social support as a moderator of life stress

PRESIDENTIAL ADDRESS—1976

Social Support as a Moderator of Life Stress

SIDNEY COBB.MD

"I assure you it is much wholesomer to be a complaisant, goodhumored, contented Courtier, than a Grumbletonian Patriot, alwayswhining and snarling."

—John Adams to his wife Abigail. The Hague, July 1, 1782

Social support is defined as information leading the subject to believe that he is cared for andloved, esteemed, and a member of a network of mutual obligations. The evidence that supportiveinteractions among people are protective against the health consequences of life stress is re-viewed. It appears that social support can protect people in crisis from a wide variety ofpathological states: from low birth weight to death, from arthritis through tuberculosis todepression, alcoholism, and the social breakdown syndrome. Furthermore, social support mayreduce the amount of medication required, accelerate recovery, and facilitate compliance withprescribed medical regimens.

Everybody talks about health, but no-body does much about it. The issue wasstated clearly by Stephen Smith (53), thefirst president of the American PublicHealth Association. He said,".. . .the cus-toms of society must be so changed that thephysician is employed to prevent ratherthan cure disease." Only recently has thisconcept begun to be implemented in theUnited States as a part of the HealthMaintenance Organization movement(37). It is, therefore, timely to address our-selves to preventive issues. As the titlesuggests, this essay will focus on socialsupport. It will examine some of the areasin which social support has been demon-strated to have dramatic health-related ef-

From the Sections of Community Health andPsychiatry of the Program in Medicine at Brown Uni-versity.

Presented in part as the Presidential Address be-fore the Society at its Annual Meeting, March 27,1976.

Address reprint requests to: Box G, Brown Univer-sity, Providence, Rhode Island 02912.

Received for publication May 6, 1976.

fects and identify some in which it seemsto have had no effects. It will not attempt toreview all the diverse literature on thissubject, for exhaustive bibliographies areavailable (23, 30, 49). Rather, it will em-phasize the way that social support acts toprevent the unfortunate consequences ofcrisis and change.

Before proceeding, we must come tosome mutual understanding of the conceptof social support. For the present discus-sion, social support is conceived to be in-formation belonging to one or more of thefollowing three classes:

1. Information leading the subject to be-lieve that he is cared for and loved.

2. Information leading the subject to be-lieve that he is esteemed and valued.

3. Information leading the subject to be-lieve that he belongs to a network of com-munication and mutual obligation.

Let us examine each in turn.Information that one is cared for and

loved or, as the Greeks might say, informa-tion about agape", is transmitted in intimate

300 Psychosomatic Medicine Vol. 38; No. 5 (September-October 1976)

Copyright ® 1976 by the American Psychosomatic Society, Inc.Published by American Elsevier Publishing Company, Inc.

SOCIAL SUPPORT AS A MODERATOR OF LIFE STRESS

situations involving mutual trust. In adyadic relationship, this informationmeets Murray et al.'s (43) need succorancefor one person, need nurturance for theother, and need affiliation for both. It isoften called emotional support.

Information that one is valued and es-teemed is most effectively proclaimed inpublic. It leads the individual to esteemhimself and reaffirms his sense of personalworth. It may be called esteem support.

Information that one belongs to a net-work of mutual obligation must be com-mon and shared. It must be common in thesense that everyone in the network has theinformation and shared in the sense thateach member is aware that every othermember knows. The relevant informationis of three kinds. The first answers thequestions: What is going on and how did itbegin? What is the relationship betweenus? How and when did we get here? Thesequestions are the essence of history. Thesecond pertains to goods and services thatare available to any member on demandand includes information about the acces-sibility of services that are only occasion-ally needed, e.g., equipment, specializedskills, technical information. The thirdcontains information that is common andshared with respect to the dangers of lifeand the procedures for mutual defense. Inthis last sense, the knowledge that a com-petently staffed hospital is available incase of need is socially supportive.

The present meaning of social supportdoes not include the activities of the hospi-tal in repairing a broken leg. Those ac-tivities are material services and are not ofthemselves information of any of the majorclasses mentioned above. This does notmean that the deferential manner of theintern may not provide esteem support orthat the tender care of the nurses may notcommunicate emotional support. It is only

to say that the services do not in them-selves constitute such support because so-cial support, being information, cannot bemeasured as mass or energy. This distinc-tion is important, for goods and servicesmay foster dependency, while the classesof information listed above do not. In fact,they tend to encourage independent be-havior.

This set of dimensions is hardly new.Angyal (1), Antonovsky (2), Fromm (22),Leighton (34), Weiss (56), and manyothers have illustrated and illuminatedthem. Perhaps the most notable illumina-tion is the novel Come Near by AlexanderLeighton (35). Gerald Caplan (9) and hiscolleagues have taught the importance ofthese concepts in community mentalhealth. But let's face it, these notions havebeen expressed over the millenia in thewritings of most of the world's religiousleaders. I have only added some precisionand emphasized that it is informationrather than goods or services that is centralto the concept.

