Social
Psychiatry
Ii,
51-58
Social Psychiatry
(1976)
9 by Springer-Verlag 1976
Friends, Confidants and Symptoms P. McC. Miller and J. G. Ingham MRC
Unit for Epiderniological
Studies
in Psychiatry,
Royal
Edinburgh
Hospital,
Edinburgh
I
Summary._ Brown, Bhrolchain & Harris (1975) have found that intimate social relationship may afford some protection against the onset of depressive illness in women. Our investigation involved 337 subjects sampled from the list of one general practice. Half had consulted their doctor within the previous 7 days and half had not. It examined the association between social support and the severity of some psychological and physical symptoms. The psychological symptoms were tiredness, anxiety, depression and irritability; and the physical symptoms were backache, headache, palpitations, dizziness and breathlessness. Social support and symptoms were assessed in a home interview. The results supported those of Brown et al. in showing that women reporting the lack of an intimate confidant had psychological symptoms of significantly greater severity than those reported by their more adequately supported counterparts. However, in contrast to Brown et al., we observed that the absence of many casual, less intimate friends was also associated with higher symptom levels (both physical and psychological). The discrepancy might arise because few of our subjects were suffering from severe psychiatric illness. The results were less clear for men but showed the same trends.
There is now much evidence that life circumstances are related to illness, both physical and psychological. Most of the interest has been in areas where the possible influence of circumstances upon health or illness has been negative in nature. There is, for example, a substantial amount of evidence of a marked link between the occurrence of stressful life events and the onset of illness (Brown, 1972; Brown and Birley, 1968; Birley and Brown, 1970; Cooper and Sylph, 1973; Brown, Harris and Peto, 1973; Uhlenhuth and Paykel, 1973; etc.) However, not all the circumstances surrounding an individual and affecting his life are negative. Some of the likely positive ones are beginning to be invesligated. One group that has been identified as likely to favour positive health includes the various forms of social support available to the individual. Brown, Bhrolchain and Harris 1975, found that an in-
1 The authors are grateful to Doctors I. R. W. Alexander, M. Paterson and M.W. Whitley for their wholehearted support in this study.
timate (but not necessarily sexual) relationship between a woman and her husband or boyfriend reduces the likelihood of depressive illness following a severe life event or major difficulty. However, social support, life stresses and illness can all differ in degree and little is known about the interaction of these factors except for the more severe levels of stress and illness. Does social support help to minimise the effects of the ordinary stresses of everyday life? Can the diffuse support of many friends and acquaintances who are not particularly close to the stressed person also help in this, or is it only a close relationship with a trusted confidant that does the trick? Is there a general alleviating effect of social support upon symptoms irrespective of how they are produced? If the answers to these questions are yes, then one would expect some (negative) association between symptom severity and support. Some data on these last two variables were available from a sludy of symptom levels and distributions in a sample of G. P. consulters as compared with a sample of non-consulting controls matched for age and sex. This comparison is reported
52 in Ingham and Miller (1976). In the present study we shall concentrate on the relationships of nine symptoms - backache, headache, palpitations, dizziness, breathlessness, tiredness, anxiety, depression and irritability - to the number of casual friends a person believes he has and to whether or not he thinks he has at least one good confidant. We shall also examine whether these two forms of support are independent. Method The Samples. Two samples were selected from the patients of an Edinburgh practice. This practice, situated in an urban residential part of Edinburgh, had 4435 patients over 16, and contained people of all ages whose occupations covered the whole social spectrum. The largest proportion, however, were skilled artisans. The type of housing was also very mixed, ranging from tenement flats to detached owner-occupied houses and bungalows. The first sample, the consulters, had just visited their doctor with a new illness episode. This