- Email: [email protected]
Evaluation of expressed emotion (EE) status in mood disorders in Japan: inter-rater reliability and characteristics of EE
Psychiatry Research 94 Ž2000. 221᎐227
Evaluation of expressed emotion ž EE/ status in mood disorders in Japan: inter-rater reliability and characteristics of EE U
Yoshio Minoa, , Shimpei Inoueb, Shinji Shimoderab, Shuichi Tanakac a
Department of Hygiene and Pre¨ enti¨ e Medicine, Okayama Uni¨ ersity Medical School, 2-5-1 Shikata-cho, Okayama 700-8558, Japan b Department of Neuropsychiatry, Kochi Medical School, Kohasu, Oko-cho, Nankoku 783-8505, Japan c Department of Psychiatry, Watarigawa Hospital, Gudo, Nakamura 787, Japan Received 20 January 1999; received in revised form 26 May 1999; accepted 12 July 1999
Abstract The reliability of expressed emotion ŽEE. ratings by the Camberwell Family Interview ŽCFI. and characteristics of EE were evaluated in families of patients with mood disorders in Japan. The subjects were 27 patients with mood disorders and 31 members of their families. The CFI was carried out with the family members. EE was rated by two raters independently, and the inter-rater reliability was evaluated according to Spearman’s correlation coefficient by ranks and the -value. The distribution of subscales of EE in these subjects was compared with that in families of patients with schizophrenia in Japan and families of patients with mood disorders abroad. Concerning critical comments ŽCC., hostility ŽH., and emotional over-involvement ŽEOI., which are important for EE rating, Spearman’s correlation coefficient and the -values were 0.4᎐0.8, and the reliability of EE ratings in mood disorders was not high. The proportion of positive agreement was particularly low in H and EOI. CCs were fewer in families of Japanese patients with mood disorders than in those with schizophrenia or families of American or European patients with mood disorders. Re-evaluation of the inter-rater reliability of EE ratings in mood disorders is needed. Expressed emotion was more suppressed in families of patients with mood disorders than in those of patients with schizophrenia. Expressed emotion was also more reserved in the Japanese subjects than in their Western counterparts. 䊚 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Depression; Bipolar disorder; Family; Camberwell Family Interview; Transcultural psychiatry
U
Corresponding author. Tel.: q81-86-235-7171; fax: q81-86-235-7178. E-mail address: [email protected] ŽY. Mino.. 0165-1781r00r$ - see front matter 䊚 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 5 - 1 7 8 1 Ž 0 0 . 0 0 1 3 2 - 3
222
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227
1. Introduction Expressed emotion ŽEE. in patients’ families has been studied concerning schizophrenia, and effects of EE on the course of schizophrenia have been confirmed in a number of cultural regions ŽLeff and Vaughn, 1985; Bebbington and Kuipers, 1994; Butzlaff and Hooley, 1998.. In Japan, which is a different cultural region compared with Western societies, relations of EE with relapse of schizophrenia, social function, and depressive symptoms have also been clarified ŽTanaka et al., 1995; Inoue et al., 1997; Mino et al., 1997, 1998.. However, the relationship between EE of families and psychiatric disorders other than schizophrenia has not been sufficiently described. The relations between EE and mood disorders including depression ŽVaughn and Leff, 1976; Hooley et al., 1986; Okasha et al., 1994; Uehara et al., 1996., in particular, must be evaluated for the following reasons. First, there is a high prevalence of mood disorders including depression. For example, the life time prevalence of major depressive disorders in the general population is 5᎐12% in males and 10᎐25% in females, and their point prevalence is 2᎐3% in males and 5᎐9% in females ŽAmerican Psychiatric Association, 1994.. The point prevalence of mood disorders is high also in the primary-care setting, and research is ongoing concerning how to screen and effectively treat them ŽCooper and Eastwood, 1992; Mino et al., 1994.. Suicide ŽGelder et al., 1996. and days lost from work ŽBroadhead et al., 1990. due to depression have also become major issues. If family intervention with mood disorders based on EE research succeeded, there would be considerable benefit not only to patients but also to society. Secondly, depression is reported to be a disease strongly influenced by the social environment ŽBrown and Harris, 1978., and effects of EE, which is a manifestation of the patient’s familial environment, cannot be ignored. The reliability of EE ratings has been studied ŽWig et al., 1987a; Orhagen and d’Elia, 1991; Bentsen et al., 1996.. Training in EE rating in Japanese has been evaluated, and a satisfactory inter-rater reliability could be attained by a training program ŽMino et al., 1995a.. However, these
studies were based on interviews with families of schizophrenic patients, and there have not been many studies in which the inter-rater reliability of EE ratings was evaluated on the basis of the Camberwell Family Interview ŽCFI. ŽLeff and Vaughn, 1985. in families of patients with mood disorders. Moreover, the distribution of EE and its subscales has been reported to differ among the cultural regions to which the patients and their families belong in studies of EE in schizophrenia ŽLeff and Vaughn, 1985; Mino et al., 1995b.. Although culture is considered to exert similar effects on EE also in families of patients with mood disorders, few studies have approached this problem. This study: Ž1. examined the inter-rater reliability of EE ratings in families of patients with mood disorders; Ž2. compared EE and its subscales between families of schizophrenic patients and those of mood-disordered patients in the same society; and Ž3. compared EE and its subscales in families of mood-disordered patients among different cultures.
