Download this article

This is the first piece in an exchange of views between Robert Harris and George W. Brown, originally published in Medical Sociology News Volumes 5 and 6, between January 1978 and January 1979. All four parts are published here, together, for the first time.

Why I Am Browned Off with Quantitative Methodology
A comment on Brown, Birley‚ Wing (1972) and Vaughn and Leff (1976)

Robert Harris
Medical Sociology Research Centre
University College of Swansea, UK

Originally published in MSN Volume 5, Issue 1, January/February 1978

Recently‚ two papers have appeared in the British Journal of Psychiatry which together claim to have established a new fact to which psychiatrists are enjoined to pay regard when making discharge decisions (Brown, Birley and Wing, 1972; Vaughn and Leff‚ 1976). The first of these studies purports to have found a causal relationship between the nature of the affective environment of discharged schizophrenics and their readmission to hospital. The second is a replicative study which substantiates what the authors call the unequivocal nature of the results of the first. On the face of it the results do indeed appear quite convincing. However‚ more detailed examination reveals that what the authors have demonstrated is a common-sense observation which I suspect has always been recognised by psychiatrists’ and others concerned with the discharge and readmission of psychiatric patients.

Of course if researchers wish to spend their time scientising common sense judgements that is entirely their own affair. It becomes a matter of debate when their findings are presented as scientific facts which should be taken into account by practitioners when deciding upon disposal options. Furthermore‚ when a common-sense judgement masquerades as a scientific fact, it gains a spurious authority which cannot be gainsaid by available put-downs such as ‘that's just an opinion’, or ‘that is generally true‚ but it doesn't apply in this case' and so forth. My personal commitment in addressing the following comments towards these studies is that I regard the form of scientism which they represent to be a positive hindrance to our understanding of schizophrenics and their families. Vaughn and Leff’s study stands or falls on the merits of the earlier research carried out by Brown et al. My remarks‚ therefore‚ are directed entirely to this. The principal findings reported by Brown et al. is shown by the following 2 x 2 contingency table. I address to this table two questions: 1. How was it produced? 2. What does it mean?

Relationship of relatives' emotion to relapse in the 9 months after discharge

Stacks Image 14059
The table shows a statistically significant direct relationship between the level of emotion expressed by members of a household group living with a discharged schizophrenic patient towards that patient and the patient’s relapse. The higher the expressed emotion‚ the more likely is the patient to relapse. Relatives' expressed emotion is an overall index arrived at by combining three indicators of emotional response towards the discharged patient. Assessment of the relative strength of these indicators were made by analysing data produced in a lengthy family interview. The three indicators are: critical comments‚ hostility and the emotional over-involvement of the relative. Relatives were categorised as expressing high or low emotion according as to whether or not they made seven or more critical comments about the ex-patient, expressed or did not express hostility about him or showed marked or low emotional over-involvement with him.

The most significant of these indicators in demonstrating high expressed emotion is the first‚ i.e. seven or more critical comments. This accounted for 35 of the relatives in that category. Hostility accounted for two and emotional over-involvement for five. The three remaining numbers of the high EE group were added following a joint interview, i.e. an interview where the patient was present. The criteria for allocating relatives into the high or low EE groups following this interview were different from those used following the joint interview. In particular‚ two or more critical comments were sufficient for allocation into the high EE group. This criterion added one relative, marked over-involvement added two, and hostility none. Hence 37 out of 45 patients lived with relatives who were assessed as being critical or hostile towards them.

Thus half of the answer to my first question‚ 'how was the association produced?'‚ is as follows: Certain discharged patients and their relatives were interviewed. What they said at the interview and how they said it was judged to show hostility‚ over-involvement and criticism. If there was judged to be
any hostility or marked over-involvement or seven or more critical comments the relatives were said to be expressing high emotion. To complete the answer to the first question we need to know how relapse was measured. Brown simply says that relapse was judged using all available information. Presumably this includes interview material and in the case of readmitted patients‚ hospital records. Twenty nine out of thirty five relapsed patients were readmitted. So much for how the table was produced. Let us now consider what it means. First of all, what do Brown and his colleagues think it means? They are in no doubt:

"…a high degree of expressed emotion is an index of characteristics in the relatives which are likely to cause a florid relapse of symptoms, independently of other factors such as length of history, type of symptomatology or severity of previous behaviour disturbance“ (Brown et al. p 242. My emphasis)

