Influence of Family Life on the Course of Schizophrenic Disorders:

A Replication
G. W. BROWN, J. L. T. BIRLEY and J. K. WING
BJP 1972, 121:241-258.
Access the most recent version at DOI: 10.1192/bjp.121.3.241

References This article cites 0 articles, 0 of which you can access for free at:
http://bjp.rcpsych.org/content/121/562/241.citation#BIBL
Reprints/ To obtain reprints or permission to reproduce material from this paper, please write
permissions to permissions@rcpsych.ac.uk

You can respond http://bjp.rcpsych.org/cgi/eletter-submit/121/562/241

to this article at
Downloaded http://bjp.rcpsych.org/ on May 24, 2012
from Published by The Royal College of Psychiatrists

To subscribe to The British Journal of Psychiatry go to:

http://bjp.rcpsych.org/site/subscriptions/
Brit. @7.Psychiat. (1972), 121, 241—58

Influence of Family Life on the Course of

Schizophrenic Disorders: A Replication
By G. W. BROWN, J. L. T. BIRLEY and J. K. WING

In spite of the new methods of treatment and ment (Brown and Birley, 1968). These facts,
care introduced during the past fifteen years, together with the contrasting but just as
schizophrenic patients are still liable to relapse handicapping reaction of schizophrenic patients
with a recurrence of florid symptoms such as to an under-stimulating social milieu, were
delusions, hallucinations and disturbed beha brought together in a more general theory of
viour, and great suffering can be caused to all environmental influences on the course of
conccrned (Brown et al., 1966). It has been schizophrenia (Wing and Brown, 1970). This
shown that the onset of florid symptoms is often also took account of the high physiological
preceded during the previous three weeks by arousal which had been found in the most
a significant change in the patient's social withdrawn schizophrenic patients (Vcnables,
environment (Brown and Birley, i g68; Birley ig68; Venables and Wing, 7962). It was argued
and Brown, 1970). Other studies have focused that in a socially intrusive environment acting
on the influence of more persistent environ upon a patient whose thought disorder was in
mental factors, such as the emotion expressed any case liable to become manifest whenever
towards the patients by relatives with whom circumstances became too complicated, a
they were living. In an exploratory survey of patient would tend to attempt to protect
discharged long-stay men it was found that himself by social withdrawal; but this process
close emotional ties with parents or wives indi might easily go too far, both in hospital and
cated a poor prognosis (Brown, Carstairs and outside it, leading to complete social isolation
Topping, 1958; Brown, 1959). In a further and inability to care for himself. The optimum
study, patients were seen in hospital just before social environment, for those who remained
discharge, and their relatives were interviewed handicapped, was seen as a structured and
at home at the same time, and both were seen neutrally stimulating one with little necessity
together at a joint interview shortly after for complex decision making.
discharge. It was found that those patients Many questions, however, were left un
who returned home to live with relatives who answered by this work. What are the com
were highly emotionally involved with them ponents of ‘¿ emotional involvement'? Is the
(as judged by ratings of the relatives' behaviour) patient's earlier disturbed behaviour the cause
were more likely to suffer a relapse of florid of the relative's emotional reaction and an
symptoms, even when the severity of psychiatric indicator of future chances of symptomatic
disturbance at the time of discharge was taken relapse? In other words, is the relative's emo
into account (Brown et al., 1962). Ratings of the tional response without causal significance;
patient's own expressed emotion showed much simply a reflection of underlying processes in
less involvement, and were much less highly the patient? Alternatively, can a highly
associated with subsequent relapse. There was emotional reaction actually cause relapse,
also a suggestion that short-term and long-term irrespective of the patient's previous degree of
influences might have a cumulative effect; for disturbance? How important are other factors
example, that a raised level of tension in the such as sex, marital status, therapeutic medica
home made relapse more likely in the event of tion, patient's attitudes, amount of contact
a critical change in the patient's social environ between patient and relative or between
241

This One
242 INFLUENCE OF FAMILY LIFE ON THE COURSE OF SCHIZOPHRENIC DISORDERS

relative and other family members, the occur to Bexley Hospital and were within five years of their
rence of environmental changes or crises, etc., first admission were also included.
in increasing or decreasing the chances of In this way, i i8 patients were selected but I 7 had
relapse? to be excluded subsequently for the following reasons:
6 patients did notwish to participate (2 at admission,
A further study was therefore designed in
4 at follow-up);
which detailed attention was paid to techniques idied in hospital;
of measurement. The instruments used to 5 remained in hospital for over one year and there
measure expressed emotion and attitudes have fore could not be followed up at home;
been described elsewhere (Brown and Rutter, 5 did not return home after discharge from hospital.
1966; Rutter and Brown, 7966). The methods The 101 patients finally included in the series were
attempt to avoid the well-known difficulties of distributed as follows:
measuring the quality of interpersonal relation Camberwell Bexley
ships by responses to standard questions. An area Hospital
experimental design, in which patients would First episode i6 II

be allocated at random to homes with high or One or more episode during

low emotional tension, was clearly out of the 5 years preceding key
admission 77 22
question, and so a prospective follow-up design
First episode more than 5
was used, in which the measurements of past years before key admission 35 —¿
behaviour, present emotional response and
future relapse were made independently of The first patient was seen in February 1966, and
each other. Alternative models were developed the final follow-up interview was completed in
as a basis for the data analysis (Rosenberg, August 7968.
1968). Since each patient in the series already The patients and their families were seen on several
occasions by members of the research team (two
had an established schizophrenic illness, the
psychiatrists and three sociologists). Eight types of
study was not intended to be concerned with interview were carried out for each patient and
the original causes of this condition but only family, and ten if the patient was readmitted in the
with factors influencing its course. follow-up period. Two interviews to establish the
The hypothesis under test is that a high current mental state of the patient and his social and
degree of expressed emotion is an index of clinical background were carried out by a research
characteristics in the relatives which are likely psychiatrist soon after the patient was admitted to
to cause a florid relapFe of symptoms, inde hospital. The main family interview was carried out
pendently of other factors such as length of at home by a research sociologist while the patient
was still in hospital. It usually took two separate
history, type of symptomatology or severity of
visits to each informant to complete and lasted
previous behaviour disturbance. about three hours in all. A husband or wife was
always seen; two parents (or married siblings or
1)ESIGN
pairs of siblings) were interviewed separately by
The case records were screened of all patients different workers.*
aged 18-64, born in the United Kingdom and living Both the current mental state and the family
with relatives at an address in Camberwell in S.E. interviews were repeated at the time of follow-up
London, who were beginning a new period of out nine months after discharge, and comparable ratings
patient or in-patient care at any one of five hospitals were made. A ‘¿ joint
interview' about two weeks after
serving the area. All those whose records indicated discharge was similar to that in the previous study
that they might be suffering from schizophrenia were
interviewed, using a semi-standardized technique to
rate and classify their clinical condition (Wing a al., * Twenty-seven parents were living together and 21