2. Materials and methods The subjects were part of the patients employed for evaluation of the relationship between the course of mood disorders and EE of the families of patients. They included 27 patients with mood disorders, who were chosen from the entry order, treated at the Department of Psychiatry, Kochi Medical School and Tosa Hospital between April 1997 and April 1998, and 31 of their family members. The diagnosis was made on the basis of DSM-IV ŽAmerican Psychiatric Association, 1994. and ICD-10 ŽWorld Health Organization, 1992.. To evaluate the inter-rater reliability of EE ratings, two raters independently evaluated the data obtained by the CFI of the 26 family members of the first 22 subjects. One of the two raters had been formally trained in EE rating and had been authorized as a rater, and the other was his trainee who attained a satisfactory level of interrater reliability Ž r s 0.86᎐0.95, P- 0.001.. The
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227
training was carried out primarily using data derived from interviews with families of schizophrenic patients. Rating was performed by the same method as the CFI in families of schizophrenic patients ŽMino et al., 1995a., using tapes in which the interviews were recorded and their transcripts. The two raters evaluated critical comments ŽCC., hostility ŽH., emotional over-involvement ŽEOI., warmth ŽW., and positive remarks ŽPR. similarly to the previous study in families of schizophrenic patients. Spearman’s correlation coefficient by ranks, value, and agreement proportion were calculated for each EE subscale as a measure of inter-rater reliability. -values were calculated by dividing the subjects into those with two or less CC and those with three or more CC, or into those with one or less CC and those with two or more CC. The subjects were divided according to: hostility of 0 or 1 or above; EOI of 2 or less or 3 or above; W of 1 or less or 2 or above; and PR of 0 or 1 or above. Since the -value was low in some subscales ŽFeinstein and Cicchetti, 1990. despite the high agreement proportion, the positive agreement proportion and the negative agreement proportion were also calculated ŽCicchetti and Feinstein, 1990.. To clarify characteristics of EE of families of patients with mood disorders, the distribution of EE subscales was compared between the families of schizophrenic patients and the subjects of this study. Data obtained in our previous study ŽMino et al., 1995b. were used as data of families of schizophrenic patients. If the ratings were different between the two raters concerning 26 of the 31 family members, they were discussed until agreement was reached. The remaining five subjects were evaluated by the authorized rater. The proportions of patients with no CC, patients with six or more CC, patients with H, and patients with EOI of 3 or more were evaluated. Moreover, to clarify characteristics of EE of families of patients with mood disorders in Japan, the results of this study were compared with the proportions of households with two or more CC or three or more CC reported in various regions of the world. If two or more members of the same household were interviewed, the greater number
223
of CC was selected for that household. Three reports from the UK ŽVaughn and Leff, 1976; Hooley et al., 1986; Hayhurst et al., 1997., a report from Canada ŽGoering et al., 1992., and a report from Egypt ŽOkasha et al., 1994. were used as references.
3. Results 3.1. Characteristics of subjects and their family members Among 27 patients, 11 Ž40.7%. were male and 16 Ž59.3%. were female. As to age: five Ž18.5%. were under 40 years old; four Ž14.8%. in their 40s; six Ž22.2%. in their 50s; and 12 Ž44.4%. in their 60s. Twenty Ž74.1%. were diagnosed as monopolar depression and seven Ž25.9%. as bipolar disorder. As to relationship to patients, of the 31 family members: 14 Ž45.2%. were husbands; 10 Ž32.2%. wives; two Ž6.5%. fathers; three Ž9.7%. mothers; and two Ž6.5%. others. As to age distribution: five Ž16.1%. were under 40 years old; five Ž16.1%. in their 40s; seven Ž22.6%. in their 50s; and 14 Ž45.2%. in their 60s or older. 3.2. Inter-rater reliability Table 1 shows Spearman’s correlation coefficients by ranks and the -values for each subscale. Spearman’s correlation coefficients by ranks were 0.7 or higher except that the coefficient was 0.48 for H. The -value was 0.46 for H and ) 0.6 for other parameters. The -value could not be calculated for EOI, because one of the raters did not rate it as 3 or higher. The agreement proportion was 0.88 or above for all items, and the negative agreement was 0.8 or above, but positive agreement was 0.5 for H and 0 for EOI. We, therefore, examined positive ratings for H and EOI. The ratings by the two raters disagreed in two subjects concerning H, and their opinions differed concerning generalization of criticism. In these cases, the number of CC were three and four. In the two cases rated differently concerning EOI, the authorized rater evaluated EOI as 3 while the other rater evaluated it as 2.
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227
224
Table 1 Inter-rater reliability of subscales of EE Ž n s 26. a
Critical comments Hostility EOI Warmth Positive remarks
r
P
0.72
- 0.001
0.48 0.70 0.79 1
0.013 - 0.001 - 0.001 - 0.001
b
0.61 0.66 c 0.46 ᎐ 0.79 1
P
A
PA
NA
0.001 - 0.001 0.019
0.88 0.88 0.92 0.92 0.92 1.0
0.67 0.73 0.50 0 0.95 1.0
0.93 0.93 0.96 0.96 0.83 1.0
- 0.001 - 0.001
a
Abbre¨ iations: EOI, emotional over-involvement; r, Spearman’s correlation coefficient by ranks; , kappa value; A, agreement; PA, positive agreement Žobserved proportion of positive agreement.; NP, negative agreement Žobserved proportion of negative agreement.. b CC were divided into the group of two or less and that of three or more. c The group of two or less and that of three or more.
3.3. Comparison with family members of schizophrenic patients Table 2 compares the distribution of EE subscales between family members of Japanese schizophrenic patients and those of mood-disorder patients. In the family members of mooddisorder patients, the percentage of those with no CC was larger, and that of those with six or more CC was smaller. On the other hand, no difference was observed in the proportion of those with H and those with EOI of 3 or above between the two groups. 3.4. International comparison of EE subscales Table 3 compares the proportions of house-
Table 2 Comparisons of EE components between families of patients with mood disorders and those with schizophrenia in Japanese sample Schizophrenia Mood disorders P Ž n s 73. Ž n s 31. value No CC 25% Six or more CC 29% Hq 12% Three or more EOI 8% U
55% 3% 10% 10%
By 2-test Žtwo-tailed., 2 value s 8.9. By Fisher’s exact test Žtwo-tailed..
UU
0.003U 0.003UU NS NS
holds with two or more CC and those with three or more CC among various countries. Since reports to date evaluated the distribution of CC in the households of mood-disorder patients rather than the distribution of EE subscales in individual family members, data concerning EE of families instead of individual family members are shown here. In this study, the proportion of households with two or more CC was 30%, and that of households with three or more CC was 19%. These values were significantly lower than the values reported from other cultural regions.
