedchanges in antibody titers to EBV and herpes simplexvirus using a type 1 antigen (HSV-1) in medical stu-dents (6) as well as decrements in HSV-1 antibodytiters in elderly adults following a relaxation inter-vention (24).It should be noted that there is some evidence for

the genetic transmission of early-onset AD (35), andthere is some suggestion that AD may have an etiol-ogy related to immune function. The early-onset formof AD probably accounts for a relatively small pro-portion of AD patients, and the types of immunologicassays we used have not been related to the presenceor absence of AD in previous studies (36). Therefore,although we used offspring caregivers, we did notexpect that their immunologic data would be differ-ent from other subjects without a family history ofAD.NK, T Lymphocyte, and T Lymphocyte Subset As-says. The percentages of total T lymphocytes, helper-to-inducer T cells, and suppressor-to-cytotoxic T cells

were determined using the monoclonal antibodiesOKT-3, OKT-4, and OKT-8, respectively (Ortho), aspreviously described (5). The Leu-11 monoclonal an-tibody (Becton-Dickerson) was used to measure thepercentage of NK cells.Briefly, lymphocytes isolated on Hypaque-Ficollgradients were washed with trypsin diluent and th enresuspended in complete RPMI 1640 medium sup-plemented with 20% fetal bovine serum. M onocyteswere removed by placing the cell suspensions inplastic tissue culture flasks and incubating at 37Cin a CO2 incubator for 2 hours. The nonadherent cellswere washed off and used to determine percentageof T-cell subsets. Lymphocytes (10e) were incubated



5/13

CHR ONIC STRESS AN D IMM UN ITY

in 0.01 ml of Leu-11, OKT-3, OKT-4, or OKT-8 mon-oclonal antibody for 30 minutes on ice. Cells werewashed with cold RPMI1640/PBS (1:1), resuspendedin goat antimouse IgG conjugated to fluorescein is-othiocyanate (Cappel Laboratories), and incubatedfor an additional 30 minutes on ice. The cells werewashed and assayed, using an Ortho System 50 flu-orescence activated cell sorter (FACS).Immuno/luorescence Assay. The indirect im-

munofluorescence (IF) assay was used to measureantibodies to EBV virus capsid antigen (VCA) (6).Antibody titers were assayed using smears of HR-1cells. Cells werefixedin acetone at room temperaturefor 10 minutes, adsorbed with twofold dilutions ofplasma prepared in phosphate-buffered saline (PBS),pH 7.4, for 30 minutes at 37C. The cells were washedwith PBS and readsorbed with goat antihuman IgGconjugated to fluore scein isothiocyanate (FITC) for30 minutes at 37C, The cells were washed w ith PBS,counterstained with Evans blue, mounted in Protex,and examined with a Zeiss UV microscope. A back-ground control for the FITC-labeled antibody wasperformed. Antibody titers were determined by thehighest dilution of plasma still able to demonstrateIF positive cells. All slides were read blind coded.

Nutritional AssaysAlbumin and transferrin, two nutritional assayswith relatively shorter and longer half-lives, wereincluded to provide objective information on the nu-tritional status of subjects. There are well docu-mented impairments in various aspects of immunefunction in undernourished individuals, and mod-erate to severe protein-caloric malnutrition is asso-ciated with increased frequency and severity of in-fection (37).Protein assays provide better information on global

nutritional status than those for carbohydrates andfats, since the former have varied nu tritional buildingblocks, as well as very complex synthetic pa thways.Different protein markers were used because of thedifferences in their half-lives; the half-life of albuminis 2-3 weeks, in comparison to 8 days for transferrin.The procedure used to measure albumin is an ad-aptation of the bromcresol green dye-binding methodof Rodkey (38), later modified by Doumas (39). Thisprocedure is recognized as a particularly good pro-cedure as compared to other dye-binding techniquesbecause of its specificity and freedom from interfer-ence.Transferrin is an iron-transporting protein. Con-

centration in plasma is affected by dietary intake ofiron. Nutritionally deficient but calorie-rich diets aregenerally lacking in iron, and, as a result, plasmairon levels tend to be low and transferrin levels high.It has been shown that estimation of transferrin levelsmay be used to assess the effectiveness of total par-enteral nutrition (40).A rate nephelometry procedure using a Beckmanhuman immunoglobulin reagent kit and a Beckmanimmunochemistry analyzer system was used to ana-lyze transferrin levels. Antibody to human transfer-rin was used in the assay, in which the peak ratesignal caused by the antigen-antibody complex isproportional to the increase in light scatter that isread by the instrument (41).