The first group, emotional support, wasinitially expressed in the need terms ofMurray (43). The second can similarly beexpressed as need recognition from theMurray lexicon. The third is clearly akin toat least two of Leighton's (34) essentialstriving sentiments: "Orientation in termsof one's place in society .. ." and "Thesecuring and maintaining of membershipin a definite human group." This meansthat the whole concept can be expressed inperson-environment fit terms (21,41) or asthe extent to which the relevant needs aremet.

Social support begins in utero, is bestrecognized at the maternal breast, and iscommunicated in a variety of ways, butespecially in the way the baby is held(supported). As life progresses, support isderived increasingly from other members

Psychosomatic Medicine Vol. 38, No. 5 (September-October 1976) 301

SIDNEY COBB

of the family, then from peers at work andin the community, and perhaps, in case ofspecial need, from a member of the helpingprofessions. As life's end approaches, so-cial support, in our culture, but not in allcultures, is again derived mostly frommembers of the family.

As will be seen in the section on themechanism of this effect, it is my currentopinion that social support facilitates cop-ing with crisis and adaptation to change.Therefore, one should not expect dramaticmain effects from social support. There areof course some main effects simply be-cause life is full of changes and crises. Thetheory says that it is in moderating theeffects of the major transitions in life and ofthe unexpected crises that the effectsshould be found. This theory is supportedby the work of Pinneau (49), who foundfew effects in cross-sectional studies.

With this background, it is time to turnto a careful examination of the extent towhich this social support protects an indi-vidual as he passes through the varioustransitions and crises of the life cycle. Wewill begin with the infant in utero and endwith death.

PREGNANCY, BIRTH,AND EARLY LIFE

The elegant study of Nuckolls et al. (44)is a good place to start this review. Thiswas a study of 170 army wives delivered ata large military hospital. Data on lifechange scores before and during pre-gnancy and on psychosocial assets werecollected. The measure of psychosocial as-sets (TAPPS) covers all three of the areasdescribed above as the main componentsof social support in a subjective way andrelatively little else except for some as-sessment of affect, which we know from

TABLE 1. Percent of Women with Complications ofPregnancy by Life Change Score and Social Support(TAPPS) [Recalculated from Data of Nuckolls et al.

(44)]

Life changescore

Social support

High Low

High3 33%(15)b 91% (11) 3.87 <0.001Low 39% (72) 49% (72) NS

Interaction I = 2.24 P < 0.05.a High both before and during pregnancy.b Numbers in parentheses are the numbers of women

in the respective cells.

other studies to be associated with sup-port.

A recalculation of the data of Nuckolls etal. (44) is presented in Table 1. Thosewomen who are designated as having highlife change scores are those who wereabove the median both before and duringpregnancy. All other women have been in-cluded in the low category. The measure ofsocial support (TAPPS) is split at the me-dian. As can readily be seen, the propor-tion of women having complications is ex-cessive (91%) only in the high lifechange /low social support cell. However,the upper left-hand cell (high life changeand high support) is the really interestingone, for here 15 women were exposed tothe same high frequency of life changesbut had no increase in complications, pre-sumably because of some protective effectexerted by the high level of social support.One useful point about this study is that, ifthe association is causal, the direction isclear, for the complications cannot haveinfluenced the TAPPS score, which wasmeasured at the first visit, or the lifechanges, almost all of which occurred wellbefore the complications. This kind ofmoderating effect will appear again andagain as this review progresses.

Another approach to social support in

302 Psychosomatic Medicine Vol. 38, No. 5 (September-October 1976)


TABLE 2. Percent of Babies with Birth Weight < 2500 g by Education, Race, and Wantedness [Recalculatedfrom Data of Morris et al. (42)]

Wanted ordid notmatter

Timing erroror unwanted

N

Less than 12 years educationBlackWhite

12 or more years educationBlack (%)White (%)

8672118

227875

10.74.9

3.12.4

16571434

257359

11.56.1

10.56.4

NSNS

<0.001<0.01

pregnancy is reported by Morris et al. in1973 (42). The data come from a study ofwantedness of babies in 60 major hospitalsin 17 cities. A simplified presentation oftheir findings is to be seen in Table 2. It isclear that, at least for women who havecompleted high school, the reporting thata baby was wanted at the time it was con-ceived is associated with a significant de-crease in the frequency of low birthweight. This is true for both blacks andwhites. Over against this, one must set aSwedish study that did not find a differ-ence in birth weight between babies forwhom abortion was requested but refusedand other babies (27). However, educationwas not used as a control variable in thisstudy. It is not reasonable to suppose that"wantedness" is information that istransrr itted from mother to fetus and in-fluences growth rate. Rather, it seemslikely that the most common reason that awoman rejects her baby is that she herselffeels inadequately socially supported. Thesocietal reaction to illegitimate pregnancyis a case in point. In this instance, thecausal direction is not so clear, for the in-quiry was made the day after delivery andit is possible that babies known to be smallwere less likely to be declared to have beenwanted than those known to be large.