4. Discussion In training in EE rating, a correlation coefficient of 0.8 or higher has been required as a criterion of inter-rater reliability. Concerning the -value, 0.75 or higher has also been considered desirable ŽFleiss, 1981.. In this study, Spearman’s correlation coefficient was 0.7 or higher except for H, but the values were lower than those in family members of schizophrenic patients ŽMino et al., 1995a.. Also, the agreement proportion was high, but the -values for CC, H, and EOI, which are important subscales for EE ratings, were 0.46᎐0.66. These values were fair to good according to the criteria of Fleiss Ž1981. but were lower than the values in schizophrenia. Concerning the -value, ratings of the two raters were often 0 or
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227 Table 3 International comparison of distributions of critical comments in families of patients with mood disorders a CC Two or more
Three or more
This study Ž n s 27. Japan
30%
19%
Vaughn and Leff Ž1976. UK Ž n s 30.
70% Ž Ps 0.002.
Hooley et al. Ž1986. UK Ž n s 39.
87% Ž P- 0.001.
Goering et al. Ž1992. Canada Ž n s 42.
57% Ž Ps 0.025.
Okasha et al. Ž1994. Egypt Ž n s 32.
84% Ž P- 0.001.
Hayhurst et al. Ž1997. UK Ž n s 39.
64% Ž Ps 0.006.
79% Ž P- 0.001.
69% Ž Ps 0.001.
a
Statistical significance was calculated by 2-test Žtwotailed., compared to Japanese sample.
negative in 2 = 2 tables, and the tables had unbalanced marginals ŽFeinstein and Cicchetti, 1990.. When the proportions of positive agreement and negative agreement were evaluated, the proportion of positive agreement was low particularly in H and EOI while the proportion of negative agreement was satisfactory. Further discussion is needed concerning the borderline judgment of H and EOI. In comparison with EE ratings in schizophrenia, the reliability of EE ratings in family members of patients with mood disorders was questionable. The primary reason for this poor reliability was that EE ratings based on the CFI were originally designed for schizophrenia. In family members of mood-disorder patients, the distributions of all subscales were closer to 0, their ranges were small, and few positive ratings were made when 2 = 2 tables were prepared. Probably for these reasons, the -value and the proportion of positive agreement were considered to be smaller than in schizophrenia. Presently, the relations between EE of family members and the course of mood disorders are being evaluated, but interna-
225
tional evaluation of the reliability of EE rating based on CFI in family members of patients with mood disorders is also required. When EE subscales are compared between family members of schizophrenic patients and those of patients with mood disorders, the proportion of individuals with no CC was greater, and the proportion of those with six or more CC was smaller, in family members of mood disorder patients. Although there have not been many studies that compared families of schizophrenic patients and those of mood-disorder patients in the same cultural region, the number of CC was distributed near 0 in a study of family members of patients with depressive psychosis conducted in the UK ŽVaughn and Leff, 1976.. We infer that similar results could be observed in other cultural regions. In this comparison, we should consider differences in characteristics of family members. More than 75% were spouses in mood-disorder cases, while parents were dominant in cases of schizophrenia. On the other hand, EOI of 3 or above was reported to have been observed less frequently in the households of patients with mood disorders ŽLeff and Vaughn, 1985., but no difference was observed in Japan, probably because the proportion of households with EOI of 3 or above was only 8% in households of schizophrenic patients. Knowledge from cross-cultural viewpoints has been accumulated concerning the distribution of EE, its subscales, and content of CC in families of schizophrenic patients ŽMino et al., 1995b; Shimodera et al., 1998.. According to a report from Japan ŽMino et al., 1995b., the proportion of high EE was smaller, the number of CC was smaller, and the proportion of H of 1 or above or of EOI of 3 or above was smaller compared with the data in Europe and America. In this study, which approached mood disorders for the first time in Asia, the proportion of two or more CC and the proportion of three or more CC, which have been suggested to be related to relapse of mood disorders, were smaller than in reports from the UK and Egypt. This means that expressed emotion is more reserved in Asia than in other cultural regions in mood disorders as well as in schizophrenia. Japanese culture has been charac-
226
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227
terized by ‘shame’ ŽBenedict, 1967.. Expressing criticism or emotion in the interview might be considered shameful by Japanese relatives. As to H, there were two spouses who showed H at the level of CC below six. This has been observed only in India ŽWig et al., 1987b., and never been found in Europe or USA. Reports concerning mood disorders from other regions of Asia are anticipated. Finally, limitations of this study are discussed. First, there is the possibility of misclassification in comparative evaluation of EE and its subscales. In fact, the second rater was not trained formally, but trained by the first rater who was certified. Therefore, there was a possibility that the second rater had not reached the standard criterion. To avoid misclassification, ratings were determined after careful discussion between the two raters if there were disagreements in their ratings. Although EE ratings in family members of schizophrenic patients have been evaluated internationally and comparatively, no substantial evaluation has been made concerning EE in family members of patients with mood disorders. International studies in this field are required for the future. Secondly, the subjects in this study included a few patients with bipolar mood disorders. As shown in Table 3, all studies with the exception of one in Egypt employed subjects with monopolar mood disorder, and this may remain as a problem. In conclusion, the reliability of EE ratings was lower in mood disorders than in schizophrenia concerning CC, H, and EOI, which are important for EE rating. Particular attention is needed concerning positive ratings for H and EOI. In Japan, expressed emotion in family members of patients with mood disorders appears to be more suppressed than in those of patients with schizophrenia. Expressed emotion of the Japanese family members also appears to be suppressed compared with European or American counterparts.