7967; Cooper, 1970; Wing, 1970). If a diagnosis of
possible or probable schizophrenia was made, the
patient was included in the study and further social
and clinical information was obtained. To obtain a
larger number of patients with recent illnesses, all
those with similar characteristics who were admitted
parents were living alone; 6 siblings were living either
alone (@) or with someone else (s). For the purposes of
this analysis siblings and parents have been placed
together. For the 30 ‘¿ pairs'of relatives we managed to
arrange separate interviews in 25 cases; in the remaining
5 cases only one informant was seen. In the interview at
readmission and follow-up only one relative was seen.
 BY G. W. BROWN,
(Brown ci al., 1962). The patient and other members
of the family were seen at home for no more than one
hour. The interview was concerned with the family's
and patient's recent contacts with medical or social
services and designed to get everyone talking together
on such topics. The scales concerning expressed
emotion were completed at the main family interviews
and at the joint interview. Patients and family were
also seen at any readmission during the nine months
after discharge.
TECHNIQUES OF MEASUREMENT
Family measures
The techniques of measuring family variables have
been described in detail elsewhere (Brown and
Rutter, mg66; Rutter and Brown, 1966). The family
interview deals not only with what happened at home
during the three months before admission (such as
who had carried out various household tasks or the
circumstances surrounding admission) but with the
feelings expressed during the interview
particular people in the home or towards recent
events. The interview was primarily designed to
obtain an account of the patient's behaviour and the
relative's feelings about him; but a somewhat
shortened form was also designed to be used when the
•¿
patient was seen. For the measurement of feeling most
reliance is placed on vocal aspects of speech—tone,
pitch and the like. Much of this material arises
spontaneously during the detailed questioning about
family activities and the development of the disorder.
Many different kinds of rating of family life were
made during the course of the study, and we cannot
do more than briefly describe those that are directly
relevant for this paper. Considerable attention was
given in the developmental work to inter-rater
reliability. No measure used in this report falls below
a product moment correlation (or comparable index
of agreement) between raters present at the same
interview of . 8o: most are in excess of this.
1. Ratings of emotional response
(a) J'Iumber of critical commentsabout someoneelse in the
home. Critical comments were judged either by tone of
voice or by content of what was said. For a remark to
be judged critical in content there had to be a clear
and unambiguous statement of resentment, disap
proval or dislike. Any remark could be rated critical
on tone alone, and in making the ratings most
emphasis was laid on the interviewer's judgement of
tone of voice. The verbal unit of assessment was a
statement terminated either by a change of topic or by
a question from the interviewer. Only one comment
could be counted per unit.
J. L. T. BIRLEY AND J. K. WING 243
(b) Hostility. Hostility was rated as present or absent.
It was defined as present if a remark was made
indicating the rejection of someone as a person; for
example, when someone was criticized for what he
was rather than for what he did. Hostility was also
regarded as present if critical comments tended to be
be generalized spontaneously; for example, when one
criticism triggered off a string of further criticisms
on unrelated topics (e.g. ‘¿ He's unhelpful, he's not
tidy and in money he's the world's worst').
(c) Dissatisfaction. According to our definition,
criticism and hostility are based on either negative
emotion or a clear statement of resentment, disap
proval, dislike or rejection. Another series of ratings
took account of any expression of dissatisfaction,
whether or not it warranted inclusion as criticism or
hostility. Dissatisfaction was rated on 4-point scales
describing eight areas of family life. An overall index
was also calculated. Many subjects who were highly
dissatisfied were rated low on hostility and criticism:
it is possible to be markedly dissatisfied yet express
towards little emotion or resentment.
(d) Warmth. This 6-point rating was based on the
amount of warmth demonstrated by the respondent
when talking about the particular person in the
home. In general, stereotyped endearments were
ignored, but positive comments, especially if made
spontaneously, were regarded as important. Sympathy
and concern, interest in the other as a person, and
expressed enjoyment in mutual activities were all
relevant. Particular attention was paid to warmth
expressed in tone of voice. Negative feelings were
deliberately ignored in making the rating, but
failure to express warmth in what seemed a relevant
situation (for example, when describing the patient's
ailments) was taken into account. A judgement was
based on the whole interview, but the expression of
warmth was most likely to occur in certain sections
dealing with leisure, marriage and communication,
and with the patient's behaviour.
(e) Emotional over-involvement. This measure was
designed to pick up unusually marked concern
about the patient. It was rated on the basis either of
feelings expressed in the interview itself or of behaviour
reported outside it. For example, a top rating on the
6-point scale was given when a mother showed
obvious and constant anxiety while describing such
minor matters as her son's diet and the setting of
his alarm clock so that he would wake in time for
work. She also showed markedly protective attitudes
about her son, who was not obviously handicapped;
for example, she said, ‘¿ could
I go out if I wanted to
go out. I don't ‘¿ cause I'm looking after Johnny'.
The rating was only made in the case of parents, as
such over-involvement was very rarely found in
244 INFLUENCE OF FAMILY LIFE ON THE COURSE OF SCHIZOPHRENIC DISORDERS

interviews with husbands and wives. Although threshold of definition (to two critical comments, and
conceptually it is the most complex of the scales, a score of 2 on the scale of emotional over-involve
agreement on a specially interviewed series of i8 ment). Only one of the three additional patients
parents was relapsed.
The relationship between the scales is much what
would be expected. Hostility and criticism are highly 2. Measures of behaviourbeforeand at the time of admission
related, and warmth is negatively related to criticism Many measures of the patient's behaviour before
and hostility. Emotional over-involvement is positively admission were employed in the study. Some of the
related to warmth, but only half of those rated as most important are:
markedly warm also showed marked over-involve (i) Work impairment. Unemployment or, for house
ment. Finally, emotional over-involvement shows a wives, marked handicap in carrying out domestic
curvilinear relationship to criticism and hostility: duties, for at least three months out of the preceding
those rated high or low on emotional over-involve two years. Time in hospital was not taken into
ment show most criticism or hostility. account.
(f) Overall index of relative's expressed emotion (EE). (ii) Disturbed behaviour. Definite aggressive or
Since the total number of patients is small, it is delinquent behaviour in the I 2 months before ad
necessary to limit the number of sub-groups produced mission (40 patients) or very markedly disturbed
by each variable. The individual scales were first behaviour at about the time of admission (this added
related to relapse (e.g. Table III) and several methods 7 patients).
of combining them into a high expressed emotion (iii) Social withdrawal. A score based on (a) contacts
(EE) group were explored. All of them produced with friends; (b) casual contacts outside the home;
much the same result. The following indices, in (c) contacts within the home.
hierarchical order, were finally used to allocate
3. Psychiatric measures
approximately half the families to a high expressed
emotion group: All patients were interviewed using the semi
N N standardized ‘¿ Present State Examination', with the
(Total N added to main object of describing fairly precisely the sympto
observed high EE matic condition of those included (Wing ci al., 7967).
in sub The CATEGO clinical classification procedure
brackets) group (Wing, Cooper and Sartorius, 7972) was then
Interview with relative applied, and only patients in the three groups
7+ critical comments .. 35 (i o i) ‘¿ typical
schizophrenia', ‘¿ delusionalpsychosis other
Marked over-involvement than mania or psychotic depression' and ‘¿ other
of parents .. .. 13 (@@) 5 schizophrenia', were included. Clinical data from the
Hostility .. .. .. i8 (ioi) 2 psychiatric history were compatible with this group
ing. Information on alcohol or amphetamine intake
Joint interview was also systematically recorded. Three groups of
2 + critical comments .. 9 (6@)
patients, with the following characteristics, were
Marked over-involvement finally set up:
of parents .. .. 9 (@@) 2 (i) Definite schizophrenia (N = 5'). The patient

Hostility .. .. .. 3 (6@) showed one or more symptoms as follows:

