Psychological Reports, 1990, 67, 279-287.

O Psychological Reports 1990

DEVELOPMENT AND PRELIMINARY VALIDATION O F A QUESTIONNAIRE ASSESSMENT O F EXPRESSED EMOTION ' NANCY M. DOCHERTY AND MARK R. SEWER

State Universiiy of New York at Binghamton PHILIP D. HARVEY

Mount Sinai School of Medicine Summary.-A questionnaire for measurement of expressed emotion was constructed. I t consisted of two scales, criticism and emotional overinvolvemenr. The questionnaire and the Camberwell Family Interview were then administered to relatives of schizophrenics, and preliminary assessments of scale reliabilities and concurrent validities were done. Both questiormaire scales were reliable. With respect to validity, the criticism scale correctly classified 88% of the subjects relative to the criticism scale of the interview. The emotional over~nvolvernentscale was less satisfactory (67%) but still assisted in classification of over-all expressed emotion. Expressed emotion, classification by the questionnaire, correctly identified 84% of the subjects with respect to the interview.

In 1958 Brown, Carstairs, and Topping found evidence that the rate of relapse among schizophrenic patients during the year following hospital discharge was related to the living situation into which they were discharged. Subsequent research showed the likelihood that a schizophrenic patient will suffer relapse depends in part on the attitudes of the relatives (parents or spouse) with whom he lives. A highly critical or emotionally overinvolved attitude on the part of a relative toward a patient predicts a higher likehood of relapse. Such attitudes have been designated high "expressed emotion." The Camberwell Family Interview was developed for assessing expressed emotion (Brown & Rutter, 1966; Rutter & Brown, 1966; Vaughn & Leff, 1976a, 1976b). This is a semistructured interview designed to tap the attitudes in question by asking relatives about the patient's behavior and the quality of the family relationships. The interview can be rated reliably on several dimensions. The dimensions which have been identified as predictive of relapse are criticism, hostility, and emotional overinvolvement. Criticism, which is the clearest predictor of relapse, refers to critical comments about specific behaviors, while hostility refers to a more generalized negative attitude toward the patient. These two ratings have considerable overlap. Both may occur together or there may be a high rating on criticism without a high rating on hostility. Very rarely does a high rating on hostility occur without a

'Address reprint requests to Philip D. Harvey, Ph.D., Department of Psychiatry, Mount Sinai School of Medicine, One Gustave L. Levy Place, New York, NY 10029.

280

N.M. DOCHERTY, ETAL.

high rating on criticism as well, and so rated hostility does not assist significantly in predicting relapse over and above ratings of criticism. Criticism ratings on the Famdy Interview are generated by actually counting critical comments made during the interview, as defined by criteria involving voice tone (emphasis, repetition, and changes in speed, pitch, and volume) and content. Emotional overinvolvement reflects overconcern, overprotectiveness, and overidentification and is rated on a 6-point scale. A relative is designated as high on expressed emotion if scores are high on criticism, emotional overinvolvement, or both. One stumbling block to research on expressed emotion has been the difficulty of assessing it. Standard procedures have been developed for administering and rating the Family Interview, and interrater reliability is typically .80 or better, but training, interviewing, and rating are long and tedious processes. The interview itself, in short form, takes one or two hours to administer to each relative, and twice that long to rate. Training to rate reliably generally takes about 70 hours, usually over a period of a month or more. Besides requiring extensive training and time of the researcher, such studies require subjects who are willing to participate in long interviews. Clearly a shorter and simpler method of assessing expressed emotion would be helpful. There have been some attempts at this, but none has succeeded in attaining ratings highly correlated with Family Interview ratings as yet, and none has been shown to predict relapse as well as the Family Interview. Kreisman, Simmens, and Joy (1979) developed the Patient Rejection Scale, an 11-item questionnaire whch overlaps conceptually with the criticism/hostility dimension of the Family Interview. This scale was administered to one relative (only) of each of 133 discharged schizophrenic patients. High scores on this scale correlated low (r = 20) with hospital readmission within four months. However, readmission is not an ideal measure of relapse, since (1) rejecting relatives might be more likely to have the person readmitted with less cause and (2) symptoms may exist in severe form in nonhospitalized schizophrenics. An attempt to rate expressed emotion from a 5-min. speech sample correlated fairly well with the Interview ratings (Magana, Goldstein, Karno, Jenluns, & F d o o n , 1986). Persons rated as high on expressed emotion on the speech sample were almost always high on the Interview as well. However, about one-third of those rated as low on the speech sample were high on the Interview, with the former underestimating the number of subjects high on expressed emotion. Magana, et al. suggest that this measure could be used for screening, with those rated as showing high expressed emotion in the speech sample not n e e l n g further evaluation and those rated as showing low expressed emotion requiring assessment with the Interview. The present study was another attempt to begin to develop a more con-