Acknowledgements The authors extend their sincere thanks to the
patients and their families who cooperated in this study. It is also mentioned with gratitude that part of this study was sponsored by a 1996 subsidy by the Uehara Memorial Foundation and a 1998᎐2000 grant-in-aid for scientific research ŽB. from the Japan Ministry of Education, Science, Sports and Culture Žgrant No. 10470108.. References American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders, 4th ed., DSM-IV. APA, Washington, DC, pp. 339᎐345. Bebbington, P., Kuipers, L., 1994. The predictive utility of expressed emotion in schizophrenia: an aggregate analysis. Psychological Medicine 24, 707᎐718. Benedict, R., 1967. The Chrysanthemeum and The Sward. Haughton Mifflin Co., Boston. Bentsen, H., Boye, B., Munkvold, O.G., Uren, G., Lersbryggen, A.B., Oskarsson, K.H., Berg-Larsen, R., Lingjaerde, O., Malt, U.F., 1996. Inter-rater reliability of expressed emotion ratings based on the Camberwell Family Interview. Psychological Medicine 26, 821᎐828. Broadhead, W.E., Blazer, D.G., George, L.K., Tse, C.K., 1990. Depression, disability days, and days lost from work in a prospective epidemiologic survey. Journal of the American Medical Association 264, 2524᎐2528. Brown, G.W., Harris, T., 1978. Social Origin of Depression. Tavistock Publication, London. Butzlaff, R.L., Hooley, J.M., 1998. Expressed emotion and psychiatric relapse, a meta-analysis. Archives of General Psychiatry 55, 547᎐552. Cicchetti, D.V., Feinstein, A.R., 1990. High agreement but low kappa: II. Resolving the paradoxes. Journal of Clinical Epidemiology 43, 551᎐558. Cooper, B., Eastwood, R., 1992. Primary Health Care and Psychiatric Epidemiology. Routledge, London. Feinstein, A.R., Cicchetti, D.V., 1990. High agreement but low kappa: I. The problems of two paradoxes. Journal of Clinical Epidemiology 43, 543᎐549. Fleiss, J.L., 1981. Statistical Methods for Rates and Proportions. John Wiley & Sons, New York. Gelder, M., Gath, D., Mayou, R., Cowen, P., 1996. Oxford Textbook of Psychiatry, 3rd ed. Oxford University Press, Oxford. Goering, P.N., Lancee, W.J., Freeman, J.J., 1992. Marital support and recovery from depression. British Journal of Psychiatry 160, 76᎐82. Hayhurst, H., Cooper, Z., Paykel, E.S., Vearnals, S., Ramana, R., 1997. Expressed emotion and depression. British Journal of Psychiatry 171, 439᎐443. Hooley, J.M., Orley, J., Teasdale, J.D., 1986. Levels of expressed emotion and relapse in depressive patients. British Journal of Psychiatry 148, 642᎐647.
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227 Inoue, S., Tanaka, S., Shimodera, S., Mino, Y., 1997. Expressed emotion and social function. Psychiatry Research 72, 33᎐39. Leff, J., Vaughn, C., 1985. Expressed Emotion in Families. Guilford Press, New York. Mino, Y., Aoyama, H., Froom, J., 1994. Depressive disorders in Japanese primary care patients. Family Practice 11, 363᎐367. Mino, Y., Tanaka, S., Tsuda, T., Babazono, A., Inoue, S., Aoyama, H., 1995a. Training in evaluation of expressed emotion using the Japanese version of Camberwell Family Interview. Acta Psychiatrica Scandinavica 92, 183᎐186. Mino, Y., Tanaka, S., Inoue, S., Tsuda, T., Babazono, A., Aoyama, H., 1995b. Expressed emotion components in families of schizophrenic patients in Japan. International Journal of Mental Health 24 Ž2., 38᎐49. Mino, Y., Inoue, S., Tanaka, S., Tsuda, T., 1997. Expressed emotion among families and course of schizophrenia in Japan: a 2-year cohort study. Schizophrenia Research 24, 333᎐339. Mino, Y., Inoue, S., Shimodera, S., Tanaka, S., Tsuda, T., Yamamoto, E., 1998. Expressed emotion of families and negativerdepressive symptoms in schizophrenia: a cohort study in Japan. Schizophrenia Research 34, 159᎐168. Okasha, A., El Akabawi, A.S., Snyder, K.S., Wilson, A.K., Youssef, I., El Dawla, A.S., 1994. Expressed emotion, perceived criticism, and relapse in depression: a replication in an Egyptian community. American Journal of Psychiatry 151, 1001᎐1005. Orhagen, T., d’Elia, G., 1991. Expressed emotion, a Swedish
227
version of the Camberwell Family Interview. Acta Psychiatrica Scandinavica 84, 466᎐474. Shimodera, S., Inoue, S., Tanaka, S., Mino, Y., 1998. Critical comments made to schizophrenic patients by their families in Japan. Comprehensive Psychiatry 39, 85᎐90. Tanaka, S., Mino, Y., Inoue, S., 1995. Expressed emotion and schizophrenic course in Japan. British Journal of Psychiatry 167, 794᎐798. Uehara, T., Yokoyama, T., Goto, M., Ihda, S., 1996. Expressed emotion and short-term treatment outcome of outpatients with major depression. Comprehensive Psychiatry 37, 299᎐304. Vaughn, C.E., Leff, J.P., 1976. The influence of family and social factors on the course of psychiatric illness. British Journal of Psychiatry 129, 125᎐137. Wig, N.N., Menon, D.K., Bedi, H., Ghosh, A., Kuipers, L., Leff, J., Korten, A., Day, R., Sartorius, N., Ernberg, G., Jablensky, A., 1987a. Expressed emotion and schizophrenia in North India. I. Cross cultural transfer of rating of relatives’ expressed emotion. British Journal of Psychiatry 151, 156᎐160. Wig, N.N., Menon, D.K., Bedi, H., Leff, J., Kuipers, L., Ghosh, A., Day, R., Korten, A., Ernberg, G., Sartorius, N., Jablensky, A., 1987b. Expressed emotion and schizophrenia in North India. II. Distribution of expressed emotion components among relatives of schizophrenic patients in Aarhus and Chandigarh. British Journal of Psychiatry 151, 160᎐165. World Health Organization, 1992. The ICD-10 Classification of Mental and Behavioural Disorders, Clinical Description and Diagnostic Guidelines. WHO, Geneva.