Moving on to the first major social de-mand that the child faces, namely,achievement of sphincter control, Steinand Susser (54) tell us that control of blad-der function at night was significantly de-layed for those children whose motherswent out to work while the children werein the second six months of life. This wasnot true if the mother went to work earlieror later in the child's life nor was it true ofthe small number of children who had asubstantial parental separation. These dataare open to a variety of possible interpreta-tions, so further study is indicated, but itseems at least possible that social supportgiven by the mother during the time whentoilet training is beginning is important tothe early acquisition of sphincter control.

When we get to the important area oflater social development, there are twostudies. Forssman and Thuwe (20) haveshown that wanted children adapt toand /or cope with the stresses of growingup better than those who started out with aparental request for abortion that was de-nied, in a study that followed a matchedseries of 120 cases and 120 controls untilage 21. The cases fared worse with respectto juvenile delinquency and need forpsychiatric treatment. Particularly strik-ing was the distribution with respect to


SIDNEY COBB

TABLE 3. The Educational Achievement of Un-wanted Children Compared with That of Controls.

[Data of Forssman and Thuwe (20)]

Advanced study

Completed requiredschooling

Retarded

Total

Unwanted

17

90

13

120

X2 = 13.4; P < 0.005.

Controls

40

74

6

120

Total

57

164

19

240

educational achievement. This is pre-sented in Table 3. It is of special interestthat the authors point out that the negativeeffects were essentially wiped out for thosecases that were reared by their two naturalparents together. The force of simpleeconomic factors in this situation cannotbe neglected.

The more recent study by Dytrych et al.(17) of the children of Czech mothers who

had been twice refused abortion for therelevant pregnancy has an appropriatelymatched control group but is subject to thecriticism, on the material so far presented,that the number of significant findingsdoes not seem to exceed the number to beexpected by chance. However, the find-ings in the area of socialization into schoolare quite suggestive because, of the 15 per-formance and behavioral items presented,all are in the predicted direction and twoare significant. We will all look forward tothe full report with interest. The possibil-ity that social support increases the effi-ciency of socialization in school must bekept in the forefront of our minds.

TRANSITIONS TO ADULTHOOD

There seems to be little informationabout the effects on the transitions to col-

lege, to first job, and to marriage. This is anarea in which the literature must be morethoroughly combed and in which specificresearch is indicated.

HOSPITALIZATION

The effects of social support on an indi-vidual in relation to hospitalization formental and physical illness are wide-spread, but much of the evidence is infe-rential. That is to say, the several studies inquestion imply differences in social sup-port without measurement. The socialbreakdown syndrome, which is so inti-mately intertwined with admission tomental hospital, can be largely prevented(24). This prevention is accomplished by a

community-oriented service providingcontinuing care from the same team withhospitalization held to the minimum.Surely, that which is provided is mostlysocial support, although some specificservices are included.

Moving on to hospitalization of chil-dren for tonsillectomy, there is a consider-able volume of clinical literature. Much ofthis was summarized by Jessner and hercolleagues back in 1952 (29). It indicatesthat supportive behavior on the part ofparents and staff is helpful in preventingpost-hospital psychological reactions. Re-calculation of Jessner's own data indicatesthat the simple provision, by the parents,of adequate information about the antici-pated hospitalization has a significant ef-fect in preventing postoperative reactions.

In concluding this section, we shouldtake note of the evidence that treating pa-tients with myocardial infarcts at homecarries no greater, and perhaps less, risk ofdeath than treating them in the hospitalintensive care unit. This is despite all theintensive care and fancy equipment that is



available in the hospital. Mather and hiscolleagues (38, 39) deserve congratula-tions for their foresight and courage in set-ting up this field experiment. Themechanism by which staying at home ex-erts its protective effect was presumablyidentified by Leigh et al. in 1972 (33),when they pointed to the association ofcardiac arrhythmias with high separationanxiety and the direction of hostility in-ward rather than outward. Both of thesepsychological sets are presumably re-duced in the supportive atmosphere of thehome. Obviously, network support is par-ticularly at issue. However, Engel (19)suggests that the effect may be due simplyto protection from environmental insults.

RECOVERY FROM ILLNESS

This section shows the importance ofthe supportive physician in recovery fromcongestive heart failure and from surgicalprocedures and the importance of psych-osocial assets in the recovery frompsychosomatic illness, especially tuber-culosis and asthma. Then the evidencethat social support keeps the patient intreatment and promotes compliance withprescribed regimens will be reviewed.

In 1953, Chambers and Reiser (10) de-scribed the association of emotionally sig-nificant events with the onset of episodesof cardiac failure. In addition, they de-monstrated the extraordinarily beneficialeffects that emotional support from thephysician could have on the course of thedisease. A related finding is reported byEgbert et al. (18). They took two compara-ble groups of surgical patients. One groupwas given special supportive care by theanesthetist and the other served as a con-trol. The surgeons managing the patientsdid not know which patient was in which

group. The patients in the special caregroup needed substantially less medica-tion for pain and were discharged on theaverage 2.7 days earlier than the controlgroup. Both findings were statisticallyhighly significant.