QUESTIONNAIRE ON EXPRESSED EMOTION

281

venient means of assessing expressed emotion that could be used for research or as an aid in screening for intervention. A self-report questionnaire was formulated, to be filled out by the relatives who lived with schizophrenic patients. The questionnaire had two different scales to measure the two dimensions predictive of relapse on the Interview. Items were generated to assess critical or emotionally overinvolved attitudes and were presented to parents or spouses of schizophrenic patients. The Interview was also administered to the same subjects and rated on the relevant dimensions. An analysis was performed to estimate the questionnaire scale reliabilities. Item-to-scale correlations were computed, and poorly correlated items were deleted. Correlations between questionnaire scores and Interview ratings were then computed to determine, in a preliminary way, the concurrent validity of the questionnaire as a measure of expressed emotion.

Subjects Subjects were 25 relatives of schizophrenic psychatric patients. The relatives were parents or spouses who were currently living with the patient (17 of 25) or who had lived with the patient at some time during the last two years (8 of 25). None of the relatives had an history of psychiatric hospitalization. Patients included both in- and outpatients, all of whom had been hospitalized at some time, all but one at a state psychiatric hospital. Appropriate patients (16) were acquired from four sources: patients on the admission wards of a state psychiatric hospital (7); subjects in another research project involving a 6-mo. follow-up of acute state hospital patients (3); patients who had been discharged from the hospital and were attending a state psychiatric outpatient clinic (excluding any who had not had an acute episode in the last year) (4); and patients whose relatives belonged to a local support group for f a d e s of mental patients (2). appropriate patients from these sources who had relatives meeting the criteria were asked for permission to contact their relatives about the project, and the relatives were contacted and asked to participate. The sample of relatives obtained included 12 mothers, 9 fathers, 3 wives, and 1 husband. Descriptive information on patients and relatives is presented in Table 1. Diagnosis of the patients was determined using the Schedule for Affective Disorders and Schizophrenia (Spitzer, Endicott, & Robins, 1978). This interview was administered to 12 of the 16 patients by a graduate student trained in its use. For the remaining four patients, who were not available for interview, the schedule was completed using hospital records. Each of these records included past and current psychiatric, psychological, and social assessments, progress notes, and hospital diagnoses, among other things. The schedule was completed based on symptoms as described in the

282

N. M. DOCHERTY,ETAL.

records. Of the patients not available for interview, only those whose records contained enough information to complete the schedule and to warrant an unequivocable diagnosis of schizophrenia were included. Using the completed schedules, diagnoses based on DSM-I11 criteria were generated independently by a doctoral level clinical psychologist and a group of graduate students. Interrater agreement, based on a larger sample of which the present is a subset, was suitable (K= .88). Differences were discussed. and a consensus was reached. All subjects in the sample received diagnoses of schizophrenia, active or residual TABLE 1 D E S C R I P ~INFORMATION ~~E Patients

M n

Age, yr. % Female Education, yt Number of Hospitalizations

16 36 25 13 3.2

Range

M

Relatives Range

25 21-56

58

30-80

60 6- 16 1- 7

14 0

7-20 0

Questionnaire Questionnaire items were each rated on a 4-point scale (0 =never or almost never, 1 = occasionally, 2 = often, 3 = always, or almost always). There were two scales, criticism and emotional overinvolvement. The items and instructions were worded to create the impression that the purpose of the questionnaire was to find out what the patient was like at home and how this affected family members. This presentation was similar to that of the Interview. Some negatively toned items were reversed such that there were roughly equal numbers of positive and negative items to avoid an over-all negative tone. Some items were based on content areas covered by the Interview, and some were formulated on the basis of current hypotheses as to characteristics distinguishing relatives high and low on expressed emotion: level of intrusiveness, emotional response, attitude toward the illness, and level of tolerance and expectations (Vaughn & Leff, 1981). The initial questionnaire consisted of 144 items. The criticism scale had 99 items, some of which concerned specific criticisms (e.g., "She looks so unkempt, it bothers me"), and some were aimed at more generalized feelings of hostility or rejection (e.g., "I would rather she didn't live with me"). There were also items attempting to obtain directly relatives' self-reports of criticism (e.g., "I criticize her behavior"). Some items were concerned with relatives' beliefs as to how much control they had over the patient's condition (e.g., "I try my best to get her to act normal") and how much control the patient has over his own behavior ("She is more competent than she pre-