RESULTSSociodemographic DataSociodemographic data are shown in

Table 1. Subjects ranged in age from 34 to82 . There were no reliable differences be-tween caregivers and comparison subjectson the matching variables of age or edu-cation, or in subjects' total family income,Fs < 1. Six of the matched pairs were tak-ing beta blocking medication, and five ofthe women in each cohort were taking anestrogen supplement.

The majority of the caregivers werespouses (n = 20), 13 were adult children,and 1 was an in-law. Half of the caregiverslived with their impaired relative (n = 17),10 AD patients were in nursing homes,and 7 of the AD victims lived alone or withanother relative but had additional careprovided by our interviewees. The lengthof time reported since the caregiver re-ported that he or she had first noticed anyAD symptoms ranged from 9 months to 16years, with a mean of 5.45 years(SD = 3.82). The time that had elapsedsince the initial physician's diagnosis ofprobable AD ranged from a new diagnosisto 11 years, with a mean and median of2.83 years and 2.04 years, respectively.



6/13

J. K. KIECOLT-CLASER et al.

TABLE 1. Subjects ' Sociodem ographic Character is t ics

SexMalesFemalesAgeMarita l statusMarriedSeparated/divorcedWidowedNumber of childrenEducation

Partial high schoolHigh school graduatePost high school business or trade school1-3 years collegeCollege graduatePostgraduate collegeAnnual family income10,000-14,99915,000-19,99920,000-29,99930,000-39,99940,000 or moreUnanswered

AD caregivers112359.32 (12.98)32112 90 (1.84)

15271182649112

Comparison subjects112360.29(13.27)2941

2.44 (1.46)7467

103564

151

TABLE 2. Me ans (SDs) for Psycholog ic Data itom th e BDI and the OAR SAD caregivers Comparison Subjects

Beck Depression Inventory short form)3Current life satisfaction3(0 = poor, 2 = good)Self-rated current mental health3Self-rated mental health, compared to5 years ago (0 = worse,3 = better)3

Number of people known well enoughto visit in their homes(0 = none, 3 = 5 or more)Frequency of phone conversationswith friends, relatives, or others inthe past week (0 = none, 3 = dailyor more)Frequency of visits with someone notliving with subject in the past week(0 = none, 3 = daily or more)Feelings of loneliness(0 = quite often, 2 = almost never)

4.88 (6.18)1.62 (0.66)1 84 (0.87)1 57 (1.07)

2.90 (0.38)2.42 (0.66)

2.12 (0.74)

1.48 (0.61)

2 48 (2.58)1.91 (0.29)2.38 (0.65)2 03 (0.75)

2 85 (0 43)2.35 (0.59)

1.88 (0.84)

1.64 (0.59)ap < 0 05.bp< 0 .01 .



7/13


Initial analyses of variance (ANOVAs)that included sex of subject revealed nodifferential sex effects for the major im-munologic and self-report variables.Therefore, data were combined for sub-sequent analyses using ANOVAs with onevariable, group membership.Self-Report DataPsychologic data are shown in Table 2.Caregivers had significantly higher scoreson the short form of the BDI than com-parison subjects, F(l,66) = 4.02, p < 0 .05.Similarly, caregivers reported signifi-cantly lower general life satisfaction thancom parison subjects, F(l,66) = 5.34,p < 0.05. Caregivers also rated their men-tal health as poorer than did comparisonsubjects on the OARS item, F(l,66) = 8.11,p < 0.01. Caregivers were significantlymore likely than comparison subjects tosay that there was a negative change intheir mental health compared to 5 yearsago, F(l,66) = 4.79, p < 0.05.There were no significant differences

between caregivers and comparison sub-jects on social support and lonelinessquestions from the OARS, as shown in Ta-ble 2. However, the pattern of correlationsamong AD impairment indices and socialcontact items shown in Table 3 suggeststhat caregivers with more impaired rela-tives had fewer social contacts with othersand we were lonelier. In addition, the sig-nificant correlations between BDI scoresand the three indices of impairment fromthe Memory and Behavior ProblemsChecklist suggest that greater impairmentin the AD patient may be associated withgreater distress in the caregiver.