sample was drawn over a i0 week period, by interviewing all the patients attending surgery on one particular day each week. All those aged between 16 and 76 who had come to consult about a new illness episode were asked if they would be willing to be interviewed at home. Data were successfully gathered on 172 of the 186 consulters approached. Four patients refused interview, while another i0 were not included for other reasons, mostly because they were not seen within a time limit of one week from consultation. For each consulter interviewed, a control of the same age and sex, who had not visited the doctor for at least three months was approached. If no interview was obtained, another control was chosen and this process was repeated until a co-operative and available control was found. In order to obtain the necessary 172 controls, 222 people had to be sought. However, only 15 of these refused outright to be interviewed; the rest were not seen either because they had left the area, or could not be traced, or because they had consulted since selection, or (in three cases only) for some other reason. In the resultant samples, 20 % were aged over 61 and 34% were under 32. Thirty-six per cent were men and 64 % women. The social class compositions were similar, but there was a slight (non-significant) tendency for the controls to be of higher social class than the consulters. Variables and Procedure. The variables relevant to the present study consisted of nine
Social P s y c h i a t r y , Vot. 11, No. 2 (1976) symptoms, whether or not the subject had a close confidant, and a measure of the number of casual acquaintances available to the subject. Each symptom was measured in two ways, both of which utilized five statements indicating different levels of the symptom - e.g. for headache, the statements were: - "I never have headaches", "Very occasionally I have a slight headache", "Quite often I have a fairly bad headache", "I get a lot of severe headaches" and "I have constant very bad headaches that are almost unbearable". In the first method, the subject was presented with selected pairs of these statements and asked to say which of the statements in each pair was nearer to the truth for him personally, during the past month. According to his answers, he was placed on an 8point Guttman type scale, and it could also be seen whether or not he was answering consistently. Fuller details about this type of scale are to be found in Ingham (1965). In the second method, the same statements were spaced in order, at 40 rnm intervals along a straight line 200 rams long, with the least severe statement at 0 and the most severe at 160 rams. The subject was requested to place a mark anywhere on the line (between statements if he wished) to indicate how bad the symptom had been for him personally during the past month. His score was the distance of this mark from zero. The nine scales used in the present study covered backache, headache, palpitations, dizziness, breathlessness, tiredness, anxiety, depression and irritability. To find out about a possible confidant, the subject was first asked what he would do if he had a personal problem to cope with, such as a leaking roof, or somebody seriously ill, or not getting on well with people at work. If he failed spontaneously to mention talking it over with somebody, he was asked "Is ihere anybody you can talk things over with?" If the answer was yes, it was further elicited whether the most personal difficulties could he discussed with this p e r s o n s whether the person lived close at hand, and w a s reasonably available and w h e t h e r the potential confidant w a s believed to reciprocate by trusting the subject with personal problems. If all these conditions w e r e fulfilled, the subject w a s held to have a confidant. A confidant thus defined could be a spouse, a family m e m b e r or a close friend, and these possibilities w e r e not distinguished further. T h e subject w a s then asked whether he felt he k n e w m a n y people in his neighbourhood and at work. If he clearly felt that he had m a n y friends in both these environments he w a s classed as having m a n y acquaintances, if in only
P.
McC.
Miller
and J.G.
Ingham:
Table
i. Numbers
Friends,
of men
Confidants
and women
and Symptoms
in the various
support
53
categories
Many acquaintances
Some acquaintances
Few acquaintances
Total
31
52
ii
94 (43.1%)
29
74
21
124 (56.9 %)
32 (14.7 %)
218
Women Good
confidant
No good
confidant
Total
60 (27.5 %)
126 (57.8 %)
Men
Table
Good confidant
22
20
5
47 (39.5 %)
No good confidant
32
35
5
72 (60.5%)
Total
54 (45.4%)
55 (46.2%)
Chi-square
for women
= 2.94
(N.S.