Evaluation of expressed emotion ž EE/ status in mood disorders in Japan: inter-rater reliability and characteristics of EE U
Yoshio Minoa, , Shimpei Inoueb, Shinji Shimoderab, Shuichi Tanakac a
Department of Hygiene and Pre¨ enti¨ e Medicine, Okayama Uni¨ ersity Medical School, 2-5-1 Shikata-cho, Okayama 700-8558, Japan b Department of Neuropsychiatry, Kochi Medical School, Kohasu, Oko-cho, Nankoku 783-8505, Japan c Department of Psychiatry, Watarigawa Hospital, Gudo, Nakamura 787, Japan Received 20 January 1999; received in revised form 26 May 1999; accepted 12 July 1999
Abstract The reliability of expressed emotion ŽEE. ratings by the Camberwell Family Interview ŽCFI. and characteristics of EE were evaluated in families of patients with mood disorders in Japan. The subjects were 27 patients with mood disorders and 31 members of their families. The CFI was carried out with the family members. EE was rated by two raters independently, and the inter-rater reliability was evaluated according to Spearman’s correlation coefficient by ranks and the -value. The distribution of subscales of EE in these subjects was compared with that in families of patients with schizophrenia in Japan and families of patients with mood disorders abroad. Concerning critical comments ŽCC., hostility ŽH., and emotional over-involvement ŽEOI., which are important for EE rating, Spearman’s correlation coefficient and the -values were 0.4᎐0.8, and the reliability of EE ratings in mood disorders was not high. The proportion of positive agreement was particularly low in H and EOI. CCs were fewer in families of Japanese patients with mood disorders than in those with schizophrenia or families of American or European patients with mood disorders. Re-evaluation of the inter-rater reliability of EE ratings in mood disorders is needed. Expressed emotion was more suppressed in families of patients with mood disorders than in those of patients with schizophrenia. Expressed emotion was also more reserved in the Japanese subjects than in their Western counterparts. 䊚 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Depression; Bipolar disorder; Family; Camberwell Family Interview; Transcultural psychiatry
U
Corresponding author. Tel.: q81-86-235-7171; fax: q81-86-235-7178. E-mail address: [email protected] ŽY. Mino.. 0165-1781r00r$ - see front matter 䊚 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 5 - 1 7 8 1 Ž 0 0 . 0 0 1 3 2 - 3
222
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227
1. Introduction Expressed emotion ŽEE. in patients’ families has been studied concerning schizophrenia, and effects of EE on the course of schizophrenia have been confirmed in a number of cultural regions ŽLeff and Vaughn, 1985; Bebbington and Kuipers, 1994; Butzlaff and Hooley, 1998.. In Japan, which is a different cultural region compared with Western societies, relations of EE with relapse of schizophrenia, social function, and depressive symptoms have also been clarified ŽTanaka et al., 1995; Inoue et al., 1997; Mino et al., 1997, 1998.. However, the relationship between EE of families and psychiatric disorders other than schizophrenia has not been sufficiently described. The relations between EE and mood disorders including depression ŽVaughn and Leff, 1976; Hooley et al., 1986; Okasha et al., 1994; Uehara et al., 1996., in particular, must be evaluated for the following reasons. First, there is a high prevalence of mood disorders including depression. For example, the life time prevalence of major depressive disorders in the general population is 5᎐12% in males and 10᎐25% in females, and their point prevalence is 2᎐3% in males and 5᎐9% in females ŽAmerican Psychiatric Association, 1994.. The point prevalence of mood disorders is high also in the primary-care setting, and research is ongoing concerning how to screen and effectively treat them ŽCooper and Eastwood, 1992; Mino et al., 1994.. Suicide ŽGelder et al., 1996. and days lost from work ŽBroadhead et al., 1990. due to depression have also become major issues. If family intervention with mood disorders based on EE research succeeded, there would be considerable benefit not only to patients but also to society. Secondly, depression is reported to be a disease strongly influenced by the social environment ŽBrown and Harris, 1978., and effects of EE, which is a manifestation of the patient’s familial environment, cannot be ignored. The reliability of EE ratings has been studied ŽWig et al., 1987a; Orhagen and d’Elia, 1991; Bentsen et al., 1996.. Training in EE rating in Japanese has been evaluated, and a satisfactory inter-rater reliability could be attained by a training program ŽMino et al., 1995a.. However, these
studies were based on interviews with families of schizophrenic patients, and there have not been many studies in which the inter-rater reliability of EE ratings was evaluated on the basis of the Camberwell Family Interview ŽCFI. ŽLeff and Vaughn, 1985. in families of patients with mood disorders. Moreover, the distribution of EE and its subscales has been reported to differ among the cultural regions to which the patients and their families belong in studies of EE in schizophrenia ŽLeff and Vaughn, 1985; Mino et al., 1995b.. Although culture is considered to exert similar effects on EE also in families of patients with mood disorders, few studies have approached this problem. This study: Ž1. examined the inter-rater reliability of EE ratings in families of patients with mood disorders; Ž2. compared EE and its subscales between families of schizophrenic patients and those of mood-disordered patients in the same society; and Ž3. compared EE and its subscales in families of mood-disordered patients among different cultures.