Immunologic DataComparisons of immunologic datashowed differences in the predicted di-rection, as seen in Table 4. The AD care-givers had significantly higher a ntibod y ti-ters to EBV VCA than comparison subjects,F(l,65) = 4.65, p < 0.05, presumably re-flecting poorer cellular immune systemcontrol of herpesvirus latency in the for-

TABLE 3. Correlations Among Impaired Ind ices for the AD Victim and Depression and Soc ial Contactsin Caregivers

1.2.3.4.5.

6.78.

MBPC frequencyMBPC upsetMBPC cross-product offrequency and upsetBDI depression scoreNumber of people knownwell enough to visit intheir homeFrequency of phoneconversations, past weekFrequency of visits, pastweekFeelings of loneliness

1.1.000.62"0.61

a

0.28b- 0 . 2 5

0.10- 0 .23- 0 4 4c

2.1.000.93a0.48c- 0 . 6 6 a

0.09- 0 . 4 3 c- 0 . 4 1 c

3.

1.000.60"- 0 . 6 3 "

0.07- 0 . 4 2 c- 0 . 4 6 c

4.

1.00-0 .4 5 c

0.16- 0 . 5 1 a- 0 . 5 0 a

5.

1.00

0.190.48c0.32b

6.

1.000.150.02

7.

1.000.35"

a p < 0.001.bp < 0.05.c p < 0.01.



8/13

I. K. KIECOLT-CLASER et al.

TABLE 4. Means (SOs) for Immu nologic and Nutritional Data

Immunologic AssaysEBV VAC*6Percentage of total T lymphocytes6Percentage of helper T lymphocytes^Percentage of suppressor T lymphocytesHelper-suppressor ratiodPercentage of NK cellsNutritional AssaysAlbuminTransferrin

AD Caregivers640.70 (570.01)48.72 (13.97)33.51 (10.86)18.30 (10 50)1.90 (0.97)12.88 (6.92)

4.23 (0.33)296.78 (48.14)

Comparison subjects376.72 (446.39)56.53 (14 15)

45.94 (12.05)21.02 (9.86)2.80 (1.86)15.35 (9.53)4.14 (0.25)284.79 (59.21)

aHigher antibody titers to a latent herpesvirus are thought to reflect poorer cellular immune system control overvirus latency.6p < 0.05.c p< 0.0001.d p< 0 . 0 1 .

mer. Caregivers had significantly lowerpercentages of both total T lymphocytes,F(l,62) = 4.87, p < 0.01, and helper Tlymphocytes than comparison subjects,F(l,62) = 18.49, p < 0.0001. Although thetwo groups did not differ in the relativepercentages of suppressor T cells,F(l,6 2) = 1.12, the re we re significant dif-ferences in the helper-to-suppressor ratio,F(l,62) = 5.77, p < 0.05. There were nosignificant differences in the percentagesof NK cells, F(l,56 ) = 1.27.

Nutritional DataWe used two nutritional m arkers, plasmaalbumin and transferrin levels, to assessthe nutritional status of the subjects, sinceinadequate nutrition is associated withimpairments in imm unity. All subjects werewithin norm al range for both markers. Thetwo groups did not show even marginald i f fe rences on e i ther a lb um in ,F(l,66) = 1.81, or transferrin, F < 1. Thus,there is no evidence that the immuno logicdifferences simply reflect underlying dif-ferences in nutrition.

Health-Related DataThe two groups did not differ reliablyin the number of physician visits duringthe previous 6 months, or in the numberof days they were ill in the same time pe-riod, Fs < 1. Similarly, there were not re-liable differences in self-ratings of curr enthealth, or health as of 1 year ago, Fs < 1.The two groups did not show differ-ences in behavior that might have immu-nologic or other health-related conse-quences, including the number of smokersin the two groups, the number of packssmoked per week, the amount of alcohol

consumed in the previous week, the av-erage number of drinks normally con-sumed in a week, or in caffeine u se, Fs < 1.There were no differences between thegroups in recent weight change,F(l,66) = 2.12. There was a significantdifference between the two groups in theamount of sleep subjects reported withinthe previous 3 days, F(l,66) = 4.60,p < 0.05, but the differences were no t large.Caregivers reported an average of 23.07(SD = 1.98) h our s, wher eas com pariso n53 0 Psychosomatic Medicine 49:523-535 (1987)


9/13


subjects reported a mean of 24.32 hours(SD = 2.83).In order to help evaluate the possibilitythat any psychologic and immunologicdifferences between the two groups sim-ply reflected diffences in the amount ofsleep, correlations were computed be-tween sleep and the immunologic param-eters in caregivers. All of the correlationswere negative, but none was significant.Sleep correlated -0 .1 0 with BDI scores,whereas correlations between sleep and theimmuno logic measu res ranged from - 0.01to - 0 . 2 3 .