Brown and his colleagues believe they have discovered a causal relationship between EE and relapse. Furthermore, EE is a measurement of characteristics in the relatives. It is thus a psychogenic variable. Later on in the same paper‚ however‚ Brown and his colleagues speak of the level of EE as ‘an enduring potential characteristic of the relative's behaviour towards the patient' (p 246). I am not at all sure that I know what they mean by an enduring potential characteristic of behaviour. Perhaps they mean that EE in an index of personality characteristics which dispose or cause the relative to behave in a particular way towards the patient. This interpretation accords well with the first statement quoted above. However‚ my interpretation does imply that the second statement is elliptic and that this ellipticism obscures the model of relatives’ behaviour on which Brown's study appears to be based. The implicit model seems to be this: Relatives have inside then certain characteristics. When exposed to a certain stimulus - the patient - these characteristics are activated. Once activated they cause the relative to behave in a particular way which, in turn‚ causes the patient to relapse. Diagrammatically Brown’s model of relatives’ behaviour seems to be this:

Stacks Image 14069

As has been frequently pointed out explanations like this deny the volitional aspects of human action and for that reason alone must be rejected as satisfactory explanations of that action. There in no denying the fact‚ however‚ that the strong statistically verified relationship shown in the table above remains. If Brown’s explanation for the relationship‚ i.e. one based on a mechanistic view of man, is inadequate‚ then what might be a more adequate explanation?

An alternative way of accounting for the association between EE and relapse is given in diagram 1. This diagram is not to be seen as a flow chart if by that is meant a sequential series of stages. Rather it is an attempt to render intelligible, a process (schizophrenic relapse) by seeing it as the praxis of relatives and patients and others with whom they come into contact‚ particularly psychiatrists. This way of seeing is, of course‚ that recommended by Laing and Esterson (1971).

As Brown himself states‚ 'expressed emotion has a highly negative connotation' (p 253). In other words‚ when Brown assesses a high level of expressed emotion he is saying‚ in effect‚ that the relatives don't like the patient very much. They are critical of him and hostile towards him.

Brown relies upon his own and his interviewers' common-sense ability to recognise hostile and critical comments when they hear them‚ to see hostility in gestures and hear it in tones of voice. There is nothing exceptional in this ability. We all have it and we use it. However‚ neither we in our everyday lives nor Brown and his interviewers in their research activity are invariably right in judging that A dislikes B. The first point to notice then is that when Brown assesses high or low EE he may be wrong or right. That is to say‚ relatives may‚ in fact‚ like the patient when Brown assesses that they don't and vice-versa. I assume that within the families studied by Brown there is a state of affairs recognised by them by which Brown may be said to be right or wrong. It seems to me to be silly to deny that our common-sense judgements of relationships may be incorrect. Insofar as Brown believes his methodology to be valid (which he does) he is denying that he may be wrong. In a separate paper (Rutter and Brown, 1966) devoted to the issue of the reliability and validity of the research instruments used in the study discussed here, Brown and his co-author describe in great detail the procedures followed to ensure validity and reliability. For him, therefore‚ expressed emotion (dislike) exists entirely
in and through the methodical practices by which it is recognised and described (Garfinkel‚ 1967).

Brown Assesses Expressed Emotion of Relatives

Stacks Image 14087
  1. He is more likely to be right since we can recognise dislike when we see it. Note that intuitive judgements and feelings disallowed.
  2. No reported cases in Brown’s series.
  3. Brown’s paradox: High E.E.: Severe Disturbance, No relapse: Patient follows medical regimen.
  4. Crucial assumption that a diagnosis of relapse is more likely following re-admission.
  5. No relapse but not because low E.E. Such cases though assumed to support Brown's theory.

Thus‚ I accept that for Brown, relatives’ dislike of patients exists in and through his methods of recognising and describing it. I do not accept that these methods constitute the totality of the affective components of the households he studies. In short‚ Brown may be wrong. He is‚ though‚ I submit‚ more likely to be right than wrong. This is not because of the excellence of his research instruments as Brown believes‚ but simply because both he‚ his co-authors and his 'hired hand' researchers are ordinary‚ competent members of society who can, as a matter of fact, recognise‚ a criticism when they hear it. But while Brown relies on common-sense to recognise a critical comment‚ at other points in his research design common-sense is thrown out of the window as being unscientific. For example, in treating critical comments as ordinal data he assumes that one critical comment carries as much censure as any others. Again‚ he denies the ambiguity of family relationships and characterises them as either hostile or not hostile. His interviewers were counselled against allowing their 'feelings' to interfere with their judgements of the affective aspects of the households they visited. All these offend common-sense. Thus, although Brown is likely to be right in his assessment of dislike as long as he relies on common-sense‚ insofar as he rejects common-sense, he is more likely to be wrong. The more scientific he is, the more wrong he is likely to be. For reasons which will become plain as we proceed, I believe that the strong association between EE and relapse is due entirely to the greater likelihood that Brown is right when he assesses that a patient is not liked very much.