This same kind of effect must be demon-strable for a variety of other conditions in-volving fragile equilibrium. It points to thefact that social support is an importantcomponent of the therapeutic process. AsFrancis Peabody (47) stated in his essayThe Care of the Patient, "One of the essen-tial qualities of the clinician is interest inhumanity, for the secret of the care of thepatient is in caring for the patient." Morerecently, Lambert (32) and Caplan (9)have emphasized the importance of socialsupport in the psychiatric management oflife crises.

Some years ago, Berle, working withHarold Wolf and others (4), developed anindex of social and psychological charac-teristics of the patient, which had substan-tial prognostic value with regard to recov-ery from psychosomatic illness. Over halfof the score on this index is easily codableto the categories of social support enumer-ated above. Holmes et al. (26) showed thatthis scale was highly predictive of the out-come of sanatorium treatment for tuber-culosis, in that all the treatment failureswere in the lowest third of the scores onthis index. This is consonant with the evi-dence reviewed by Chen and Cobb (11),suggesting that tuberculosis is a disease ofsocial isolation, i.e., low social support.

More recently, de Araujo and van Ars-del, working with Holmes and Dudley(16) in Seattle, used this Berle Index toshow a remarkable interaction of socialsupport with life change with respect tothe need for steroid therapy in adult as-thmatics. Table 4 summarizes their data.The figures in the cells are average daily


SIDNEY COBB

TABLE 4. Average Daily Steroid Dosage in Millig-rams per Day for Patients with Asthma by LifeChange Score and Social Support (Berle Index), [de

Araujo et al. (16)]

Life change score

Berle Index

High Low

HighLow

5.6 (12)a

5.0(10)19.6 ( I I ) 6

6.7 (4)

a Number of cases.b This cell is significantly different from each of the

other cells (P < 0.01).

doses of steroids (prednisone or equival-ent). It is clear that those with low lifechange scores needed only small doses ofsteroids and that those with a lot of lifechanges and a low Berle Index neededthree to four times as much. The importantpoint about this table is that those with ahigh Berle Index were protected from theneed for high doses of steroids that pre-sumably would be generated by a high lifechange score. It is this interaction betweenlife change and support, in a way that sug-gests a protective effect of support, that isthe focus of this review.

There are a lot of pathways to the abovementioned effects. They fall into two majorclasses. The first is a direct effect throughneuroendocrine pathways. This is theone that we in psychosomatic medicineare most apt to emphasize. The second isthrough promoting compliant behavior onthe part of the patient. There is a largequantity of very consistent evidence point-ing to the fact that those patients who arenot socially isolated and are well sup-ported are, in our casual lingo, good pa-tients, in that they stay in treatment andfollow our recommendations. The evi-dence on dropping out of treatment issummarized by Baekeland and Lund wall(3), who state, "The importance of socialisolation and /or lack of affiliation was in-

dicated in 19 out of 19 studies (100%) thatconsidered them." In a review of com-pliance with therapeutic regimens thatdoes not overlap with the above, Haynesand Sackett (25) indicate that only one of22 articles in which social support-relevant variables were measured gaveevidence contrary to the hypothesis thatsocial support is positively associatedwith compliance. Six of the 22 articlesfound no evidence either way. As data onsuch matters go, this association ofcooperative patient behavior with variouscomponents of the social support complexis one of the best established facts aboutthe social aspects of medical practice.

LIFE STRESS

Here data will be presented on the dan-gers of stopping drinking without socialsupport and the way in which social sup-port protects against depression in the faceof extensive life changes. After that, wewill go on to two particular life changes:job loss and bereavement.

Joan Jackson, in a chapter in Sparer'sbook Personality, Stress and Tuberculosis(28), presented some truly remarkabledata, the significance of which did not re-ceive adequate recognition at the time. Ihave recalculated her data in Table 5. Shecompared the men sent to the police farmfor alcoholism with the alcoholics admit-ted to the tuberculosis sanatorium with re-spect to the frequency of men attemptingto stop drinking in the preceding year withand without the support of an institution-ally based program and /or AlcoholicsAnonymous. As you can see, the frequencydistributions are highly significantly dif-ferent (P<0.005).Therelativerisk calcula-tion says in essence that men who tried tostop drinking on their own, i.e., withoutthe support of an organized program, had



TABLE 5. The Risk of Tuberculosis to Men Stopping Drinking in the Preceding 12 Months without SocialSupport [Recalculated from Data of Jackson (28)J

Stopped drinking during past 12 monthsWithout supportWith support

Did not stop drinking

Total

X2 = 14.8; P < 0.005; relative risk = 1 8 X 3 3 = 20.

Alcoholic admissions to

Police farm

15

28

34

TB Sanatorium

182

27

47

TABLE 6. Percentage of a Random Sample WhoHad a Recent Onset of Affective Disorder by life

Stress and Social Support [Brown et al. (7)]

Life stress

Severe event ormajor difficulty

Neither

Social support

Confidant

4% (2/45)

1% (1/82)

No confidant

38% (17/45)a

3% (1/34)

a Significantly different from each of the other cells (P< 0.001). Interaction: t = 3.60, P < 0.001.