(a) thought intrusion, broadcast, commentary or
45 insertion;
The joint interview was considerably shorter than (b) delusions of control;
that with the relative alone, and rather less emotion (c) auditory hallucinations not based upon de
was expressed, the relative presumably being more pressed or elated mood.
restrained in the presence of the patient. Only three If patients with such a condition had been taking
additional cases were added to the group by using the amphetamine or alcohol, and the diagnosis was in
doubt for this reason, they were included in group (ii).
data from the joint interview. Even this small number
(ii) Possible schizophrenia (N = 42). There were
was only reached by considerably lowering the
two main groups:
* The published reliability study was based on 30
(a) Patients with definite schizophrenia as defined
married couples: a similar (unpublished) check with i8 above, but who were also taking alcohol or
parental families produced almost the same results amphetamine, so that there was a measure of
concerning the reliability of ratings. doubt about the aetiology (N = 8).
 BY G. W. BROWN, J. L. T. BIRLEY AND J. K. WING 245
(b) Patients who did not show the symptoms listed TABLE I
above but who described other delusions or Relapse, symptomatology and re-admission
hallucinations (for example, persecutory, refer
ence, religious, grandiose, somatic, sexual or Re re
admittedTotalRelapse admittedNot
fantastic delusions; visual or olfactory halluci
nations). Symptoms such as elation, or de (N = 35)
pressive delusions or hallucinations, were not Typel .. . i8

predominant (N = 34). TypeIl

12No .. .. . ii5 i23
(iii) Doubtful schizophrenia (N = 8). Patients with
no delusions or hallucinations who presented with relapse (N = 66)
stupor, excitement or muteness in the absence of No schizophrenic
depression or elation. These patients were all given symptoms .. . 3

hospital diagnoses of schizophrenia of non-paranoid Schizophrenic

17Total symptoms + .. . 046
type. One patient, for example, had recurrent 1749

attacks of retardation and stupor without affective .. .. .. 326g101

changes from which she appeared to recover com
pletely without treatment apart from admission to
hospital. Another patient had two attacks in which
she appeared disoriented, with markedly incoherent
speech, and seemed to be hallucinated. She recovered
on each occasion, but was unable to remember what
had occurred.
4. Relapse
The criteria used to assess relapse were the same
as those used in our studies of the social precipitants
of acute schizophrenia, which were based upon the
same series of patients as the present one (Brown and
Birley, 1968; Birley and Brown, 1970).
We distinguished between two types of relapse:
Type I involved a change from a ‘¿ normal' or ‘¿ non remained steady or showed only mild fluctuations.
schizophrenic' state to a state of ‘¿ schizophrenia'as
defined in the previous section. Type II involved a
marked exacerbation of persistent schizophrenic
symptoms. At our interviews at follow-up or read
mission, we made a judgement, based on all available
information, on two points: firstly, whether the
patient had experienced schizophrenic symptoms at
some time during the follow-up period; and, secondly,
whether those patients who had been continuously ill year after discharge.
had experienced a marked exacerbation of these
symptoms.
Table I shows the relationship between relapse,
symptoms and readmission of the 35 patients who
had relapsed during the follow-up period. Twenty
nine of the 35 were readmitted. The remaining 6 all
had schizophrenic symptoms at their follow-up
interview; five of these patients had been well for
some time after their key discharge, while the sixth
had been continuously ill and had experienced a
marked exacerbation of symptoms shortly before
interview.
Three of the readmitted patients had not relapsed.
One woman had been admitted to avoid possible
friction with her son who was returning home from
prison. The two others were both women who had
taken small overdoses of phenothiazines, apparently
on impulse. One had been upset by her poor memory
when asked questions about her baby in a welfare
clinic. (She had originally been admitted with
puerperal schizophrenia and had received an ex
tensive course of ECT). The other said she was ‘¿ fed
up with taking tablets' and had decided to take them
all at once. All three patients remained well during
the rest of the follow-up period.
Seventeen of the ‘¿ notrelapsed' patients had ex
perienced schizophrenic symptoms continuously
during the follow-up period, but these symptoms had
5. Time relationships between variables
The indices of work impairment and disturbed
behaviour relate to events that are past by the time
the patient is admitted to hospital. The scales of
expressed emotion are rated while the patient is still
in hospital or shortly after discharge. The index of
symptomatic relapse relates to events during the
These indices measure conditions which are them
selves changing over time. For example, at the time of
follow-up, g months after discharge, 45 patients were
rated as showing markedly improved behaviour
compared with the time of key admission, while the
other 45 showed moderate improvement or none.
The number of criticisms made by relatives also
decreased considerably. Thus, at the time of key
admission 29 out of @orelatives (32 per cent) made
no criticisms, while 27 made seven or more (30 per
cent). At the time of follow-up 43 made no criticisms
(47 per cent) and only 13 made seven or more (14 per
cent). The greatest reduction in number of relatives'
critical comments occurred with respect to those
patients who had markedly improved; on average
246 INFLUENCE OF FAMILY LIFE ON THE COURSE OF SCHIZOPHRENIC DISORDERS

they received only one quarter of the original number parents are used, all based upon the interview
of critical comments compared to the rest who still with the relative alone. (Where more than one
received about one half. The other indices of expressed person in the family was seen, the higher rating
emotion, such as warmth and over-involvement, was used). In each case there is a significant
showed less of a tendency towards decreased emotion
association with relapse.
at follow-up.
We are making the assumption, therefore, that a T@LE III
high degree of expressed emotion on one occasion is
a measure of the relative's propensity to react in that Relationshipof threemeasuresof relative'semotionto
relapsein the9 monthsafter discharge
way to that particular patient, even though other
factors may be needed to precipitate this. The same (i) Number critical comments
relative would not necessarily respond to other people
in the same way. For example, there is very little No
correlation between the amount of emotion expressed relapseo Relapse relapse%
by a parent towards the patient and the amount of
emotion expressed by the same parent towards his or 6 26

her spouse. The measure reflects a quality of relation i—6 .. .. .. 9 25 26

ship with a particular person (the patient), not a 7+.. .... .. 57(ii)
.. 20 15

=
general tendency to react to everyone in a similar way. (rows.. i and 2 combined)‘9
Thus the level of expressed emotion at the time of
the patient's key admission will be taken to represent HostilityNo
an enduring potential characteristic of the relative's
behaviour towards the patient.
relapseNo Relapse relapse%
RESULTS

Relationship between index of expressed emotion 28 59

Yes.. .... .... 50

and relapse ..
y =
7
•¿ 3632
7

The proportions of patients with relatives in

the high and low EE groups who relapsed are (iii) Emotional over-involvement of parents
shown in Table II. There is a significant
association between high EE and relapse No %
Relapse relapse relapse
(@,= .75 I df, p < .ooi)*
TASi.z II Low (o to 3)
Relationshipof relatives'emotionto relapsein the 9 months Marked (4 an..d @).. 62y
..15 827 536
= •¿ 48
after discharge