Let us assume that Brown is right when he assesses that a patient is not liked very much. Further, let's try and see the family situation from the point of view of the patient by imaginatively taking his position. If we are in a situation where we are not liked we may reach a point where we try to escape from it. Escape may involve physical removal from the situation or social withdrawal. The second is effective only as far as the others allow you to withdraw. Both forms of withdrawal require a level of command over scarce resources‚ namely accommodation‚ either a separate and private room in the household home‚ or accommodation outside the household home. We need to ask, therefore: Has the disliked patient somewhere to go apart from hospital? If the answer is 'Yes' the patient leaves the situation in either of the above senses. If the answer is ‘No’, we than have to ask: Does the patient resist readmission? If the answer is ‘Yes’ we have then to ask two questions. Firstly‚ what resources are available to the patient to resist readmission and remain sane in the household? Whatever the available resources‚ among them is likely to be one of the major tranquillisers routinely prescribed for discharged schizophrenics. Secondly‚ what is likely to occur in a household in which the patient is disliked and refuses to leave? My common-sense hunch would be that the level of conflict in the household is likely to increase. We would not be surprised‚ therefore‚ to find some cases of high dislike‚ relatives’ reports of severe disturbance in the patient‚ the patient following the medical regimen (he takes tranquillisers because he himself defines his need for them) and no relapse (if he relapses the patient knows he will be readmitted). And this is precisely what we do find among Brown's data. Since Brown's implicit theory effectively denies human volition he regards this state of affairs as a paradox. We now see it not as paradoxical at all but as a state of affairs brought about by specifically human action.

The disliked patient, however, may not resist readmission and may indeed, actively seek it as a means of egress from the household. Should he be readmitted I hypothesise that
either he will be assumed to have relapsed, or any assessment of his behaviour will be biased towards a finding of relapse once the patient has been readmitted. I regard this hypothesis as plausible on three counts: Firstly, relapse was assessed by Brown using all available information. Presumably this includes relatives’ reports of the patient’s behaviour. Relatives who dislike the patient are more likely to define the patient’s behaviour in a way that renders it perceivable as symptomatic as a means of extruding him from the household. Secondly‚ the patent himself may mimic psychiatric symptoms as a means of escaping from the household (Braginsky and Braginsky, 1969). The belief of medical and nursing personnel that psychiatric patients can act insane is documented in the literature and the simulation of psychiatric symptoms is a recognised syndrome in clinical psychiatry: the Ganser syndrome. Thirdly‚ other research has shown psychiatric diagnosis to be manifestly influenced by social context. The patient’s presence in the hospital requires psychiatric legitimation. A finding of relapse is, therefore, likely.

An important feature of this rendition of Brown’s significant association is that the problem of cases which don't fit Brown’s hypothesis disappears. Cases which deviate from Brown‘s hypothesis are cases either where his assessment of dislike is wrong or the patient has somewhere to go or the patient resists readmission.

It might of course‚ be argued that even allowing the plausibility of my rendition of Brown‘s research I have not disproved his hypothesis. This is perfectly true‚ but do we really need research of this kind to convince us that discharged patients living with people who don't like them are more likely to be readmitted than those who live in a more sympathetic emotional atmosphere? If we do, we seem in great danger of substituting a mystifying welter of reified statistics for our common sense.

Stacks Image 32

Braginsky‚ R. and Braginsky ‚ B. (1969) Methods of Madness: The Mental Hospital of Last Resort, New York, Holt, Rinehart and Winston.
Brown, G.‚ Birley‚ J. and Wing‚ J. (1972) Influence of family life on the course of schizophrenic disorder: a replication". British Journal of Psychiatry‚ 121, 241-258.
Garfinkel‚ H. (1967) Studies in Ethnomethodology‚ New Jersey, Prentice Hall.
Laing, R. D. and Esterson‚ A. (1971) Sanity, Madness and the Family‚ second edition‚ London, Penguin.
Rutter‚ M. L. and Brown, G. W. (1966) The reliability and validity of measures of family life and relationships in families containing a psychiatric patient‚ Social Psychiatry‚ 1‚ 38-53.
Vaughn‚ C. E. and Leff‚ J. P. (1976) The influence of family and social factors on a course of psychiatric illness: A comparison of schizophrenic and depressed neurotic patients‚ British Journal of Psychiatry‚ 129, 125-237.

Next: 'Science and Common Sense: A Reply' by George Brown