Chi-square
for men
= 0.77
(N. S. )
2. Standard
score
means
i0 (8.4%)
119
P<0.3)
on psychological
symptoms
for women
Many acquaintances
Some acquaintances
Few acquaintances
Tiredness
Confidant No confidant
-0.176 (N=31) 0.040 (N=29) -0.068*
-0.186 (N=52) 0.016 (N=74) -0.085
-0.164 (N=ll) 0.691 (N=21) 0.264
-0.175" 0.249
Anxiety
Confidant No confidant
-0. 161 0.078 -0.042
-0. 198 0.050 -0.074
-0.223 0.562 0.170
-0.194 0.230
Depression
Confidant No confidant
-0.103 0.035 -0.034
-0. 250 0.034 -0.108
-0.035 0. 620 0.293
-0.129 0.230
Irritability
Confidant No confidant
-0. 148 0. 176 0.014
- 0.151 -0.038 -0.095
0.121 0.419 0.270
-0.059 0.186
Confidant No confidant
-0. 587 0.328 -0.130
-0. 784 0.063 -0. 361
-0.301 2.292 i. 000
-0.557 0.894
Total Psychological Symptoms
*Figures underlined 0.186 - 0.164)
are means
of means;
e.g.
-0.068
= pl. (-0.176 + 0.040) and -0.175
= ~i(-0"176
54 Table
Social Psychiatry, 3. Standard
score means
on physical
symptoms
Vol.
Ii, No.
2 (1976 )
for women
Many acquaintances
Some acquaintances
Few acquaintances
Backache
Confidant No confidant
-0.072 (N=31) -0.060 (N=29) -0.066*
-0,079 (N--52) 0.018 (N=74) -0.031
-0.346 (N=II) 0.505 (N=21) 0.080
-0.166" 0.154
Headache
Confidant No confidant
-0.051 0.192 0.071
-0.091 -0.047 -0,069
-0.180 0. 294 0.057
-0.107 0,146
Palpitations
Confidant No confidant
0.060 -0.042 0.009
-0.164 -0.090 -0.127
-0.056 0.721 0.333
-0.053 0.196
Dizziness
Confidant N o confidant
-0.085 0.014 -0.036
-0.097 -0.060 -0.079
0. 128 0.491 0.310
-0.027 0.148
Breathlessness
Confidant No confidant
-0.160 0.035 -0.063
-0. 225 -0.064 -0.145
0. 245 0.669 0.457
-0.047 0.213
Confidant No confidant
-0. 307 0,139 -0.084
-0. 657 -0.243 -0.450
-0. 209 2.680 1.236
-0.391 0.859
Total Physical Symptoms
*See footnote to Table
Table 4. Standard
2.
score means
on psychological Many acquaintances
symptoms
for men
Some aquaintance s
Few acquaintances 0.116"
Tiredness
Confidant No confidant
-0.059 (N=22) -0,175 (N=32) -0.117"
0,076 (N=20) -0.053 (N=35) 0.012
0.330 (N=5) i. 114 (N=5) 0.722
b.295
Anxiety
Confidant No confidant
-0. i00 -0.051 -0,076
-0.070 -0.064 -0.067
0.833 0.659 0.440
~.181
Depression
Confidant No confidant
-0.114 0.051 -0.032
-0. 314 0,116 -0.099
0. 300 0.325 0.313
Irritability
Confidant No confidant
-0.020 -0.177 -0.099
-0.032 0.052 0.010
0.436 0.587 0.512
Confidant No confidant
-0. 300 -0.352 -0.326
-0. 340 0.051 -0. 145
i. 898 2.685 2.292
Total Psychological Symptoms
*See footnote to Table 2.
0.221
-0,043 0.164
0.128
o.154 0.419 0.795
P.
McC.
Miller
and J.G.