Residence of AD Victim andFunctioning of CaregiverIn order to evaluate possible differencesamong caregiver functioning as conse-quence of where their AD relative lived(with them, in an institution, or else-where), we compared demographic, self-report, immunologic, and n utritional data.The caregivers whose relative lived else-where were younger (mean age of 45.57,SD = 9.62) than those whose relative livedwith them (mean = 62.06, SD = 10.00) orthose whose relative was institutionalized(mean = 65.80, SD = 12.87), F(2,31) =7.42, p < 0.01. There were no differencesin the total time spent each day in care-

giving activ ities, F(2,31) = 2.20, in part asa function of the very wide variabilitywithin each of the three groups: those liv-ing with their impaired relative reportedspending an average of 9.87 hours a day(SD = 9.99), those whose relative was in-stitutionalized reported a mean of 3.1 hourseach day (SD = 2.33), and those whoserelative lived elsewhere reported an av-erage of 6.42 hours per day (SD = 7.93).Those caregivers whose relatives were in-stitutionalized reported a significantly

longer duration of illness, F(2,31) = 12.18,p < 0.0001. The two groups did not differreliably on any of the three memory andbehavior indices, Fs < 1.There were no reliable differences onBDI scores, life satisfaction, or self-ratedmental or physical health. Th ere were sig-nificant immunologic differences only forpercentages of NK cells, F(2,27) = 6.49,p < 0.01, with those whose relative wasinstitutionalized having the highest val-ues, with a mean of 19.26 (SD = 9.24),compared to those who lived with theirAD relative (mean = 10.17, SD = 4.07),and those whose relative lived elsewhere(mean = 11.40, SD = 4.25). Nutritio nalcom parisons sho wed no significant differ-ences among these three groups.

Differences in AD Caregiversas a Function of SupportGroup AttendanceWe compared the 14 caregivers who at-tended a support group with the 20 whodid not. The two groups did not differ re-liably on age, edu cation , or family inc ome .Support group members had been caregiv-ing for substantially longer periods of time,F(l,31) = 7.66, p < 0.01. Sup port groupmembers rated themselves as significantlyless lonely, F(l,31) = 6.11, p < 0.05, andhad significantly high er perce ntages of NK

cells than nonmembers, F(l,29) = 7.10,p < 0.01. The differences on o ther p sy-chologic or immu nologic parame ters werenot significant.

DISCUSSIONTaken together, these data support theprimary hypothesis of the investigation:caregivers appear more distressed and have



10/13

J. K. KIECOLT-GLASER et al.

poorer immune function than their well-matched age peers. We found no evidencethat the observed differences were a func-tion of nutrition, alcohol use, or caffeineintake. Although the caregivers reportedless sleep than comparison subjects, aswould be expected from other studies (42),the amount of sleep was not reliably cor-related with immune function or mood incaregivers.

The differences in immunity betweenthe caregiving and comparison subjectgroups are particularly noteworthy, be-cause these caregivers are less distressedthan other caregiver samples described inthe literature (15-17). These differencesmay be related to the fact that our care-givers were relatively well-educated, andincome data suggest that they have morefinancial resources than described in othersimilar samples (33, 43). Given these rel-ative advantages, it is reasonable to sug-gest that these data represent a "best casescenario" (44). The persistence of signifi-cant psychologic and physiologic differ-ences between our two cohorts in spite ofthese relative advantages strongly sup-ports the hypothesized negative impact ofcaregiving responsibilities.In contrast to other studies (43, 45), wedid not find that caregivers were more likelythan individuals without comparable re-sponsibilities to become isolated from their

usual com panions an d social activities be-cause of the time demands. However, wedid find correlation evidence within thecaregiver cohort that linked greater im-pairment in the AD victim with fewer so-cial contacts and greater depression andloneliness. If caregivers experience in-creased social isolation as their relative'scondition deteriorates, it could have im-portant consequences; research with botholder and younger adults suggests that so-cial support may moderate stress-related

depression or dysphoria and may also berelated to morbidity and mortality (46,47).The immunologic data provide evi-dence of persistent alterations in cellularimmunity associated with a chronic psy-chosocial stressor, in contrast to the longer-term adaptation or enhancement of im-munity found in some studies with ro-dents (10,11). Related evidence consistentwith possible longer-term immunologicalterations in humans was provided in re-search on immune function, marital qual-ity, and marital disruption (8); however,immunologic changes in these ind ividualswere more closely tied to their adaptationto an acute stressor, marital disruption.