QUESTIONNAIRE ON EXPRESSED EMOTION

283

tends to be"; see Berkowitz, Eberlein-Vries, Kuipers, & Leff, 1984; Hooley, 1985; Vaughn & Leff, 1981). There was one item relating to the item that correlated with expressed emotion on Berkowitz, et al.'s (1984) Knowledge Interview, to the effect that the patient is or is not his normal self between episodes ("Between episodes, she behaves normallyn). The emotional overinvolvement scale had 45 items relating to the three areas defined as reflecting emotional overinvolvement in the Interview: overprotectiveness (e.g., "I have to stick up for her, because she won't stick up for herself"), overconcern (e.g., "My life centers around her"), and overidentification (e.g., "She and I get so close we are like one person"). Included in these were items concerned with emotional overreaction (e.g., "I am a nervous wreck from worrying about her"), interdependency (e.g., "She and I take care of each other"), and extreme devotion and self-sacrifice (e.g., "I have made many sacrifices for her"). Procedure Relatives who agreed to participate were sent the questionnaire by mail. About two weeks after they had completed the questionnaire, the Interview was administered to them in their homes by the first author. Usually the Interview focuses on the three-month period prior to hospitalization, but because some subjects in this study were not currently hospitalized, the focus in this study was on the period around the most recent episode, which was always within the last year. Although a few of the patients were not living with their relatives at the time, all had been in close contact with them. Interview tapes were rated for criticism and emotional overinvolvement, blind to questionnaire scores, by the same author, who had been trained on the rating system by Robert Cole, Ph.D., and who had attained a reliability level of .96 for criticism ratings and .80 for emotional overinvolvement ratings on criterion Interview tapes.

ScaIe Reliability There were several steps in the analysis. First the over-all internal consistencies of the two scales were assessed using coefficient alpha. Second, a cluster analysis was performed on the items from the completed questionnaires. Scale-to-scale correlations were computed, and item-to-scale analyses were done to test each item's correlation with its cluster score. Items poorly correlated with their cluster scores were then removed. Item-to-item analyses were also done, to test for redundancy of items within the scales. No items were highly redundant, so none were removed on that account. All 25 subjects were included in the analyses of criticism scale scores. The four spouses were excluded from the analyses of emotional overinvolvement since this typically applies only to parents.

284

N. M. DOCHERTY, ETAL.

Initial estimates of internal consistency were high for both of the questionnaire scales. For the 99-item criticism scale alpha was .94 and for the 45-item emotional overinvolvement scale alpha was .85. After deleting the poorly correlated items ( r < .3O), the revised scales consisted of 70 and 29 items, respectively. The mean item-to-scale correlation of items retained was .52, and for those items deleted r was .03. For these modified scales of criticism alpha was .96 and of emotional overinvolvement alpha was .90. The two scales were not significantly correlated with each other ( r = -.33, p.80) and for the questionnaire scale ( >.96). Two of the three misses in the dichotomized correlation between questionnaire and Interview criticism ratings were based on two subjects who scored well into the high range on the questionnaire but made no critical comments on the Interview. The most likely explanation for this discrepancy is that the face-to-face nature of the interview inhibited these subjects from voicing the criticisms expressed in their questionnaire responses. If this is true, then some subjects' attitudes may be more accurately reflected in their questionnaire responses than by their Interview scores. It would be necessary to test the predictive validity of the questionnaire scale with respect to relapse to determine whether this is the case. The emotional overinvolvement scale scores were less satisfactory than those of the criticism scale. This was to be expected, since the former construct is more complex and less clearly defined than is criticism. It involves overprotectiveness, overconcern, and overidentification. One, two, or all three of these facets are present in various forms in high-scoring subjects. The high-scoring population as defined is rather diverse, and it is not yet clear whlch aspects of the construct are most predictive of relapse. Further clarification of the relevant aspects of emotional overinvolvement would make it easier to assess accurately by questionnaire, interview, or any other means. Patients and subjects included in this study varied a little from the usual sample in research on expressed emotion. It was necessary to modify the usual criteria somewhat to have a sample of 25. Patients as a group tended to be somewhat older, and some of the relatives (8 of the 25) no longer lived with the patients. I t was to be expected that expressed emotion, and especially the numbers of critical comments made, would be somewhat lower in such a sample. There was less immediate pressure on the relatives, because patients' symptoms were less acute and contact was somewhat less frequent for the group as a whole. The mean number of critical comments of 4.24 (somewhat lower than usual) is not surprising. However, this discrepancy from more typical results on expressed emotion highlights the need to bear in mind the limitations of the small, imperfect sample in interpreting the present findings. While results were promising in this preliminary- study, especially for the criticism scale, further work needs to be done toward validation of the questionnaire. This study could be replicated using the modified scales, with a larger sample of subjects more typical of past research on expressed emotion, that is, only those whose schizophrenic family member was very recently hospitalized and who had lived with the patient for the 3-mo. period prior to hospitalization. Ideally, separate samples could be used for the reliability and validity analyses. The cut-off scores derived here could be tested and perhaps modified to suit the more typical sample. I t might also