Ingham:
Friends,
Confidants
one of them, some acquaintances, and if in neither, few acquaintances. These data were gathered as part of a home interview lasting about one hour, and taking place, for the consulters, within one week of their surgery visit. Symptom levels were assessed at the start of the interview, the questions concerning social support were asked towards the end. Analysis. Mean symptom levels, as assessed by both methods, were computed for each of the six possible combinations of the acquaintance and confidant variables for men and women separately in both samples (consulters and controls). On the whole, mean symptom levels were higher in the consulters (see Ingham and Miller) but the trends within the samples and the methods were otherwise similar. Accordingly the samples were combined (with men and women still kept separate) and the two method scores for each subject were converted to standard scores and averaged. Two-way analyses of variance were then done on these scores for each symptom separately with the two types
Results Table 1 shows the relationship between the acquaintance and confidant variables in the combined samples. There is no evidence of an for m e n Some acquaintances (N=20) (N=35)
0.621 -0.321 0.150
(N=5) (N=5)
0.224* -0.137
Confidant No confidant
Headache
Confidant No confidant
0.009 0.045 0.027
-0.198 -0.042 -0.120
0.706 0.057 0.382
0.172 0.020
Palpitations
Confidant No confidant
0.066 0.070 0.068
-0.046 -0.002 -0.024
-0. 364 -0. 173 -0.269
-0.115 -0.035
Dizziness
Confidant No confidant
0.076 -0.089 -0.007
-0.117 -0.085 -0.i01
1.065 -0.153 0.456
0.341 -0.109
Breathlessness
Confidant No confidant
0.352 -0.215 0.069
0.026 -0.014 0.006
-0.076 -0.105 -0.091
0.i01 -0.iii
Confidant No confidant
0.566 -0.345 0.iii
- 0. 349 -0.075 -0.212
i. 953 -0.695 0.629
0.723 -0.372
*See footnote to Table 2.
-0.013 0.067 0.027
Few acquaintances
Backache
Total Physical Symptoms
0.063 (N=22) -0.156 (N=32) -0.047*
55
of support as independent variables and using the method of least squares to allow for the unequal n u m b e r s in the cells. Similar analyses w e r e done on the s u m m e d scores for five s y m p t o m s which s e e m e d to be m o r e physical and loealised0 i.e. backache, headache, palpitations, dizziness and breathlessness (physical s y m p t o m s ) and on the s u m m e d scores for the four remaining s y m p t o m s , i.e. tiredness, anxiety, depression and irritability (psychological symptoms). Covariance adjustments to allow for the possible effects of imbalances in age and social class w e r e considered, but in view of the near zero correlations of these variables with the s y m p t o m s , it w a s decided that such adjustments w e r e unnecessary. Seven subjects had to be excluded f r o m the analyses because data for t h e m w a s incomplete.
Table 5. Standard score m e a n s on physical s y m p t o m s Many acquaintances
and Symptoms
56
Social Psychiatry,
association between them. A somewhat higher proportion of men than women have many acquaintances. Mean symptom levels, expressed as standard scores are given in Tables 2-5 and analysis of variance results are set out in Table 6. For the women, having a good confidant appears to be associated with lower levels of tiredness, anxiety and depression. The acquaintance variable reaches significance for tiredness, depression, palpitations and breathlessness; in each case having few acquaintances accompanies higher symptom levels. When the symptoms are combined into psychological and physical symptoms, significant results are obtained on acquaintances for the physical symptoms and on both types of support for the psychological symptoms. For the men the trends among the four psychological symptoms are similar to those
Table
6. Analysis
of variance
Vol. 11, No. 2 (1976)
for the women, but probably owing to smaller numbers, fewer clearcut results are obtained. However, for tiredness, anxiety and the combined psychological symptoms, having few acquaintances is once again associated with a higher symptom level. For the physical symptoms no significant results are apparent.