Although we did not find health differ-ences between the two cohorts, the ob-served immunologic differences are con-sistent with the kinds of changes, (thoughof a much lesser magnitude) that are ob-served in immun e-suppressed patients. Forexample, transplant patients and AIDS pa-tients are good examples of immunesuppression, although resulting from dif-ferent processes; within both groups it isknown that reactivation of latent virus oc-curs, including reactivation of EBV withassociated increases in EBV antibody ti-ters (49). There are higher EBV antibodytiters among elderly than among youngeradults (50), consistent with evidence thatthe former have relative deficits comparedto their younger counterparts on somefunctional or qualitative immunologic as-says (25, 51).Data from other stud ies suggest there m aybe a number of other stress-related im-munologic changes associated with acutestress, and such changes may have healthconsequences. For example, medical stu-dents followed over an academic yearshowed higher EBV antibody titers con-comitant with a decrease in specific cellkilling of EBV-infected cells during ex-



11/13


amination, compared with samples takenbetween examination periods (52). Otherconcurrent immunologic changes in-cluded m arkedly lower production of 7 in-terferon by lymphocytes stimulated withCon A associated with examination stress.The activity of a lymphokine, leukocytemigration-inhibition factor, normally sup-pressed during recrudescence of herpessimplex virus type 2 infections, was al-tered during examination periods, and anincrease in both plasma and intracellularlevels of cyclic AMP was associated withexamination stress as well. The medicalstudents also reported more illness duringexamination periods.

Chronic distress-related immunosup-pression may have its most importanthealth consequences in groups with other,preexisting immunologic impairments,such as older adults with age-related dec-rements in immune function (48). Indi-rect evidence consistent with this prem-ise is provided by the remarkably highmortality rates for elderly psychiatric pa-tients from pneumonia: within the firstyear after psychiatric admission, there arefifty times more deaths from this infec-tious disease than found among age-matched general population counter-parts. The ratio drops to twenty times thatof their age-matched counterparts by thesecond year of institutionalization, sug-gesting that the hospital environment perse may be a less critical factor in mortal-ity than the transition (53). In this con-text it should be noted that depression is

the leading reason for psychiatric hospi-talization in the elderly (54).Since the modal age of onset for AD is65-69 years (14), AD caregivers are them-selves most often middle-aged or elderly(55). The significant functional immuno-logic decrements that accom pany aging (25)are thought to be associated with the in-creased morbidity and mortality of infec-tious disease in the elderly (49, 52). Lon-gitudinal studies with chronically stressedat-risk groups like caregivers may providevaluable information on the contributionof psychosocial v ariables to mor bidity an dmortality.

This study was supported in part by agrant from the American Cancer ResearchFund, by grant No. 1 ROl MH40787 fromthe National Institute 0/ Mental Health,an d by the Ohio State University Com-prehensive Cancer Center Grant No. CA-16058, from the National Cancer Institute,National Institutes 0/ Health.We are very grateful to Enid Light atNIMH, who suggested that we pursue thequestion of immune /unction in caregiv-ers. We are also very appreciative 0/ theefforts 0/Ann a Maria Malia, the ExecutiveDirector of the Columbus A lzheimer's Dis-ease and Related Disorders Association.We thank Roy Donnerberg for his assis-tance with subject recruitmen t. We appre-ciate the kind help with interviewing andlaboratory assistance provided by SusanMalkojf, Paul Vespa, Dawn Reese, and JaneHoJJiday.