2. Materials and methods The subjects were part of the patients employed for evaluation of the relationship between the course of mood disorders and EE of the families of patients. They included 27 patients with mood disorders, who were chosen from the entry order, treated at the Department of Psychiatry, Kochi Medical School and Tosa Hospital between April 1997 and April 1998, and 31 of their family members. The diagnosis was made on the basis of DSM-IV ŽAmerican Psychiatric Association, 1994. and ICD-10 ŽWorld Health Organization, 1992.. To evaluate the inter-rater reliability of EE ratings, two raters independently evaluated the data obtained by the CFI of the 26 family members of the first 22 subjects. One of the two raters had been formally trained in EE rating and had been authorized as a rater, and the other was his trainee who attained a satisfactory level of interrater reliability Ž r s 0.86᎐0.95, P- 0.001.. The
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227
training was carried out primarily using data derived from interviews with families of schizophrenic patients. Rating was performed by the same method as the CFI in families of schizophrenic patients ŽMino et al., 1995a., using tapes in which the interviews were recorded and their transcripts. The two raters evaluated critical comments ŽCC., hostility ŽH., emotional over-involvement ŽEOI., warmth ŽW., and positive remarks ŽPR. similarly to the previous study in families of schizophrenic patients. Spearman’s correlation coefficient by ranks, value, and agreement proportion were calculated for each EE subscale as a measure of inter-rater reliability. -values were calculated by dividing the subjects into those with two or less CC and those with three or more CC, or into those with one or less CC and those with two or more CC. The subjects were divided according to: hostility of 0 or 1 or above; EOI of 2 or less or 3 or above; W of 1 or less or 2 or above; and PR of 0 or 1 or above. Since the -value was low in some subscales ŽFeinstein and Cicchetti, 1990. despite the high agreement proportion, the positive agreement proportion and the negative agreement proportion were also calculated ŽCicchetti and Feinstein, 1990.. To clarify characteristics of EE of families of patients with mood disorders, the distribution of EE subscales was compared between the families of schizophrenic patients and the subjects of this study. Data obtained in our previous study ŽMino et al., 1995b. were used as data of families of schizophrenic patients. If the ratings were different between the two raters concerning 26 of the 31 family members, they were discussed until agreement was reached. The remaining five subjects were evaluated by the authorized rater. The proportions of patients with no CC, patients with six or more CC, patients with H, and patients with EOI of 3 or more were evaluated. Moreover, to clarify characteristics of EE of families of patients with mood disorders in Japan, the results of this study were compared with the proportions of households with two or more CC or three or more CC reported in various regions of the world. If two or more members of the same household were interviewed, the greater number
223
of CC was selected for that household. Three reports from the UK ŽVaughn and Leff, 1976; Hooley et al., 1986; Hayhurst et al., 1997., a report from Canada ŽGoering et al., 1992., and a report from Egypt ŽOkasha et al., 1994. were used as references.
3. Results 3.1. Characteristics of subjects and their family members Among 27 patients, 11 Ž40.7%. were male and 16 Ž59.3%. were female. As to age: five Ž18.5%. were under 40 years old; four Ž14.8%. in their 40s; six Ž22.2%. in their 50s; and 12 Ž44.4%. in their 60s. Twenty Ž74.1%. were diagnosed as monopolar depression and seven Ž25.9%. as bipolar disorder. As to relationship to patients, of the 31 family members: 14 Ž45.2%. were husbands; 10 Ž32.2%. wives; two Ž6.5%. fathers; three Ž9.7%. mothers; and two Ž6.5%. others. As to age distribution: five Ž16.1%. were under 40 years old; five Ž16.1%. in their 40s; seven Ž22.6%. in their 50s; and 14 Ž45.2%. in their 60s or older. 3.2. Inter-rater reliability Table 1 shows Spearman’s correlation coefficients by ranks and the -values for each subscale. Spearman’s correlation coefficients by ranks were 0.7 or higher except that the coefficient was 0.48 for H. The -value was 0.46 for H and ) 0.6 for other parameters. The -value could not be calculated for EOI, because one of the raters did not rate it as 3 or higher. The agreement proportion was 0.88 or above for all items, and the negative agreement was 0.8 or above, but positive agreement was 0.5 for H and 0 for EOI. We, therefore, examined positive ratings for H and EOI. The ratings by the two raters disagreed in two subjects concerning H, and their opinions differed concerning generalization of criticism. In these cases, the number of CC were three and four. In the two cases rated differently concerning EOI, the authorized rater evaluated EOI as 3 while the other rater evaluated it as 2.
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227
224
Table 1 Inter-rater reliability of subscales of EE Ž n s 26. a
Critical comments Hostility EOI Warmth Positive remarks
r
P
0.72
- 0.001
0.48 0.70 0.79 1
0.013 - 0.001 - 0.001 - 0.001
b
0.61 0.66 c 0.46 ᎐ 0.79 1
P
A
PA
NA
0.001 - 0.001 0.019
0.88 0.88 0.92 0.92 0.92 1.0
0.67 0.73 0.50 0 0.95 1.0
0.93 0.93 0.96 0.96 0.83 1.0
- 0.001 - 0.001
a
Abbre¨ iations: EOI, emotional over-involvement; r, Spearman’s correlation coefficient by ranks; , kappa value; A, agreement; PA, positive agreement Žobserved proportion of positive agreement.; NP, negative agreement Žobserved proportion of negative agreement.. b CC were divided into the group of two or less and that of three or more. c The group of two or less and that of three or more.
3.3. Comparison with family members of schizophrenic patients Table 2 compares the distribution of EE subscales between family members of Japanese schizophrenic patients and those of mood-disorder patients. In the family members of mooddisorder patients, the percentage of those with no CC was larger, and that of those with six or more CC was smaller. On the other hand, no difference was observed in the proportion of those with H and those with EOI of 3 or above between the two groups. 3.4. International comparison of EE subscales Table 3 compares the proportions of house-
Table 2 Comparisons of EE components between families of patients with mood disorders and those with schizophrenia in Japanese sample Schizophrenia Mood disorders P Ž n s 73. Ž n s 31. value No CC 25% Six or more CC 29% Hq 12% Three or more EOI 8% U
55% 3% 10% 10%
By 2-test Žtwo-tailed., 2 value s 8.9. By Fisher’s exact test Žtwo-tailed..
UU
0.003U 0.003UU NS NS
holds with two or more CC and those with three or more CC among various countries. Since reports to date evaluated the distribution of CC in the households of mood-disorder patients rather than the distribution of EE subscales in individual family members, data concerning EE of families instead of individual family members are shown here. In this study, the proportion of households with two or more CC was 30%, and that of households with three or more CC was 19%. These values were significantly lower than the values reported from other cultural regions.