relapseHigh
EE of relativesNo relapseRelapse% Warmth expressed towards the patient was
not used in the overall index. The scale shows
a curvilinear association with relapse; patients
Low.. .... .... ..19
4726 958 i6 in the middle range showing the lowest relapse
rates. This was because relatives rated as
This result is confirmed when various other showing little warmth tended to be highly
indices of high expressed emotion are examined. critical, while those rated as showing consider
For example, Table III shows the proportion of able warmth tended also to be emotionally
relapses when measures of critical comments, over-involved. Marked warmth free from these
hostility or emotional over-involvement of the unfavourable factors was associated with a low
rate of relapse. Only one patient relapsed out
* y is used as the measure of association. It is appropriate of i i from families characterized by marked
for ordinal material and provides an estimate of the pro warmth in the absence of high EE (df. @,
portional reduction in error (Goodman and Kruskal,
1954). A coefficient of i @oregisters a one-way association,
p < @OI)
i.e. gamma is I@o if there is just one zero in a 2 X 2 The other major affective variable measured
contigency table (Mueller and Schuessler, 1961). was dissatisfaction. There was an overall
S
 BY G. W. BROWN, J. L. T. BIRLEY AND J. K. WING 247
association between dissatisfaction on the part tension in the home. The relatives' expressed
of relatives and the patient's relapse, but only emotion may be caused by the patient's past
within the high EE group. This suggests that behaviour, the emotional expression may have
dissatisfaction as such did not cause relapse. caused the behaviour, or each may have
Ratings of expressed emotion in the patients influenced the other; it is impossible to say what
showed far less than in relatives (only 10 per is primary in a study of this kind.
cent of patients were markedly critical about The three possibilities may be expressed
relatives at home compared with 34 per cent of graphically as follows:
relatives about patients) and there was no (Bi) I/D-÷EE--'.R
relationship to relapse. (B2) EE-÷R
@I/D
The @ffect
of previous behavioural and work (B3) I/D@—÷EE-÷R
impairment In each of the cases Bi, B2 and B3, it is the
Work impairment during the two years before relationship between EE and R which can be
key admission was related to relapse (y = .40), differentiated from that in case (A).
as was disturbed behaviour (y = .32) An index It is possible to discriminate between these
combining these two factors did not much two hypotheses, since if relatives' EE independ
increase the association (y = .47) Both were ently contributes to relapse various conditions
closely related to EE (work impairment, y = @73; should hold. (The relevant associations are
behavioural disturbance, y = .82). Two-thirds of given in Fig. i and in Appendix I which shows
patients who had been disturbed or had had the associations between the main variables
work difficulties lived with a relative with a high used in this paper.)
level of EE, but only 14 per cent without such (i) In the first place, when previous work
behaviour did so. impairment and behavioural disturbance are
There are two alternative ways of explaining controlled by standardization* the association
this set of correlations. The first and most between EE and relapse should not be much
straightforward hypothesis is that past impair reduced. This is the case (gamma becomes
ments and disturbances are predictive of future •¿ instead
63 of .75), and these two background
behaviour such as relapse because some under factors cannot be producing a spurious effect
lying ‘¿ process'links them together. At the same (Blalock, 1960).
time, the more disturbed the patient's be (ii) In the second place, since in the three
haviour, the more likely are the relatives to alternatives EE is either an intervening variable
respond with criticism, hostility and emotional (Bi and B3) or a common cause of the other
over-concern. Thus the correlation between factors (B2), the association between impair
expressed emotion and relapse is spurious in so ment/disturbance and relapse should be greatly
far as it is mediated by the patient's own decreased when EE is controlled (Blalock, 1964).
behaviour, as shown in diagram (A): In fact it almost disappears (@47 to o8).
(iii) Finally, it follows from these same
EE (expressed emotion)
considerations that the association between
I/D impairment/disturbance and relapse (@47)
(A) (impairment/disturbance) should be weaker than that between impair
ment/disturbance and EE (.84) or between EE
R (relapse) and relapse (.m@). These conditions also hold.
Table IV shows why this is so. Three
The second hypothesis is that the relatives' quarters of the patients in the series fall into one
expressed emotion contributes to relapse inde of two categories; either they have been dis
pendently of the patient's past behaviour. In
other words the patient would be unlikely to * Standardization
effect of apparently
has been
important
used to
variables
control for
(Rosenberg,
the

relapse, however disturbed or impaired his 1962; Atchley, 1970). The raw data used to standardize
past behaviour, if there were no ‘¿ emotional' the associations may be obtained from the senior author.
248 INFLUENCE OF FAMILY LIFE ON THE COURSE OF SCHIZOPHRENIC DISORDERS

Fio. i.—Relationship of the main variables to each other and to relapse

TABLE IV past and are living with relatives with a low

Relationshipbetweenbehaviouraldisturbance,work degree of expressed emotion. Only one-quarter
impairment, expressedemotional and relapse of the patients are non-congruent for the two
Past factors; but here, it is plain, the degree of
behavioural Expressed expressed emotion is related to relapse, and the
relapseRelapseTotal++151934+015ii60+47II0032840Total6635101Past
disturbance• emotionNo degree of past disturbance is irrelevant (x2 =
I0@28, df i, p < .ooi). Precisely the same is
true of past work impairment.

Clinical picture
Patients who showed ‘¿ typical'schizophrenic
symptoms (thought intrusion, delusions of
control, etc.) at the time of present state exami
nation were more likely to relapse than those
workExpressedNoImpairmentemotionrelapseRelapseTotal++121426+073100+712190040646Total6635101
in the second group (y = .45). The third (non
paranoid) group is too small for separate analysis
and is included with the second.
This association between type of clinical
condition and relapse is not mediated by a
higher degree of past work impairment or
behavioural disturbance (y for the association
between clinical condition and impairment!
turbed in behaviour in the past and are living disturbance is only . i8; with relapse controlled
with relatives with a high degree of expressed it is reduced to zo). Moreover, the association
emotion, or they were not disturbed in the between EE and relapse remains strong (y= . 79)
 BY G. W. BROWN, J. L. T. BIRLEY AND J. K. WING 249
when type of clinical condition is controlled, alternative indicators of it. Since they add
and the association between type of clinical nothing to the ability of the index to predict
condition and relapse is increased slightly relapse they can be ignored. Whether a patient
(y= .49) when EE is controlled. has been previourly admitted or not relates
Thus, type of clinical condition is not asso somewhat to the impairment/disturbance index
ciated with any particular degree of emotional (y = .32). When it is controlled, the association
expression (y=. io) and seems to be independ between EE and relapse is not reduced and so
ently related to relapse (y .49). it can be ignored. Length of time since first
onset and duration of present episode appear to
Marital and parental homes
have little significance.
When this analysis is repeated separately for
The living group and sex differences are more
patients living with parents and patients living
difficult to explain. The relapse rate for men is
with marital partners, the main results are
double that for women, and the rates for the
repeated in every particular. Fewer married
unmarried are somewhat greater than for the
patients are rated as having been disturbed or
married:
impaired in the past, but this does not affect
Married:
the hypothesis under test.
men 44%; women 17%; total 26%
Other factors related to outcome Unmarried:
@a)Background factors. Table V lists other men 53%; women 26%; total 42%
background variables that show some relation The difference between the unmarried and the
ship to relapse. The three indices of previous married is a consequence of the fact that if
occupation are all highly correlated with the either of both parents live with no one other
index of work impairment and are best seen as than the patient a greater amount of emotion

T@@rn@x
V
The relationshipof additional ‘¿ background'factors
to relapse
0/
/0Nrelapse
gamma(a)Best

occupational level (men; N = 47)Skilled

....1443.17(b)Decline or better
or unskilled ..
occupation level (men; N = 47)Semi None .. .. .. ..
23 35
65.56*(c)Most Decline .. .... ..33
2352
recent occupation level (men; N = 47)Skilled or better ..
56.68*(d)Time Semi or unskilled .... ..10 3720
since first onset< i year .. .. 7
i—5years .. .. .. 52 38
35.17(e) 5years+ .. .... ..I 2629
previously admitted .. .. ..
Yes .. .. .. .. 41 54
Living group at dischargeNo
.34(g)Sex
(f)Whether Marital .. .. .. 46 26
Parental/sibling .... ..59 5522 42.6i*
of patientFemale .. ..
50.59*(h)Age Male .. .. .... ..53 4821
of admission45 plus .. ..
39.45*(i)Peak —¿ 45 .. .. .... ..25 7620
of heterosexual performanceProlonged
relationship..3029.33(j)Decline
prolonged relationship
in peak heterosexual performanceNo Decline .. .. .. 29 24
No decline .. .. .. 52
@ 8
Never sexual contact.. ..62 1447 36f3•28
Note: Variation in totals due to cases not known.
* p <•o@.