References

1. Solomon GF, Am kraut AA: Psyc honeur oendoc rinologic al effects on the immu ne response. Annu RevMicrobiol 35:155-184, 1981

2. Ader R (ed): Psychoneuroim munology. New York, Academic, 1981



12/13

J. K. KIECOLT-GLASER et a l .

3. Kiecolt-Glaser JK, Garner W, Speicher CE, Penn G, Glaser R: Psychosocial modifiers of immunocom-petence in medical students. Psychosom Med 46:7-14, 19844. Kiecolt-Glaser JK, Glaser R, Strain E, Stout J, Tarr K, Holliday J, Speicher C: Modulation of cellularimmunity in medical students. J Behav Med 9:5-21, 19865. Glaser R, Kiecolt-Glaser JK, Stout JC, Tarr KL, Holliday JE: Stress-related impairments in cellularimmunity. Psychiatry Res 16:233-239, 19856. Glaser R, Kiecolt-Glaser JK, Speicher CE, Holliday JE: Stress, loneliness, and changes in herpesviruslatency. J Behav Med 8:249-260, 19857. Glaser R, Rice J, Speicher CE, Stout JC, Kiecolt-Glaser JK: Stress depresses interferon production byleukocytes concomitant with a decrease in natural killer cell activity. Behav Neurosci, 100:675-678,19868. Kiecolt-Glaser JK, Fisher LD, Ogrocki P, Stout JC, Speicher CE, Glaser R: Marital qu ality, maritaldisruption, and immune function. Psychosom Med 49:13-34, 19879. Stein M, Keller SE, Schleifer SJ: Stress and immunom odulation: The role of depression and neu roen-docrine function. J Immunol 135:827s-833s, 198510. Monjan AA, Collector MI: Stress-induced modulation of the immune response. Science 196:307-308,197711. Sklar LS, Anisman H: Stress and coping factors influence tumor growth. Science 205:513-515, 197912. Reisberg B (ed): Alzheimer's Disease: The Standard Reference. New York, Free Press, 198313. Heckler MM: The fight against Alzheimer's Disease. Am Psychol 40:1240-1244, 198514. Heston LL, Mastri AR, Anderson VE, White G: Dementia of the Alzheimer type. Arch Gen Psychiatry38:1085-1091, 198115. Fiore J, Becker J, Coppel DB: Social network interactions: A buffer or a stress? Am J Community Psychol11:423^439, 198316. Crook TH, Miller NE: The challenge of A lzheimer's Disease. Am Psychol 40:1245-1250, 198517. Eisdorfer C, Kennedy G, Wisnieski W, Cohen D: Depression and attributional style in families copingwith the stress of caring for a relative with Alzheimer's Disease. Gerontologist 23:115-116, 198318. George LK, Gw yther LP: The dy namics of caregiver burden: Changes in caregiver well-being over time.Paper presented at the annual meeting of the Gerontological Society of America, San Antonio, November198419. Brocklehurst JC, Morris P, Andrew s K, Richards B, Laycock P: Social effects of stroke . Soc Sci Med15A:35-39, 198120. Sainsbury P, Grad de Alarcon J: The psychiatrist and the geriatric patient: The effects of communitycare on the family of the geriatric patient. J Geriatr Psychiatry 4:23-41, 197021. Schleifer SJ, Keller SE, Meyerson AT, Raskin MJ, Davis KL, Stein M: Lymphocyte function in majordepressive disorder. Arch Gen Psychiatry 41:484486, 198422. Kiecolt-Glaser JK, Stephens RE, Lipetz PD, Speicher CE, Glaser R: Distress and DNA repair in humanlymphocytes. J Behav Med 8.311-320, 198523. Krauss IK: Between- and within-group comparisons in aging research. In Poon LA (ed), Aging in the1980s: Psychological Issues. Washington, DC, 1980, pp 542-55124. Kiecolt-Glaser JK, Glaser R, Williger D, Stout J, Messick G, Sheppard S, Ricker D, Romisher SC, BrinerW, Bonnell G, Donnerberg R: Psychosocial enhancement of immunocompetence in a geriatric popu-lation. Health Psychol 4:25-41, 198525. Goodwin JS, Searles RP, Tung KSK: Immunological responses of a healthy elderly population. ClinExp Immunol 48:403-410, 198226. Goodwin JS, Messner RP, Williams RC: Inhibitors of T-cell mitogenesis: Effects of mitogen dosage. CellImmunol 45:303-308, 197927. Rook KS: The negative side of social interaction: Impact on psychological well-being. J Pers Soc Psychol46:1097-1108, 198428. Beck AT: Depression: Clinical, Experimental, and Theoretical Aspects. New York, Harper and Row,196729. Hammen CL: Assessment: A clinical and cognitive emphasis. In Rehm LP (ed). Behavioral Therapy forDepression: Present Status and Future Directions. New York, Academic, 1981