20 times the likelihood of being admittedto the tuberculosis sanatorium as theirpeers who did not try to stop or who triedwith support. In fact, the numbers suggestthat the risk of tuberculosis is reduced forthose who try with support, but they aretoo small to be convincing. This is an im-pressive difference and deserves attentionfrom those who work with alcoholics.Again, the social support dimensionemerges as especially important for tuber-culosis.

Turning to the work of George Brownand his collaborators (7), one findsanother striking datum. In Table 6 isshown the percent of a random sample ofwomen classified as having a severe affec-tive disorder. They are divided by whetheror not they had a confidant. A confidant

was defined as a person, usually male,with whom the woman had "a close, inti-mate and confiding relationship." Thetable makes it clear that those women whohad severe events and lacked a confidantwere roughly 10 times more likely to bedepressed than those in any of the otherthree cells. If one assumes that the eventshad some causal relationship here, one isforced to the conclusion that the intimaterelationship is somehow protective.Although one must be cautious about in-terpreting the effect of life events on de-pression because depressed people over-report unpleasant events, I have seen noevidence that depressed people tend todeny their confidants.

Alcoholism is related to depression, soit is not surprising that Quinn (50) foundthat escapist drinking, but not other formsof drinking, is significantly elevated onlyamong those who have high job stress andare not supported by their supervisors.These data come from the National Qualityof Employment Survey, 1972-73 (50).

EMPLOYMENT TERMINATION

This is a major life crisis for most men,particularly if they have dependents andhave been stably employed for some years.


SIDNEY COBB

My colleagues and I have made a specialstudy of this matter. The doctoral disserta-tion of Susan Gore (23) was specificallyfocused on the moderating effect of socialsupport on a selected set of outcome vari-ables. Our total study examined a widevariety of economic, social, psychological,and medical variables in 100 men whosejobs were abolished and 74 men whoseemployment was stable. The men were vis-ited by public health nurses before thetermination, soon after the termination,and then at 6,12, and 24 months after theplant closings. The men were all marriedblue-collar workers with about 19 years ofseniority. In this study, the measure fo-cused almost entirely on network support,although there was one item on subjectiveemotional support. This measure hadmoderating effects on some physiologicalvariables and some indicators of illness,but not with regard to others. Cholesteroland uric acid levels in the serum werehigher during the weeks surrounding thetermination for those with little supportthan for those with adequate support. Nosuch differences were observed with re-spect to norepinephrine in the urine orcreatinine in the serum, although impor-tant changes over time were noted withrespect to each and both were modified bythe level of psychological defense (12).The changes in level of complaints notedon a symptom check list was significantlymoderated by high social support, but noeffect was noted on hypertension or, sur-prisingly enough in view of the findings ofCobb et al. (13), on peptic ulcer. At theother end of the social support scale, thefinding that marital hostility is associatedwith ulcer disease was replicated.

When it came to arthritis, the data pre-sented in Table 7 emerged. Here it can beseen that there was a 10-fold increase inthe proportion of men found to have two or

TABLE 7. The Effect of Sodal Support in Prevent-ing Joint Swelling in Relation to Job Loss

Two or more jointswith observed swelling

Other

Total

Percent with two ormore joints swollen

Social support

High Med

1

27

28

4%

5

36

41

12%

Low

12

17

29

41%

Total

18

80

98a

aTwo cases have missing data.y = 0.732; P < 0.0003.

more joints swollen as one went from thehighest to the lowest quartile of the socialsupport dimension. This was not pre-dicted in advance, so the matter was ex-amined with some care and the findingstands up no matter how you look at it.

BEREAVEMENT

Everyone acts as though social supportwere important to those who are bereavedand many authors suggest that this shouldbe appropriate behavior, but hard evi-dence on the subject seems difficult tocome by. Parkes (45) presents some sug-gestive evidence that opportunities for af-filiation and affiliative behavior correlatepositively with good psychological state13 months after bereavement. The moststriking data are provided by Burch (8).Recalculating her data, it is possible toconclude that a married man who lost hismother had no increased probability ofsuicide. However, if he were single or hismarriage had been terminated, his risk ofsuicide was increased ninefold by thedeath of his mother. Dr. Burch also showedthat those men who had less contact with



relatives had a greater probability ofsuicide than those with more contact. Con-sidering the importance of the questionand the potential of the study design, theanalyses presented by Gerber (22a) are dis-appointing. One wishes that the data weremore fully presented and more approp-riately analyzed. However, on four of thesix health dimensions presented, the be-reaved subjects who had received profes-sional support were better off during theperiod 5-8 months after bereavement.Further research in this area is clearly in-dicated.

AGING AND RETIREMENT

In this area, as in hospitalization of chil-dren, there are a lot of strong impressionsabout the importance of social support inprotecting people from the consequencesof the stress of growing old and infirm. Thebest data that have come to hand are thoseof Blau (5). She supports the view thatbeing married or being employed or hav-ing substantial social activity ("participa-tion") is protective against the develop-ment of low morale. Similarly, Lowenthaland Haven (36) find in a sample of 280persons aged 63 and older that 85% ofthose with low social interaction were de-pressed, whereas only 42% of those withhigh social interaction were depressed.