*
250 INFLUENCE OF FAMILY LIFE ON THE COURSE OF SCHIZOPHRENIC DISORDERS

is expressed. The reason will be discussed later; outcome, but only modestly, and just fails to
it should be noted, however, that it is not a reach statistical significance (y = —¿â€˜¿ 41).
How
measurement artefact. The difference remains ever, the results are not straightforward. Table
when allowance is made for the fact that the VI shows that drugs appear to have no effect
highest rating of either parent is used to arrive on patients living with relatives rated low on
at the EE index. EE, suggesting that medication might serve
The large difference between men and women mainly to protect patients who live with relatives
is puzzling. The rate of relapse in men is showing a high EE. The numbers are small,
double that in women and this difference however, and the difference in outcome between
remains when marital status and level of EE patients in the high EE group who were taking
are controlled. Extensive analysis failed to and those who were not taking drugs regularly is
suggest any reason for it. It does not, however, not statistically significant. So far as the
affect the main association between EE and patients in the low EE homes are concerned,
relapse, which is high in both sexes (y . 8@ for however, there is no trend at all.
women and ‘¿ 70 for men). Previous research suggested another variable
Age at admission and peak heterosexual of importance: that the patient who returned to
performance are related very highly to sex and a high EE home could to some degree improve
marital status, due to the fact that women in his chance of avoiding relapse by seeing less of
the series are older than men and more likely his relatives. More than 35 hours per week of
to be married. face to face contact with adults in the home
Decline from peak sexual performance @s,if seemed to be the critical amount of time; less
anything, related to a lower rate of relapse. As contact than this seemed to provide some
will be seen later, there is a general suggestion protection when the patient returned to a high
from a number of results that ‘¿ social
withdrawal' EE home (Brown et al., 1962). In the present
can protect a patient from relapse. study we made a similar estimate, at the time of
Social class measures based on the occupation the follow-up interview, of the amount of face to
of the male patients, the father's occupation face contact between relatives and patients
for women living with parents, or the husband's living at home before relapse.
occupation for married female patients, showed Table VII shows that the previous result was
no consistent relationship with relapse; nor did repeated: the amount of contact makes no
length of time since first onset or length of difference for those living with low EE families,
current episode. but a very large one for those living in high EE
(b) Factors after discharge. Two-thirds of the homes. These results hold for both drug and
patients took one of the major tranquillizing non-drug groups, and both factors are probably
drugs for a large part of the follow-up period or of importance in determining the conditions
until their relapse. A larger proportion of under which the home environment may have
patients failed to take drugs in high EE homes a deleterious effect.
(idf, p < .oi). Drug taking does relate to It is possible that a general coping style is
T@.rn..zVI
Relationshipof relatives' EE, drug taking after dischargeand relapse

drugsDrugsNo
Relatives'
EENo
relapseHigh relapseRelapse% relapseNo relapseRelapse%

Low..
i@Total..171445491726
..6 II12 266 1513 36II 646
 BY 0. W. BROWN, J. L. T. BIRLEY AND J. K. WING 251

T@LE VII some way was related to the likelihood of relapse

Relationshipof relatives' EE, time spentin face toface after discharge. Patients who objected were less
contactper weekafter dischargeand relapse likely to relapse (Table VIII). There are no
in face to face obvious reasons for this result, which is made
contact
relativesLess
with more curious by the fact that resistance to ad
mission is itself related to a series of factors which
hoursNothan 35 otherwise appear to make relapse more likely.
EE Rejection of admission, for example, is associated
%
relapse Relapse relapse with number of critical comments (‘44),over
High .. .. .. 15 6 29 involvement (‘@@), work impairment (-@4) and
LowRelatives' ..
25y=.()9Relatives' ..Time .. 15 5 disturbed behaviour (.63). Those who had
shown most impairment and disturbed be
haviour and who had provoked relatives into
the most marked emotional expression were
hoursNothan 35 also most likely to have resisted admission, but
EEMore
% they had a lower chance of relapse. Extensive
relapseHigh relapse Relapse analysis revealed only two clues to this paradox.
In the first place, patients who had resisted
.. 4 15 79 admission did better when they took drugs after
Low..
discharge (Table IX). Out of 25 patients who
.. .... .. 29 4 12

V = ‘¿ 92
had resisted admission, living in high EE homes,
(Hours of contact not known in 8 cases) most of whom would have been expected to
relapse, only 8 did so, and 6 of these were not
taking medication regularly, whereas 13 of the
involved; a tendency for some patients when in 17 patients who remained well took regular
difficulty to withdraw from close social contacts. medication. In the second place, they were
There is some support for this view. Several much more likely to have been markedly
indices of level of contact with others in the socially withdrawn either at home or in their
home in the three months before admission were
associated with subsequent outcome, parti T@usi.zVIII
cularly in patients living with parents or Patients' attitude to admission and relapse
siblings. Patients who had been most involved in
family life relapsed more often after discharge. No %
For example, the following factors were asso relapse Relapse relapse
ciated with a higher relapse rate (all but the Accepting....322544Rejecting..
first two measures are based on the relative's 10
reports): patient's report of a dominating parent y=.. ‘¿ 41,p34 <@05,Idf23
(y = . ag), patient's report of unreduced com
munication with parents (.50), one day or
Ti@rn@n
IX
more per week leisure activity with parents Patients' attitude to keyadmissionand relapseby
(‘@@), a moderate or high level of communica whetherdrugsweretakenafter discharge
tion with parents (.42), unreduced joint leisure
activities (.31), holiday with parents in year
Attitude
(.29), joint contact with patient outside home
to EE No No
(‘28), 6 or more evenings spent at home per admissionDrugs relapse RelapseNorelapsedrugs
RelapseRejecting
week (.27). Only two interactional measures
failed to show such a trend. The result did not, High
not, however, hold for married patients. Rejecting Low13
IAccepting 132 04 36
High47I5Accepting
(c) Patient's attitude towards admission. Whether Low2065I
a patient accepted admission or resisted it in
 252 INFLUENCE OF FAMILY LIFE ON THE COURSE OF SCHIZOPHRENIC DISORDERS
leisure activities in the two years before ad as ‘¿ personality' factors played some part in
mission—often in both (y = .49 p < .05). determining their response. It is important to
Although this particular measure of social investigate this possibility, since if expressed
withdrawal was unrelated to relapse (y = —¿ . o6) emotion is strongly related to some other
various results have suggested that ‘¿ social variable that acts independently of the patient
withdrawal' can lessen chances of relapse of it would be even less likely that the relationship
schizophrenic patients. between EE and relapse could be explained by
Rejecting admission, taking drugs regularly, some enduring clinical characteristic. One
and low face to face contact with relatives after possibility is that the reaction of relatives is
discharge were all associated with a favourable related to the amount they see of people other
outcome. Of patients who were characterized than the patient. It seemed possible that there
by all three factors none relapsed; of those was a relationship between the amount of
characterized by two, 23 per cent relapsed, and contact that relatives had with others, their
of those characterized by only one, 46 per cent dependence on the patient, and their tolerance
relapsed. of the patient's behaviour; in short, that the more
(d) Summary of other factors relating to relapse. isolated relatives would show the least tolerance
Thus, typical schizophrenic symptoms, male because they were the most dependent on the
sex, acceptance of admission, lack of regular patient for practical and emotional support. It
medication and high contact in the home are was therefore predicted that a high rate of contact
associated with a higher chance of relapse. with relatives outside the home, or the presence of
Table X provides a check that the relationship others in the home, would reduce dependence
between EE and relapse is not reduced when on the patient, and lead to less expressed
these factors are controlled. The presence of emotion.
any such factor scored one point, giving an For patients living with parents these ideas
overall score from o to @.Relapse is more likely were confirmed: a parent living alone, except
the higher the score, but the association between for the patient, or living only with a spouse,
EE and relapse holds within each score, con expressed more emotion than those who had
firming that it is independent of these factors. others living in the house, apart from the
patient. More emotion was also expressed by
Determinants of the relatives' emotional responses those with low rates of contact with friends
Not all relatives who had lived with patients and relatives. The 20 per cent who were most
characterized by long-term work impairment or isolated were almost all rated as expressing a
behaviour disturbance had a high EE score; high degree of emotion (Table XI).
@ as many as 20 of the who did so showed a However, for married patients this result was
low score (in contrast to 36 of the 42 remaining reversed. The isolated relatives expressed the
relatives). It is possible that situational as well lowest emotion, and it was the 40 per cent who
T@LE X
Associationbetweenemotionalresponseand relapsecontrollingfor thepresenceof fivefactors: typeof clinical condition,
sex, acceptanceor rejection of admission, time in face-to-face contact and medication

score five factors

5*and4Overall
2,ando*No 3on

RelapseHigh relapse RelapseNo relapse RelapseNo relapse RelapseNo relapse

EE .. ..
5 16 4 3
Low EE .. .. ..
3 4 14 2 i8 I II I
y .. .. .. .. @58
.93 @92 .53
Overall relapse.. rate as %I 694 503 1910 i6

* One patient scored 5, and four scored o.