Discussion This type of enquiry, which is essentially correlational in nature, can only show that an association exists between two sets of variables. The precise nature of the relationship remains to be determined. Brown et al. (1975) claim to have shown that severe life events and longterm difficulties play a causal role in bringing about depressive and other affective disorders in women. They also suggest that there are a number of factors, including lack of a close
F-values
C 0 nfi dant ! " No confidant
Confidant x acquaintances
Ac quaint anc e s
Men df=l,ll3
Women df=l,212
Men dr=2,113
Women df=2,212
Men df=2,113
Women df=2,212
Tiredness
0.06
7.33***
3.72**
4.13"*
1.16
2.09
Anxiety
0.003
6.15"*
3.38**
1.54
0.06
1.04
Depression
2.54
5.63**
0.79
3.31"*
0.38
0.82
Irritability
0.01
2.42
2.08
2.26
0.27
0.31
Total Psychological Symptoms
0.15
7.49***
3.23**
3.85**
0.12
1.08
Backache
0.68
1.82
0.29
0.82
1.26
2.16
Headache
0.29
1.56
1.22
0.82
0.80
0.71
Palpitations
0.11
0.94
0.59
5.02***
0.09
2.40
Dizziness
0.97
0.60
1.42
2.83
1.94
0.36
Breathlessness
2.62
2.74
0.92
6.37***
1.14
0.21
Total Physical Symptoms
0.68
2.75
0.20
4.80***
0.90
1.70
** P < 0.05 *** P < 0.01
P. M c C .
Miller and J.G. Ingham: Friends, Confidants and S y m p t o m s
57
acquaintance variable is held constant, there relationship with a husband or boyfriend, which also appears to be s o m e tendency for unemployincrease the chances of developing a psychiatric ed people to have higher s y m p t o m levels. The disorder when a serious life event or major best conclusion s e e m s to be that both lack of a long-term difficulty is present. In support of job and lack of casual friends are probably this position they divide their sample into those associated with higher s y m p t o m levels. At most, who have an intimate relationship with a husthe u n e m p l o y m e n t variable could only account band or boyfriend and those who do not, and infor part of the differences found. In any case, to those who suffer a severe life event or difas these two variables appear to be intrinsically ficulty and those who do not. They then show confounded and there is no obvious w a y of saythat the one group in which disorder is very ing which is the m o r e fundamental, the authors frequent comprises people who have no intimate feel that no modification of the m a i n conclusions relationship and who also suffer a severe event is necessary. or major difficulty. In the other three groups The interaction between social support and of their sample, disorder was rare. life events can be seen m o r e clearly in a subIn the present study we have been concerned sample of 68 people to w h o m w e administered with both close support (similar but not identian interview based on Brown's procedure for cal to the index of intimate relationship used by the assessment of life events. Because n u m b e r s Brown et al. ) and the more diffuse support are small, firm inferences are again not poswhich stems from having many friends and sible but these data suggest that (i) serious life acquaintances. The results cited are consistent with Brown et alts. position in that lack of a events tend to raise psychological s y m p t o m good confidant is associated with the presence scores, (2) having a good confidant and (partiof symptoms in women, especially symptoms cularly) having at least s o m e acquaintances m a y we know to be associated with depressive illafford partial, and only partial, protection ness. Brown et al. also state, however, that against this rise, (3) psychological symptom "there is no suggestion that any other kind of levels probably vary with the social support confidant, no matter how often seen, serves to variables even when there is no serious life modify the impact of a severe event or difficulevent present. ty". In contrast, in our sample there appears to At first sight these findings and tentative be a lower level of psychological symptomatoloconclusions might be thought to conflict with gy in both sexes, among those who say that they those of Brown et al. who found that an intimate know at least several people, if only at a superrelationship afforded virtually complete protecficial level either at home or at work. This tion against onset of depressive illness and variable seems to be particularly important for that no other relationship was much help. Howf the men, but less important than the confidant ever, although no psychiatric assessment was variable for the women (Table 6). made, we eanbe reasonably sure that few, if A possible difficulty arises in its interpretaany, of our sample were suffering from frank tion, however, in that it is to some extent conpsychological illness of recent onset. founded with