4. Discussion In training in EE rating, a correlation coefficient of 0.8 or higher has been required as a criterion of inter-rater reliability. Concerning the -value, 0.75 or higher has also been considered desirable ŽFleiss, 1981.. In this study, Spearman’s correlation coefficient was 0.7 or higher except for H, but the values were lower than those in family members of schizophrenic patients ŽMino et al., 1995a.. Also, the agreement proportion was high, but the -values for CC, H, and EOI, which are important subscales for EE ratings, were 0.46᎐0.66. These values were fair to good according to the criteria of Fleiss Ž1981. but were lower than the values in schizophrenia. Concerning the -value, ratings of the two raters were often 0 or
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227 Table 3 International comparison of distributions of critical comments in families of patients with mood disorders a CC Two or more
Three or more
This study Ž n s 27. Japan
30%
19%
Vaughn and Leff Ž1976. UK Ž n s 30.
70% Ž Ps 0.002.
Hooley et al. Ž1986. UK Ž n s 39.
87% Ž P- 0.001.
Goering et al. Ž1992. Canada Ž n s 42.
57% Ž Ps 0.025.
Okasha et al. Ž1994. Egypt Ž n s 32.
84% Ž P- 0.001.
Hayhurst et al. Ž1997. UK Ž n s 39.
64% Ž Ps 0.006.
79% Ž P- 0.001.
69% Ž Ps 0.001.
a
Statistical significance was calculated by 2-test Žtwotailed., compared to Japanese sample.
negative in 2 = 2 tables, and the tables had unbalanced marginals ŽFeinstein and Cicchetti, 1990.. When the proportions of positive agreement and negative agreement were evaluated, the proportion of positive agreement was low particularly in H and EOI while the proportion of negative agreement was satisfactory. Further discussion is needed concerning the borderline judgment of H and EOI. In comparison with EE ratings in schizophrenia, the reliability of EE ratings in family members of patients with mood disorders was questionable. The primary reason for this poor reliability was that EE ratings based on the CFI were originally designed for schizophrenia. In family members of mood-disorder patients, the distributions of all subscales were closer to 0, their ranges were small, and few positive ratings were made when 2 = 2 tables were prepared. Probably for these reasons, the -value and the proportion of positive agreement were considered to be smaller than in schizophrenia. Presently, the relations between EE of family members and the course of mood disorders are being evaluated, but interna-
225
tional evaluation of the reliability of EE rating based on CFI in family members of patients with mood disorders is also required. When EE subscales are compared between family members of schizophrenic patients and those of patients with mood disorders, the proportion of individuals with no CC was greater, and the proportion of those with six or more CC was smaller, in family members of mood disorder patients. Although there have not been many studies that compared families of schizophrenic patients and those of mood-disorder patients in the same cultural region, the number of CC was distributed near 0 in a study of family members of patients with depressive psychosis conducted in the UK ŽVaughn and Leff, 1976.. We infer that similar results could be observed in other cultural regions. In this comparison, we should consider differences in characteristics of family members. More than 75% were spouses in mood-disorder cases, while parents were dominant in cases of schizophrenia. On the other hand, EOI of 3 or above was reported to have been observed less frequently in the households of patients with mood disorders ŽLeff and Vaughn, 1985., but no difference was observed in Japan, probably because the proportion of households with EOI of 3 or above was only 8% in households of schizophrenic patients. Knowledge from cross-cultural viewpoints has been accumulated concerning the distribution of EE, its subscales, and content of CC in families of schizophrenic patients ŽMino et al., 1995b; Shimodera et al., 1998.. According to a report from Japan ŽMino et al., 1995b., the proportion of high EE was smaller, the number of CC was smaller, and the proportion of H of 1 or above or of EOI of 3 or above was smaller compared with the data in Europe and America. In this study, which approached mood disorders for the first time in Asia, the proportion of two or more CC and the proportion of three or more CC, which have been suggested to be related to relapse of mood disorders, were smaller than in reports from the UK and Egypt. This means that expressed emotion is more reserved in Asia than in other cultural regions in mood disorders as well as in schizophrenia. Japanese culture has been charac-
226
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227
terized by ‘shame’ ŽBenedict, 1967.. Expressing criticism or emotion in the interview might be considered shameful by Japanese relatives. As to H, there were two spouses who showed H at the level of CC below six. This has been observed only in India ŽWig et al., 1987b., and never been found in Europe or USA. Reports concerning mood disorders from other regions of Asia are anticipated. Finally, limitations of this study are discussed. First, there is the possibility of misclassification in comparative evaluation of EE and its subscales. In fact, the second rater was not trained formally, but trained by the first rater who was certified. Therefore, there was a possibility that the second rater had not reached the standard criterion. To avoid misclassification, ratings were determined after careful discussion between the two raters if there were disagreements in their ratings. Although EE ratings in family members of schizophrenic patients have been evaluated internationally and comparatively, no substantial evaluation has been made concerning EE in family members of patients with mood disorders. International studies in this field are required for the future. Secondly, the subjects in this study included a few patients with bipolar mood disorders. As shown in Table 3, all studies with the exception of one in Egypt employed subjects with monopolar mood disorder, and this may remain as a problem. In conclusion, the reliability of EE ratings was lower in mood disorders than in schizophrenia concerning CC, H, and EOI, which are important for EE rating. Particular attention is needed concerning positive ratings for H and EOI. In Japan, expressed emotion in family members of patients with mood disorders appears to be more suppressed than in those of patients with schizophrenia. Expressed emotion of the Japanese family members also appears to be suppressed compared with European or American counterparts.