 BY 0. W. BROWN, J. L. T. BIRLEY AND J. K. WING 253
T@nLEXI where they lived, the frequency of contacts with
Relationshipof socialcontactsof relativesinsideand kin and friends one year later when the patient
outsidethehometo expressed emotion was often a good deal improved, and we also
asked whether the patient's illness had changed
Expr essed em
‘¿ Isolation'
indexotion%Homeside the amount of contact with relatives and
friends. These data did not suggest that the
Out severity of the patient's disorder could explain
LowHighhigh(a) the results.
Parentsand,siblings DIscussIoN

Yes* Marked The main results of the previous family

study of discharged schizophrenic patients

27(b) NoYes**No Low II10

@}
490
37 have been replicated
respects extended.
and in certain important

MarriedYesYes In the first place, the ‘¿ emotionalinvolvement'

8IIIYes Marked of relatives with patients has been more pre
. 6 cisely specified and the term ‘¿ expressedemotion'
>Medium >
INO xesj 9 3J substituted. The most measurable component of
NoNo@lNo Low 8ol Ii12 58 expressed emotion is the number of critical
* Yes = Relative living alone except for spouse.
comments made by the key relative about the
** Yes = With an average rate of fewer than 3 patient. The whole analysis could have been
contacts per week in the 3 months before undertaken using this measure alone. However,
admission. the other components are important in inter
preting the nature of ‘¿ expressedemotion'. These
were least isolated who most often expressed are ‘¿ emotional
over-involvement' and ‘¿ hostility'.
high emotion (Table XI). ‘¿ Warmth',which was formerly considered to
Thus there is some suggestion that the be a component, is now seen to be a complex
relatives' social life plays some role in deter variable. A high rating of warmth is often
mining their reaction to the patient, but the accompanied by a high rating on emotional
theory as stated applies only to parents. In their over-involvement, while a low rating on warmth
case the presence of others in the home makes usually implies a high level of critical com
them less dependent on the patient, but when ments. Patients whose relatives showed marked
the patient is a husband or wife the presence of warmth without also expressing criticism or
others in the home (who will most often be over-involvement had a significantly better
children) may make the patient's spouse more outcome. Warmth was consequently not in
dependent on the patient. Furthermore, a cluded in the index of expressed emotion.
spouse may depend more upon the patient ‘¿ Expressed emotion', therefore, has a mainly
fulfilling a parental, breadwinning or house negative connotation.
keeping role and upon a ‘¿ joint'front before Using this index, a high degree of emotion
guests and when others are contacted outside expressed by relatives at the time of key ad
the home. It is possible that such ‘¿ structural' mission war found to be strongly associated with
features could be influenced by the patient's symptomatic relapse during the nine months
disorder and therefore produce a spurious following discharge. This replicates our principal
result. It is not difficult to imagine brothers previous finding. The question arises how far
and sisters leaving home more often when the this relationship can be explained by the fact
patient had been badly disturbed, and this kind that patients with the most severe behavioural
of selective process could have brought about disturbance and work impairment before the
the results. Some estimate could be made of time of key admission had the greatest liability
such contamination. We knew the number of to relapse. In our previous study, both factors
potential kin who might be contacted and were independently related to outcome. The
254 INFLUENCE OF FAMILY LIFE ON THE COURSE OF SCHIZOPHRENIC DISORDERS
present results are unequivocal, however, in the critical period; above this, the chances of
suggesting that expressed emotion is inde relapse were greatly increased. The same held
pendently associated with relapse, while pre true in the present analysis. Several other
vious work impairment and behavioural disturb findings contributed towards a strong impression
ance are only associated with relapse because that social withdrawal could be a protective
of their association with level of expressed factor, particularly for unmarried patients.
emotion. Moreover, the ability of the index of Other factors were independently linked to
expressed emotion to predict symptomatic chances of relapse, such as age (under 45), sex
relapse is not explained away by the action of (male), admission status (not first admission),
any other factor that we have investigated (such recent occupational level (unskilled manual),
as age, sex, previous occupational record, length decline in occupational level and failure ever to
of clinical history, type of illness, etc.). It will achieve a satisfactory sexual adjustment.
therefore be assumed, during the remaining One other factor deserves special mention;
discussion, that the level of relatives' expressed schizophrenic subtype. The patients were divi
emotion must be taken into account as one of ded into three groups, as described on pages
the factors that cause relapse. 244—45, of which the first was composed of
It is worth mentioning that dissatisfaction on patients with clear-cut and typical schizo
the part of relatives was only associated with phrenic illnesses (Class S in the CATEGO
relapse if criticism was also present. Dissatisfac system; Wing, Cooper and Sartorius, 1972).
tion need not necessarily be critical. The other two groups of patients could have
The action of this factor can be mitigated to been diagnosed in various ways (paranoid state,
some extent by two other important environ alcoholic hallucinosis, recurrent atypical psy
mental variables which might help to give the chosis, oneirophrenia, etc.) although they were
clinician some control over events. The first of in fact, all labelled schizophrenia. The relation
these is regular phenothiazine medication, which ships between expressed emotion, relapse, and
shows a close to significant relationship with a other factors were identical within the first
favourable outcome. In view of the results compared with the second and third groups. The
obtained in a recent controlled trial of preventive first (most typical) group had a worse prognosis
phenothiazine medication (Leff and Wing, than the other two, but this was not due to a
1971) this finding must be taken seriously. It higher degree of expressed emotion in the
is the more significant since, as Leff and Wing patients' relatives. It seems that type of clinical
found, preventive medication is only effective condition is an independent variable, and this
for certain groups of schizophrenic patients. Our raises the question whether the factors we have
present results suggest that patients living with been considering are in any way specific for
relatives who expressed high emotion at the schizophrenia at all. This is not, however, a
time of key admission (and who are therefore question we can answer from our present
most vulnerable to relapse) are also most material. So far, we have been discussing the
susceptible to protective drug effects. There is relationship between expressed emotion and
one other clue to the sort of patients most outcome, in the light of other factors which
responsive to phenothiazines; those who resisted exercise an independent, mitigating or delete
admission were particularly like to be living rious effect upon it. The results may be ex
with relatives who expressed high emotion, and pressed in the same terms as those of our previous
they were particularly helped by drugs. article, i.e. that three types of factor independ
The second mitigating factor is the extent to ently predict outcome; previous history (age,
which the patient can avoid a too close contact sex, decline in occupational level etc.), clinical
with a highly emotional relative. The most condition (schizophrenic subgroup, severity of
clear-cut measure of this is the number of hours symptoms at discharge—not measured in the
per week that the patient spends in actual face present study), and factors following discharge
to-face contact with the relatives. In our pre (relatives' expressed emotion, medication,
vious study we found that 35 hours per week was amount of face-to-face contact).
 BY 0. W. BROWN, J. L. T. BIRLEY AND J. K. WING 255
Now we shall turn out attention to the deter It is difficult to reconcile this example with
minants of the relatives' degree of expressed model B2, in which the degree of expressed
emotion. We have shown that there is a con emotion is held to be primary. A circular effect
siderable association with the patient's beha (which may be vicious or benign) seems much
viour and employment record. We cannot more probable. Another point is also worth