THREAT OF DEATH

Life threats are most striking in battleand as life's end approaches. These twosituations will be examined in turn. Thereare many studies of the effects of morale,which is presumably a derivative of social

support on the frequency of neuro-psychiatric disorders. Rose's (52) ratherclean study illustrates the point. He com-pared two battalions from the same regi-ment with respect to morale and neuro-psychiatric casualties. These battalionswere otherwise quite similar. The battal-ion with high morale had roughly half ashigh a rate of psychiatric casualties as thebattalion with low morale. To the extentthat high morale involves high self-esteemand group cohesiveness, mutual esteemsupport and network support are central tothe maintenance of morale.

In a similar vein, Swank (55) found thatevery soldier in the Normandy campaignwho had lost 75% or more of his buddiesdeveloped combat exhaustion. Reid (51)reports almost identical findings forbomber crews in the Royal Air Force. Justto clinch the matter firmly, I quote from themilitary experience of two distinguishedpsychiatrists. Francis Braceland (6) said,"It became obvious early in the course ofthe war that the most important prophylac-tics against psychiatric casualties in themilitary forces were proper individualmotivation and high morale...." WilliamMenninger (40) added, "We seemed tolearn anew the importance of group ties inthe maintenance of mental health."

It would be improper to close this re-view without drawing attention to the lifesparing effects of anticipated ceremonialoccasions. Phillips and Feldman (48) in atruly remarkable paper showed in five dif-ferent populations that deaths are reducedin the 6 months preceding birthdays andincreased in the succeeding 6 months. Asummary of their data is presented in Fig-ure 1. They went on to hypothesize that, ifthis were a social support effect, it shouldbe more striking for the most disting-uished. This they found to be dramaticallyconfirmed.


SIDNEY COBB

Fig 1. Number of deaths before, during, and afterbirth month (all five samples combined).

DISCUSSION

We have seen strong and often quitehard evidence, repeated over a variety oftransitions in the life cycle from birth todeath, that social support is protective.The very great diversity of the studies interms of criteria of support, nature of sam-ple, and method of data collection isfurther convincing that we are dealingwith a common phenomenon. We have,however, seen enough negative findings tomake it clear that social support is not apanacea.

The conclusion that supportive interac-

tions among people are important ishardly new. What is new is the assemblingof hard evidence that adequate social sup-port can protect people in crisis from awide variety of pathological states: fromlow birth weight to death, from arthritisthrough tuberculosis to depression, al-coholism, and other psychiatric illness.Furthermore, social support can reducethe amount of medication required and ac-celerate recovery and facilitate com-pliance with prescribed medical regimens.

In a number of the studies that have beencited, it is possible to suggest alternativesto the social support explanation for the



findings. For example, the birth weightfindings of Morris et al. (42) might be dueto a reporting artifact, in that high schoolgraduate mothers appreciate the signifi-cance of low birth weight and thereforetend to over-report their small babies asunwanted, while the less well-educatedmothers do neither. Similarly, the excessof suicides among sons not currently mar-ried described by Burch (8) could con-ceivably be related to unresolved maternalties, rather than to a lack of social support.At the other extreme, such studies as thoseofNuckollsetal. (44),deAraujoetal. (16),Egbert et al. (18), or my own on job termi-nation are considerably less amenable toalternative interpretations because the so-cial support was measured in advance, thestress was defined, and the outcomes weresufficiently specific to avoid major con-founding. The crux of the matter seems tobe that, although one or two of the studiespresented might turn out on further inves-tigation to be truly irrelevant, the series isso long and so diverse that it demandsattention.

I may well have missed important find-ings both positive and negative. I wouldwelcome additions to my files from read-ers who can identify such omissions.However, it should be clear that I havefocused on the interaction of social sup-port with environmental stress and haveintentionally omitted a modest volume ofstudies that demonstrate a simple directeffect of social support on health. Thesestudies are mostly included in the reviewsby Gore (23), Kaplan et al. (30), and Pin-neau (49). The effect on tuberculosis isnoted by Chen and Cobb (11). The fullrange of data on coronary heart disease hasnever been assembled in one place. Thisdeserves attention, for the collective effectof the full set is impressive. These data arenot presented here because they are mostly

reported as main effects rather than asmoderating the effects of social stress.

What remains is to consider possiblemechanisms for the observed protective ef-fects. At the present time, the most attrac-tive theory about the nature of thisphenomenon involves pathways throughfacilitation of coping and adaptation.(Coping in my language means manipula-tion of the environment in the service ofself and adaptation means change in theself in an attempt to improve person-environment fit.) It would not be un-reasonable to suppose that esteem supportwould encourage a person to cope, i.e., togo out and attempt to master a problem.Likewise, emotional support and a sense ofbelonging might provide the climate inwhich self-identity changes can most read-ily take place.