Acknowledgements The authors extend their sincere thanks to the
patients and their families who cooperated in this study. It is also mentioned with gratitude that part of this study was sponsored by a 1996 subsidy by the Uehara Memorial Foundation and a 1998᎐2000 grant-in-aid for scientific research ŽB. from the Japan Ministry of Education, Science, Sports and Culture Žgrant No. 10470108.. References American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders, 4th ed., DSM-IV. APA, Washington, DC, pp. 339᎐345. Bebbington, P., Kuipers, L., 1994. The predictive utility of expressed emotion in schizophrenia: an aggregate analysis. Psychological Medicine 24, 707᎐718. Benedict, R., 1967. The Chrysanthemeum and The Sward. Haughton Mifflin Co., Boston. Bentsen, H., Boye, B., Munkvold, O.G., Uren, G., Lersbryggen, A.B., Oskarsson, K.H., Berg-Larsen, R., Lingjaerde, O., Malt, U.F., 1996. Inter-rater reliability of expressed emotion ratings based on the Camberwell Family Interview. Psychological Medicine 26, 821᎐828. Broadhead, W.E., Blazer, D.G., George, L.K., Tse, C.K., 1990. Depression, disability days, and days lost from work in a prospective epidemiologic survey. Journal of the American Medical Association 264, 2524᎐2528. Brown, G.W., Harris, T., 1978. Social Origin of Depression. Tavistock Publication, London. Butzlaff, R.L., Hooley, J.M., 1998. Expressed emotion and psychiatric relapse, a meta-analysis. Archives of General Psychiatry 55, 547᎐552. Cicchetti, D.V., Feinstein, A.R., 1990. High agreement but low kappa: II. Resolving the paradoxes. Journal of Clinical Epidemiology 43, 551᎐558. Cooper, B., Eastwood, R., 1992. Primary Health Care and Psychiatric Epidemiology. Routledge, London. Feinstein, A.R., Cicchetti, D.V., 1990. High agreement but low kappa: I. The problems of two paradoxes. Journal of Clinical Epidemiology 43, 543᎐549. Fleiss, J.L., 1981. Statistical Methods for Rates and Proportions. John Wiley & Sons, New York. Gelder, M., Gath, D., Mayou, R., Cowen, P., 1996. Oxford Textbook of Psychiatry, 3rd ed. Oxford University Press, Oxford. Goering, P.N., Lancee, W.J., Freeman, J.J., 1992. Marital support and recovery from depression. British Journal of Psychiatry 160, 76᎐82. Hayhurst, H., Cooper, Z., Paykel, E.S., Vearnals, S., Ramana, R., 1997. Expressed emotion and depression. British Journal of Psychiatry 171, 439᎐443. Hooley, J.M., Orley, J., Teasdale, J.D., 1986. Levels of expressed emotion and relapse in depressive patients. British Journal of Psychiatry 148, 642᎐647.
Y. Mino et al. r Psychiatry Research 94 (2000) 221᎐227 Inoue, S., Tanaka, S., Shimodera, S., Mino, Y., 1997. Expressed emotion and social function. Psychiatry Research 72, 33᎐39. Leff, J., Vaughn, C., 1985. Expressed Emotion in Families. Guilford Press, New York. Mino, Y., Aoyama, H., Froom, J., 1994. Depressive disorders in Japanese primary care patients. Family Practice 11, 363᎐367. Mino, Y., Tanaka, S., Tsuda, T., Babazono, A., Inoue, S., Aoyama, H., 1995a. Training in evaluation of expressed emotion using the Japanese version of Camberwell Family Interview. Acta Psychiatrica Scandinavica 92, 183᎐186. Mino, Y., Tanaka, S., Inoue, S., Tsuda, T., Babazono, A., Aoyama, H., 1995b. Expressed emotion components in families of schizophrenic patients in Japan. International Journal of Mental Health 24 Ž2., 38᎐49. Mino, Y., Inoue, S., Tanaka, S., Tsuda, T., 1997. Expressed emotion among families and course of schizophrenia in Japan: a 2-year cohort study. Schizophrenia Research 24, 333᎐339. Mino, Y., Inoue, S., Shimodera, S., Tanaka, S., Tsuda, T., Yamamoto, E., 1998. Expressed emotion of families and negativerdepressive symptoms in schizophrenia: a cohort study in Japan. Schizophrenia Research 34, 159᎐168. Okasha, A., El Akabawi, A.S., Snyder, K.S., Wilson, A.K., Youssef, I., El Dawla, A.S., 1994. Expressed emotion, perceived criticism, and relapse in depression: a replication in an Egyptian community. American Journal of Psychiatry 151, 1001᎐1005. Orhagen, T., d’Elia, G., 1991. Expressed emotion, a Swedish
227
version of the Camberwell Family Interview. Acta Psychiatrica Scandinavica 84, 466᎐474. Shimodera, S., Inoue, S., Tanaka, S., Mino, Y., 1998. Critical comments made to schizophrenic patients by their families in Japan. Comprehensive Psychiatry 39, 85᎐90. Tanaka, S., Mino, Y., Inoue, S., 1995. Expressed emotion and schizophrenic course in Japan. British Journal of Psychiatry 167, 794᎐798. Uehara, T., Yokoyama, T., Goto, M., Ihda, S., 1996. Expressed emotion and short-term treatment outcome of outpatients with major depression. Comprehensive Psychiatry 37, 299᎐304. Vaughn, C.E., Leff, J.P., 1976. The influence of family and social factors on the course of psychiatric illness. British Journal of Psychiatry 129, 125᎐137. Wig, N.N., Menon, D.K., Bedi, H., Ghosh, A., Kuipers, L., Leff, J., Korten, A., Day, R., Sartorius, N., Ernberg, G., Jablensky, A., 1987a. Expressed emotion and schizophrenia in North India. I. Cross cultural transfer of rating of relatives’ expressed emotion. British Journal of Psychiatry 151, 156᎐160. Wig, N.N., Menon, D.K., Bedi, H., Leff, J., Kuipers, L., Ghosh, A., Day, R., Korten, A., Ernberg, G., Sartorius, N., Jablensky, A., 1987b. Expressed emotion and schizophrenia in North India. II. Distribution of expressed emotion components among relatives of schizophrenic patients in Aarhus and Chandigarh. British Journal of Psychiatry 151, 160᎐165. World Health Organization, 1992. The ICD-10 Classification of Mental and Behavioural Disorders, Clinical Description and Diagnostic Guidelines. WHO, Geneva.