specify the direction of cause and effect, but the emphasizing, since our previous results have
fact that a decrease in expressed emotion at sometimes been misinterpreted. We do not
follow-up accompanies an improvement in the wish to suggest that the behaviour of the
patient's behaviour strongly suggests that there relatives was in general abnormal or excessive.
is a two-way relationship, each depending on We have found a high degree of expressed
the other. Striking examples of such an inter emotion in a substantial proportion of relative
action of factors can be found in our material. patient pairs, but not in relative-relative pairs.
For example, one married patient had been We have the impression that the relatives of
disturbed in behaviour for about six years, other kinds of handicapped individuals might
following a surgical operation. She experienced also tend to develop such relationships, though
auditory hallucinations and had periodic out whether the tendency would be as strong as
bursts of shouting. She was suspicious of every with schizophrenia deserves investigation. It
thing and accused and threatened her husband. would seem to be a reasonable assumption that
The patient sat behind closed curtains most of it would depend upon the degree of behavioural
the time, smoking cigarettes, and she cooked a disturbance involved.
meal only once a week. There was ‘¿terrible
a We are unable to comment on claims that
atmosphere' in the home. Her husband often factors in the relatives' personality and handling
threatened to leave her unless she accepted of the patient as a child cause the first onset of
treatment, but did not actually do so. (‘Ifelt the illness, except to say that the fact that
sorry for her' . . . ‘¿ althoughlove has gone I expressed emotion acts as strongly in marital
think there is some goodness there'). His degree partners as in parents argues for a reactive
of expressed emotion was high. She was finally rather than a causal model.
admitted to hospital. At the time of follow-up The only substantial body of work which
she was reported to be quite well apart from a attempts to show a measurable abnormality in
tendency to grumble and ‘¿ to
go on and on parents (which would, however, be compatible
talking'. She was taking no drugs, was coping with a genetic or an environmental explanation,
well with housework and cooked each day. or a mixture of both) is that of Wynne and his
There was much less tension (low expressed colleague (Wynne, i g68). However, their results
emotion) and the couple now spent most of their have not been replicated in recent work by
spare time together, chatting and watching Hirsch and Leff (i@7i). This discrepancy could
television. We would argue that in such a case perhaps be explained by the very different
the effect of hospital treatment was to mitigate diagnostic practices which have been shown to
the severity of the patient's symptoms; this exist in the U.S. and the U.K. (Kendell et al.,
lowered the level of tension in the home and 1972) in which case Wynne and Singer's work is
the expressed emotion of the relative, which in also of less relevance to our own. The patients
turn reduced the chances of relapse, of work in the present study, as in Hirsch and Leff's
impairment, behavioural disturbance and symp investigation, were diagnosed according to a
tomatic exacerbation in the patient. This is standardized clinical examination and classifi
model B3 (Low I/D'@—-*Low EE-÷No Relapse). cation procedure (Wing et al., 1967; Wing,
There was no preventive medication in this 1970). Such a group of patients would also
case, but we would suggest that it was probably be recognized as schizophrenic in the U.S. but
unnecessary. If the relative had continued to would probably form only a minority of any
express a high degree of emotion, however, pre group diagnosed there, the majority suffering
ventive medication and a reduction in amount of from mania, depression or personality disorders
face-to-face contact would have been useful. according to British practice.
256 INFLUENCE OF FAMILY LIFE ON THE COURSE OF SCHIZOPHRENIC DISORDERS
Our own theory (Brown and Birley, 1968; social environment would be structured, with
Wing and Brown, 1970) is that many schizo clear-cut roles, only as much complexity as any
phrenic patients remain very highly sensitive to given individual can cope with, and with
their social environment, even when there are neutral but active supervision to keep up
no apparent symptoms. The optimum ‘¿ arousal' standards of appearance, work and behaviour.
level (Venables, 1964) may very easily be upset Thus we postulate interactions between bio
if the patient finds himself in overstimulating or logical and social factors which will have pre
understimulating social conditions. In the dictably different results in different social
presence of a socially intrusive relative, for environments.
example, he is unable to withdraw, and any Finally, we should like to make some tentative
residual or latent thought disorder will become suggestions as to the ways in which our findings
manifest as expressed delusions or odd beha might be of practical interest to clinicians who
viour. This condition will be accompanied by are trying to help schizophrenic patients and
a high level of ‘¿ arousal'. their families.
Epstein and Coleman (1970) suggest that this (i) Certain factors which can potentially lead
arousal ‘¿broader
is than fear and anxiety in the to relapse are to some extent under the patient's
usual sense of these terms, as it involves a control; whether he takes medication, whether
reaction to the combined stimulation from all he remains in face-to-face contact with a socially
sources, including positive and negative emo intrusive relative and, in other ways, whether he
tions and impulses, as well as external stimuli'. withdraws from socially complicated situations
We have demonstrated that schizophrenic (disturbing sexual relationships, for example).
patients are susceptible to both positive and (2) Reducing the degree of disturbance of a
negative emotional stimuli (Brown and Birley, patient's behaviour or of his work impairment
1968). Life changes and crises are most often may act to reduce the level of emotion expressed
unpleasant and lead to negative emotion, but by a key relative.
by no means always. We may assume that (3) Long-term medication may be parti
patients living with relatives who express a cularly effective with patients who have resisted
great deal of emotion are likely to have height the idea of entering hospital for treatment and
ened arousal levels over long periods of time. who return to live with highly emotional
The occurrence of a critical event during such a relatives.
period is particularly likely to precipitate a (4) Applying techniques of reducing the
relapse, although either factor acting alone may number of hours spent in face-to-face contact
occasionally do so. Social environments other with highly emotional relatives may make the
than the home may also provide conditions for difference between stability and relapse.
long-term over-arousal but they have not been (5) Therapy with relatives should take
systematically studied. However, too enthu account of their liability to develop highly
siastic attempts at reactivating unprepared emotional responses to the patient. In the light
long-stay patients have been shown to lead to of present knowledge, it should not be too
sudden relapse of symptoms that had not been readily assumed that the parents' handling
present for years (Wing, Bennett and Denham, of the patient when a child has caused
1964). On the other hand, if the patient is schizophrenia to develop; such an assumption
allowed to withdraw, for example in the may be wrong, in which case harm may be
understimulating social environment of an old done both to relative and patient.
fashioned ment@il hospital (or even in an attic (6) The best coping patterns may differ for
at home), what might in other circumstances single patients living with parents and for
have been a protective withdrawal may go too married patients living with their partners.
far, and the picture of the typical ‘¿ back ward' Extra social contacts on the part of the parents
patient develops (Wing and Freudenberg, may be helpful if their friends and relations
1961; Wing and Brown, 1970). Such a patient accept the patient as handicapped. The friends
too, shows a high level of ‘¿ arousal'.The optimum and relations of a patient's husband or wife, on
 BY G. W. BROWN, J. L. T. BIRLEY AND J. K. WING 257
the other hand, may have different expectations BROWN, G. W., MONCK, E. M., CARFrA.nts, G. M., and
WING, J. K. (1962). ‘¿ Theinfluence of family life on
of the patient, which can cause considerable
the course of schizophrenic illness.' Brit. 3. prey. soc.
@ embarrassment and concern to the non-schizo Med., i6,
phrenic spouse. It may be possible to anticipate —¿ Boz'@z, M., DALISON, D., and WING, J. K. (i966).