QUESTIONNAIRE O N EXPRESSED EMOTION

287

be possible to shorten the scales by deleting more items, given the h g h internal consistency found even with our very small sample. Most importantly, to extend the criterion validation beyond simple concurrent validity, the rate of relapse needs to be assessed. Although scores on the questionnaire predicted scores on the Interview fairly well in this preliminary study, and scores on the Interview have predicted relapse in past research, this does not necessarily mean that questionnaire scores will predict relapse. Before the questionnaire can be used with confidence, its predictive validity must be demonstrated. If the questionnaire is indeed a predictor of relapse, it could be a valuable research tool, both for easier assessment of expressed emotion and for refinement of the construct. REFERENCES BERKO~Z R.,, EBERLEIN-VRIES, R., KUIPERS,L., & LEFF, J. P. (1984) Educating relatives about schizophrenia. Schizophrenia Bulletin, 10, 418-429. BROWN,G. W., BIRLEY,J. L., & WING,J. K. (1972) Influence of family life on the course of schizophrenic disorders: replication. British Journal of Psychiatry, 121, 241-258. BROWN, G. W., CARSTAIRS, G. W., & TOPPING,G. C. (1958) The post-hospital adjustment of chronic mental patients. L n c e t , 2, 685-689. BROWN,G. W., MONCK,E. M., CARSTAIRS, G. W., & WING, J. K. (1962) The influence of family life on the course of schizophrenic disorders. British Journal of Preventive Social Medicine, 16, 55-68. BROWN,G. W.. & R ~ RM., L. (1966) The measurement of family activities and relationships. Human Rekations, 19, 241-263. HOOLEY,J. M. (1985) Expressed emotion: review of the critical literature. Clinical Psychology Review, 5 , 119-140. KREISMAN,D. E., S ~ E N SS., J., &JOY, V. D. (1979) Rejecting the patient: preliminary validation of a self-report scale. Schizophrenia Bulletin, 5, 220-222. LEFF, J., KUIPERS, L., BEFSCOWITZ, R., EBERLEIN-VRIES, R., & STURGEON, B. (1982) A controlled trial of social intervention in the families of schizophrenic patients. British Journal of Psychiatry, 141, 121-134. MAGANA, A . B., GOU)STEM,M. J., KARNO,M., J E ~ s J., , & FALLOON, I. R. H. (1986) A brief method for assessing expressed emotion in relatives of psychiatric patients. Psychiatry Research, 17, 203-212. R m , M., & BROWN,G . W. (1966) The reliability and validity of measures of family life and relationships in families containing a psychiatric patient. Social Psychia~ry,1, 38-53. SPITZER, R. L., ENDICO'IT,J., & ROBINS,L. (1978) The Schedule for Affective Disorders and Schizophrenia. New York: Biometrics Research. VAUGHN, C. E., & LEFF, J. P. (1976a) The influence of family and social factors on the course of psychiatric illness. British Journal of Psychiatry, 129, 125-137. VAUGHN,C. E., & LEFF,J. P (1976b) The measurement of expressed emotion in the families of psychiatric patients. British Journal of Social and Clinical Psychology, 15, 157-165. V A U G I C. ~ , E., & LEFF, J. P. (1981) Patterns of emotional response in relatives of schizophrenic patients. Schizophrenia Bulletin, 7, 43-44.

Accepted July 24, 1990.

Development and preliminary validation of a questionnaire assessment of expressed emotion.

A questionnaire for management of expressed emotion was constructed. It consisted of two scales, criticism and emotional overinvolvement. The question...
363KB Sizes 0 Downloads 0 Views