As evidence that mastery of a new tasktakes place best under supportive condi-tions, I would draw attention to the classi-cal experiment of Coch and French (15) inthe pajama factory. They found that "par-ticipation" markedly reduced the timeneeded to get back to full production aftera change in the nature of the task. An invi-tation from management to participate inthe planning and implementation of achange is certainly direct esteem support.Participation may incidentally increaseother forms of support, but that is not cen-tral to the argument that esteem supportfacilitates coping.

Parsons' (46) study clarified the sickrole. Since successful role changes involveidentity changes, it is logical to presumethat movement in and out of the sick role isassisted by those things that facilitate therelevant identity changes. Clearly, socialsupport facilitated remaining in treatment(3) and may speed recovery (18). Thehypothesis is strong but, as far as I know,untested that social support facilitates


SIDNEY COBB

identity change, which, in turn, facilitatesrole change.

Further research on the proposedmechanism through which social supportmight operate is clearly indicated. In addi-tion, there is an obvious need for furtherinvestigation of the moderating effects ofsocial support on the consequences of thefollowing transitions: entry into primaryschool, entry into college, first job, mar-riage, residential change, and bereave-ment. Surely, investigation will proceedon the effect of social support on the out-come of medical treatment because the re-sults are likely to point to methods for re-ducing the costs of medical care. This re-view has focused on acute stress. Thereremains an important question as to

whether social support can moderate theeffects of chronic stress such as that ex-perienced by air traffic controllers (14).

There appears to be enough evidence onthe importance of social support to war-rant action, although, of course, all the de-tails as to the circumstances under whichit is effective are not yet worked out. Fol-lowing the behest of Stephen Smith (53)cited at the beginning of this review, weshould start now to teach all our patients,both well and sick, how to give and receivesocial support. Only in rare instances ofclear psychiatric disability should this in-struction require a psychiatrist. It seems tome that this is the real function for whichRichard Cabot designed the profession ofmedical social work.

REFERENCES

1. Angyal A: Neurosis and Treatment: A Holistic Theory (edited by E Hanfmann and RM Jones). New York,Wiley, 1965

2. Antonovsky A: Conceptual and methodological problems in the study of resistance resources andstressful life events, in Stressful Life Events: Their Nature and Effects, Chap. 15 (edited by BS Dohren-wend and BP Dohrenwend). New York, Wiley, 1974

3. Baekeland F, Lundwall L: Dropping out of treatment: a critical review. Psychol Bull 82:738-783,19754. BerleBB, Pinsky RH, Wolf S, Wolf HG: Berle index: a clinical guide to prognosis in stress disease. J Am

Med Assoc 149:1624-1628, 19525. Blau ZS: Old Age in a Changing Society. New York, New Viewpoints, 19736. Braceland FJ: Psychiatric lessons from World War II. Am J Psychiatry 103:587-593, 19477. Brown GW, Bhrolchain MN, Harris T: Social class and psychiatric disturbance among women in an urban

population. Sociology 9:225-254, 19758. Burch J: Recent bereavement in relation to suicide. J Psychosom Res 16:361-366, 19729. Caplan G: Support Systems and Community Mental Health. New York, Behavioral Publications, 1974

10. Chambers WN, Reiser MF: Emotional stress in the precipitation of congestive heart failure. PsychosomMed 15:38-60, 1953

11. Chen E, Cobb S: Family structure in relation to health and disease. J Chron Dis 12:544-567, 196012. Cobb S: Physiological changes in men whose jobs were abolished. J Psychosom Res 18:245-258,197413. Cobb S, Kasl SV, French JRP Jr, Norstebo G: The intrafamilial transmission of rheumatoid arthritis.

VII: Why do wives with rheumatoid arthritis have husbands with peptic ulcer? J Chron Dis 22:279-293,1969

14. Cobb S, Rose RM: Hypertension, peptic ulcer and diabetes in air traffic controllers. J Am Med Assoc224:489-492, 1973

15. Coch L, French JRP Jr: Overcoming resistance to change. Hum Rela 11:512-532, 194816. de Araujo G, van Arsdel PP, Holmes TH, Dudley DL: Life change, coping ability and chronic intrinsic

asthma. J Psychosom Res 17:359-363, 1973



17. Dytrych Z, Matejcek Z, Schuller V, David HP, Friedman HL: Children born to women denied abortion.Fam Plann Perspect 7:165-171, 1975

18. Egbert LD, Battit GE, Welch CE, Bartlett MK: Reduction of post-operative pain by encouragement andinstruction of patients. N Eng J Med 270:825-827, 1964

19. Engel GL: Psychologic factors in instantaneous cardiac death. N Eng J Med 294:664-665, 197620. Forssman H, Thuwe I: One hundred and twenty children born after application for therapeutic abortion

refused. Acta Psychiatr Scand 42:71-88, 196621. French JRP Jr, Rodgers W, Cobb S: Adjustment as person-environment fit, in Coelho GV, Hamburg DA,

Adams JE, Coping and Adaptation. New York, Basic Books, 197422. Fromm E: The Sane Society. New York, Rinehart, 195522a. Gerber I, Wiener A, Battin D, Arkin A: Brief therapy to the aged bereaved, in Bereavement: Its Psychosocial

Aspects, (edited by B Schoenberg et al.). New York, Columbia University Press, 197523. Gore S: The influence of social support and related variables in ameliorating the consequences of job loss.