and deal with such problems. Schizophreniaand Social Care. Maudsicy Monograph
(7) If likely ‘¿ precipitating' events can be No. 17. London: Oxford University Press.
anticipated it may be possible to prevent an —¿ and RUTFER, M. L. (1966). ‘¿ The measurement of

family activities and relationships.' Human Relations,

unwanted reaction, by adjusting the dose of 19, 24!.
medication, or by preparing patient and —¿ and BIRLEY, J. L. T. (ig68). ‘¿ Crisis and life changes

relatives, or even by avoiding the situation and the onset of schizophrenia. 3. of Health and Social
altogether in some cases. Behaviour, 9, 203—14.
These are fallible rules. The clinician will, in COOPER, J. E. (1970). ‘¿ Theuse of a procedure for stan
dardizing psychiatric diagnosis,' in Psychiatric Epide
all cases, need to decide what is best in the
miology (eds. Hare, E. H., and Wing,J. K.). London:
light cf all the circumstancer. Moreover, we Oxford University Press.
have not here discussed the important rules of Epsmm, S., and COLEMAN, M. (1970). ‘¿ Drive theory of
management which can be derived from a schizophrenia.' Psychosomatic Med., 32, 113—40.
study of the processes leading to secondary GOODMAN, L. A., and Kaus@, W. H. (ig@4). ‘¿ Measures

handicaps (Wing and Brown, 1970). However, of association for cross classification.' 3. Amer. stat.
Assoc., 49, 732—64.
we would suggest that work like that described
HIRScH, S. R., and LEFF, J. P. (,g7,). ‘¿ Parental abnor
in this paper has important practical, as well as malities of verbal communication in the transmission
theoretical, aspects. of schizophrenia.' Psychol. Med., i@ 118-27.
KENDELL, R. E., COOPER, J. E., COPELAND, J. R. M. et al.
ACKNOwLEDGEMENTS (1972). Psychiatric Diagnosis in Xew Tork and London.
We are very grateful for the co-operation extended to Maudsley Monograph No. 20 (to be published by
us by the medical, nursing, records and social work staff Oxford University Press).
at Bethiem Royal, Bexley, Cane Hill, Maudsley and LEFF, J. P., and WING, J. K. (i 97 i). ‘¿ Trial of maintenance
St. Olave's Hospitals and of the local authority. Much of therapy in schizophrenia.' Brit. med. 3., 3, 599—604.
the social interviewing for this project was undertaken by MUELLER, J. H., and SCHUESSLER, K. F. (196,). Statistical
MissJudy Clarke, Miss Rita Lang, Mrs. Margaret Rayfield Reasoningin Sociology.Boston: Houghton and Muffin.
and Miss Susan Reid. The value of the project depends ROSENBERG, M. (1962). ‘¿ Testfactor standardization as a
very largely upon their skill. Miss Clarke also assisted with method of interpretation.' Social Forces, 4!, 53—61.
the data analysis.
—¿ (i968). The Logic of Survey Analysis. New York: Basic
The work was undertaken while the authors were
Books.
members of the scientific staff of the M.R.C. Social
Psychiatry Unit, Institute of Psychiatry, London. Ru'rrER, M. L., and BROWN, G. W. (i966). ‘¿ The reliability
and validity of measures of family life and relation
REFERENCES
ships in families containing a psychiatric patient.'
Soc. Psychiat., I, 38.
ATCHLEY, R. C. (i97o). ‘¿ A
qualification of test factor
VENABLES, P. H. (i 964). ‘¿ Performance and level of activa
standardization: a methodological note.' Social
tion in schizophrenics and normals.' Brit. 3. Psychol.,
Forces,49, 84—85.
55, 207—18.
BIRUtY,J. L. T., and BRowN, G. W. (‘970).‘¿ Crisis and life
changes preceding the onset or relapse of acute —¿ (1968). ‘¿ Experimental psychological studies of

schizophrenia: clinical aspects.' Brit. J. Psychiat., i i6, chronic schizophrenia,' in Studies in Psychiatry (eds.
Shepherd, M., and Davies, D. L.). London: Oxford
327—33.
BLALOCK, H. M. (ig6o). Social Statistics. New York: University Press.
McGraw Hill. —¿ and WING, J. L. (1962). ‘¿ Level of arousal and the

—¿ (1964). ‘¿ Controlling for background factors: spurious

subclassification of schizophrenia.' Archives of gen.
ness versus developmental sequences.' Social Enquiry, Psychiat., 7, 114—19.
34, 28—40. WING, J. K. (i@'o). ‘¿ standard
A form of psychiatric
BROWN, G. W. (1959). ‘¿ Experiencesof discharged chronic Present State Examination and a method of stan
schizophrenic mental hospital patients in various dardizing and classification of symptoms,' in Psychi
types of living group.' Milbank Man. Fund quart., 37, atric Epidemiology (eds. Hare, E. H., and Wing,J. K.).
105. London: Oxford University Press.
—¿ CARSTAIRS, G. M., and Toppiuo, G. G. (1958). ‘¿ The —¿ and FREUDENBERG, P. K. (i 961). ‘¿ The response of

post-hospital adjustment of chronic mental patients.' severely ill chronic schizophrenic patients to social
Lancet, ii, 685. stimulation.' Amer. 3. Psychiat., ii8, 311—22.
258 INFLUENCE OF FAMILY LIFE ON THE COURSE OF SCHIZOPHRENIC DISORDERS

WING, J. K., BENNETF, D. H., and DENHAM, J. (1964). WING,J. K., COOPER, J. E., and SARTORIUS, N. (1972).
The Industrial Rehabilitation of Long-stay Schizophrenic MeasurementandClasszficationofPsychiatricSymptomatology
Patients. Medical Research Council Memo No. 42. (Instruction Manual to P.S.E. and CATEGO)
London: H.M.S.O. (To be published).
—¿ BntLEY, J. L. T., COOPER, J. E., GitaHAa, P., and

Is@cs, A. D. (1967). ‘¿ Reliability of procedure for WYNNE, L. C. (ig68). ‘¿ Methodologic and conceptual
measuring and classifying “¿ Present Psychiatric issues in the study of schi.zophrenics and their
State―.'Brit. 3. Psychiat., 113, 499—515. familes,' in The Transmission of Schizophrenia (eds.
—¿ and BROWN, G. W. (ig7o). Institutionalism and Schizo Rosenthal, D., and Kety, S. London and New York:
plzrenia. Cambridge: Cambridge University Press. Pergamon.

APPENDIX I

Interrelation of various variables (y)

I2345678910II12I.

....—2.
Relapse
emotion..75—3.
Exp.
impair...4073—4.
Work
beh...328257—5.
Disturbed
....451048—o6—6.
Symptoms
admiss...41—40—44—63i8—7.
Accept.
hsehold...344321—05—o6‘7—8.
Type
contact..29—33—17—1714i6—38—9.
Amount
....—4'—4212—362530o621—io.Sex..
Drugs
....59280105371245o804—ii.
Prey.
73319—12. admiss...6i51580212—2044— I

Age at admiss...45—20—43282843I
7—II00592I—I3.Impalrmentor478@——i8—5715—ii—IIi63217disturbance..

Italic numbers = Statistically significant (.05 level).

Key
i. i Yes; 2 No 5. I Typical; 2 other 9. Yes;2 No
2. I High; 2 Low 6. Accept; 2 Reject 10. I Male; 2 Female
3. i Yes; 2 No 7. Par/sib; 2 Marital II. , Yes; 2 No
4.I Yes;2 No 8. I High; 2 Low 12. 1—45; 2—45+

A synopsis of this paper was published in the April 1972 Journal.

G. W. Brown, Ph.D., Reader in Sociology, Bedford College, University of London, Regent's Park, N. W. i.
J. L. T. Birley,B.M.,M.R.C.P.,
M.R.C.Psych.,
Dean,Instituteof Psychiatry,
De Crespigny
Park,London,
S.E.5
J. K. Wing, M.D.,Ph.D.,F.R.C.Psych.,
ProfessorSocialPsychiatry,Instituteof Psychiatryand London
School of Hygiene; Director, M.R.C. Social Psychiatry Unit, Institute of Psychiatry, De Crespigny Park,
London, S.E.5

(Received 26 October i,çi',)