534 Psychosomatic Medicine 49: 523 -53 5 (1987)


13/13


30. Fillenbaum GG, Smyer MA: The development, validity, and reliability of the OARS MultidimensionalFunctional Assessment Questionnaire. J Gerontol 36:428-434, 198131. Reifler BV, Eisdorfer C: A clinic for the impaired elderly and their families. Am J Psychiatry 137:1399-1403,198032. Zarit SH, On NK, Zarit JM: The Hidden Victims of Alzheimer's Disease: Families Under Stress. NewYork, New York University Press, 198533. Zarit SH, Todd PA, Zarit JM: Subjective burden of husbands and wives as caregivers: A longitudinalstudy. Gerontologist 26:260-266, 198634. Reinherz EL, Schlossmen SF: Regulation of immune responseInducer and suppressor T-lymphocytesubsets in human beings. N Engl J Med 303:370-373, 198035. Heston LL: Genetic relationships in early-onset Alzheimer's dementia, pp 566-8. In Cutler NR, mod-erator, Alzheimer's disease and Down's syndrome: New insights. Ann Inter Med 103:566-578, 198536. Leffell MS, Lumsden L, Steiger WA: An analysis of T lymphocyte subpopulations in patients withAlzheimer's Disease. J Am Geriatr Soc 33:4-8, 198537. Chandra RK, Newberne PM: Nutrition, Immunity and Infection: Mechanisms of Interactions. New York,Plenum, 197738. Rodkey FL: Direct spectrophotometric determination of albumin in human serum. Clin Chem 11:478487,198539. Doumas BT, Watson WA, Biggs HG: Albumin and the measurement of serum album in with bromcresolgreen. Clin Chim Acta 31:87-96, 197140. Keyser JW: Human Plasma Proteins. New York, Wiley, 197941. Buffone GJ: Transferrin, ICS-14. Fullerton CA, Beckman Instruments, 198042. Rabins PV, Mace HL, Lucas MJ: The impact of dementia on the family. JAMA 248:333-335, 198243. George LK, Gwyther LP: Caregiver well-being: A multidimensional examination of family caregiversof demented adults. Gerontologist 26:253-259, 198644. Gwyther LP, George LK: Caregivers for dementia patients: Complex determinants of well-being andburden. Gerontologist 26:245-247, 198645. Johnson CL, Catalano DJ: A longitudinal study of family supports to impaired elderly. Gerontologist23:612-618, 198346. Blazer D: Social support and mortality in an elderly community population. Am J Epidemiol 115:684-694,198247. Cohen CI, Teresi J, Holmes D: Social networks, stress, and physical health: A longitudinal study of aninner-city elderly population. J Gerontol 40:478^186, 198548. Kiecolt-Glaser JK, Glaser R: Interpersonal relationships and immune function. In Carstensen L, NealeJ (eds), Mechanisms of Physiological Influence on Health. New York, Wiley, in press.49. Glaser R, Gottlieb-Stematsky T (eds): Human Herpesvirus Infections: Clinical Aspects. New York, MarcelDekker, 198250. Glaser R, Strain EC, Tarr K, Holliday JE, Donnerberg RL, Kiecolt-Glaser JK: Changes in Epstein-Barrvirus antibody titers associated with aging. Proc Soc Exp Biol Med 179:253-355, 198551. Roberts-Thomson IC, Whittingham S, Youngchaiyud U, MacKay IR: Aging, immune response, andmortality. Lancet 2:368-370, 197452. Glaser R, Rice J, Sheridan J,Fertel R, Stout J,Speicher CE, Pinsky D, Kotur M, Post A, Beck M, Kiecolt-Glaser JK: Stress-related immune suppression: Health implications. Brain Behav Immun, in press.53. Craig TJ, Lin SP: Mortality among elderly psychiatric pa tients: Basis for preventive intervention. J AmGeriatr Soc 29:181-185, 198154. Solomon K: The depressed patient: Social antecedents psychopathologic changes in the elderly. J AmGeriatr Soc 29:14-18, 198155. Zarit SH, Reever KE, Bach-Peterson J: Relatives of the impaired elderly: Correlates of feelings of burden.Gerontologist 20:649-655, 1980

Psychosomatic Medicine 49 :523 -53 5 (1987) 535

Kiecolt-Glaser et al.[1]

Documents

Transcript of Kiecolt-Glaser et al.[1]