Doctoral dissertation, University of Pennsylvania, 197324. Gruenberg E, Snow HB, B ennett CL: Preventing the social breakdown syndrome, in Social Psychiatry, Vol

47 (edited by FC Redlich). Baltimore, ARNMD, 196925. Haynes RB, Sackett DL: A Workshop /Symposium: Compliance with Therapeutic Regimens—Annotated

Bibliography. Department of Clinical Epidemiology and Biostatistics, McMaster University MedicalCentre, Hamilton, Ontario, 1974

26. Holmes TH, Joffe JR, Ketcham JW et al.: Experimental study of prognosis. J Psychosom Res 5:235-252,1961

27. Hultin M, Ottosson MO: Perinatal conditions of unwanted children. Acta Psychiatr Scand [Suppl]221:59-76, 1971

28. Jackson JK: The problem of alcoholic tuberculous patients, in Sparer PF, Personality Stress and Tuber-culosis. New York, International Universities Press, 1954

29. Jessner L, Blom GE, Waldfogel S: Emotional implications of tonsillectomy and adenoidectomy onchildren. Psychoanal Study Child 7:126-169, 1952

30. Kaplan BH, Cassel JC, Gore S: Social support and health. Paper presented at meetings of American PublicHealth Association, Nov 9, 1973

31. Kramer M, Pollack ES, Redick RW, Locke BZ: Mental Disorders /Suicide. Cambridge, Harvard UniversityPress, 1972

32. Lambert K: Agape" as a therapeutic factor in analysis. J Anal Psychol 18:25^16, 197333. Leigh H, Hofer MA, Cooper J, Reiser MF: A psychological comparison of patients in "open" and "closed"

coronary care units. J Psychosom Res 16:449-457, 197234. Leighton AH: My Name Is Legion. New York, Basic Books, 195935. Leighton AH: Come Near. New York, Norton, 197136. Lowenthal ME, Haven C: Interaction and adaptation: intimacy as a critical variable. Am Sociol Rev

33:20-30, 196837. MacLeod GK, Pressin JA: The continuing evolution of health maintenance organizations. N Eng J Med

228:439-443, 197338. Mather HG: Intensive care. Br Med J 2:322, 197439. Mather HG et al.: Acute myocardial infarction: home and hospital treatment. Br Med J 3:334-338,197140. Menninger WC: Psychiatric experience in the war 1941-1946. Am J Psychiatry 103:587-593, 194741. Moos RM, Insel PM: Issues in Social Ecology. Palo Alto, National Press Books, 197442. Morris NM, Udry JR, Chase CL: Reduction of low birth weight rates by prevention of unwanted pregnan-

cies. Am J Pub Health 63:935-938, 197343. Murray HA et al.: Explorations in Personality. New York, Oxford, 193844. Nuckolls KB, Cassel J, Kaplan BH: Psychosocial assets, life crisis and the prognosis of pregnancy. Am J

Epidemiol 95:431^141, 197245. Parkes CM: Bereavement: Studies of Grief in Adult Life. New York, International Universities Press, 197246. Parsons T: Definition of health and illness in the light of American values and social structure, in Patients,

Physicians and Illness (edited by EG Jaco). Glencoe Free Press, 195847. Peabody FW: The care of the patient. J Am Med Assoc 88:877-882, 1927


SIDNEY COBB

48. Phillips DP, Feldman KA: A dip in deaths before ceremonial occasions: some new relationships betweensocial integration and mortality. Am Soc Rev 38:678-696, 1973

49. Pinneau SR: Effects of social support on psychological and physiological strains. Dissertation, Universityof Michigan, 1975

50. Quinn RP: Personal communication, 197351. Reid DD: Some measures of the effect of operational stress on bomber crews, in Great Britain Air Ministry,

Psychological Disorders in Flying Personnel of the R.A.F. London, His Majesty's Stationery Office, 194752. Rose AM: Factors in mental breakdown in combat, in Mental Health and Mental Disorder—A Sociological

Approach (edited by AM Rose). London, Routledge and Kegan Paul, 195653. Smith S: On the limitations and modifying conditions of human longevity, the basis of sanitary work, in

Public Health Reports and Papers presented at the meetings of the American Public Health Association inthe year 1873. New York, Riverside, 1874

54. Stein Z, Susser M: Social factors in the development of sphincter control. Dev Med Child Neurol9:692—706, 1967

55. Swank RL: Combat exhaustion: a descriptive and statistical analysis of causes, symptoms and signs. JNerv Ment Dis 109:475-508, 1949

56. Weiss RS: The fund of sociability. Transactions 6:36-^3, 1969


social support as a moderator of life stress

Documents