whether or not the subject is in Although we have at present no direct employment. Subjects cannot be rated as having knowledge of the relationship between Brown's "many acquaintances" unless they are workin'g concept of onset and our symptom scale scores, and subjects who are not employed have a in a sample of psychiatric outpatients of both greater chance of being classed as having "few sexes (total n = 86) the average psychological acquaintances". This confounding is not entirely symptom score, computed using the parameters an artefact of our measuring procedure. It of the present (G. P. + control) sample in the would almost certainly exist no matter how one standardisation process, was 5.37 (as against' attempted to measure number of acquaintances. 0.0 for the present sample). Therefore it would However, it is of interest to see to what extent seem that most of our subjects were probably a being in employment might explain the result good deal less disturbed than the depressed obtained. Fifteen men and 77 women were not subjects of Brown et al. The results of both working at the time of interview. Although ours and their studies might be explained on the following model: numbers are too small to permit clear conclusions, the results for the acquaintances variPeople differ in the degree of social support able seem to hold, by and large, amongst emwhich they enjoy. Those that have lesser degrees ployed men taken alone, employed women taken of close and diffuse support tend to have a alone and unemployed women taken alone. There slightly raised level of psychological symptomaare too few unemployed men for meaningful tology and to be more vulnerable to life stress. comparisons. On the other hand, when the -The onset of depressive illness is probably
58 best conceived as a vicious spiral in which a life event may occur which depresses the person somewhat and both because of this depression and because of the effect on other circumstances, makes another life event more likely, and so on until a breakdown occurs. (Examples of such spirals would be loss of spouse followed by mild depression leading to loss of job due to poor work followed by further depression and finally onset of a full-blown depressive illness; or a mild depression leading to careless driving and a serious accident followed by further depression and the loss of one's job as a van driver followed by further depression and illness onset. ) In the early stages of this malevolent process any type of social support is helpful in reducing symptom levels. As it proceeds, however, the only effective help is that provided by a close confidant. The spiral may, and often does, have only one twist. This would correspond to the formulation by Brown et al. that a serious life event causes depression in L the absence of favourable protective circumstances. However, we would contend that they have, at most, established only that life events sometimes cause depression. They have not (and do not claim to have) ruled out the possibility that in some cases depression may come before the life event. In our view which causes which is often difficult to resolve for an individual instance. The association between life events and psychiatric illness is, however, quite clear. Further research effort should now be directed towards a careful elucidation of the exact nature of this association. References Birley, J. L. T., Brown, G.W. : Crises and life changes preceding the onset or relapse of acute schizophrenia: Clinical aspects. Br. J. Psychiat. 116, 327-333 (1970)
Social Psychiatry,
Vol.
ii, No.
2 (1976)
Brown, D.G. : Stress as a precipitant factor in eczema. J. Psychosornat. Res. 16, 321328 (1972) Brown, G.W., Birley, J.L.T.: Crises and life changes and the onset of schizophrenia. J. Health Soc. Behav. 9, 203-214 (1968) Brown, G.W., Bhrolchain, M.N., Harris, T.O.: Social class and psychiatric disturbance among women in an urban population. Sociology 9, 225-254 (1975) Brown, G.W., Harris, T.O., Peto, J.:Life events and psychiatric disorders. Part 2: Nature of causal link. Psychol. Med. 3, (1973) Cooper, B. , Sylph, J. : Life events and the onset of neurotic illness: an investigation in general practice. Psychol. Med. 3; 421-435 (1973) Ingham, J.G. : A method for observing symptoms and attitudes. Br. J. Soc. Clin. Psychol. 4, 131-140 (1965) Ingham, J.G., Miller, P. McC. : The concept of prevalence applied to psychiatric disorders and symptoms. Psychol. Med. (1976) in press Uhlenhuth, E.H., Paykel0 E.S.:Symptom intensity and life events. Arch. Gen. Psychiat. 28, 473-477 (1973)
Dr. P. McC. Miller M R C Unit for Epidemiological Studies in Psychiatry Royal Edinburgh Hospital Morningside Park Edinburgh, EHI0 5 H F Scotland, U.K.