Since the introduction of the concept of expressed emotion (EE) (Brown et al, Reference Brown, Monck and Carstairs1962, Reference Brown, Birley and Wing1972), that is emotion expressed by close relatives towards a family member with schizophrenia, there have been several studies focusing on its influence on the course of schizophrenia and other disorders. Although most studies have supported the hypothesis that high EE predicts relapse of schizophrenia, a few good quality studies have failed to do so (e.g. MacMillan et al, Reference MacMillan, Gold and Crow1986, Reference MacMillan, Crow and Johnson1987). A recent investigation using meta-analysis of data from 27 EE studies reported a significant association between high EE and schizophrenia relapse (Reference Butzlaff and HooleyButzlaff & Hooley, 1998). However, the studies included were largely from Western countries, all of them were cross-sectional, and problems regarding the cross-cultural validity of assessments of EE were not discussed.

A number of relevant issues have been raised over the years, including the role of EE as a trait (that exists among relatives of patients before the onset of the disorder) or state (that occurs among relatives in reaction to the onset of the disorder); the frequency of patients' contact with their relatives before a relapse; the timing of EE assessment (before or during a relapse); the stability of EE levels over time; differential EE levels among relatives living with the patient; potential confounding/inter-acting effects on EE level and the course of the disorder of socio-demographic factors, social support, life events, medication and psychosocial treatment; cross-cultural validity of the concept and assessment of EE (Reference Jenkins and KarnoJenkins & Karno, 1992; Reference KavanaghKavanagh, 1992). The cross-cultural validity of EE has been examined in a few non-Western cultures, with controversial findings. Interestingly, a recent study in China, although it failed to find a significant association between high EE and relapse of schizophrenia (Reference Phillips and XiongPhillips & Xiong, 1995), demonstrated a significant contribution of high EE to the level of stigma (Reference Phillips, Pearson and LiPhillips et al, 2002, this issue).

IS THE STUDY OF EE AN ETIC PHENOMENON?

The original concept of EE includes a set of positive and negative emotions, among which three major negative components — criticism, hostility and emotional over-involvement — have become the focus of research in both Western and non-Western countries. It has been pointed out, however, that what counts as criticism, hostility and emotional overinvolvement is a matter of cultural definition (Reference Jenkins and KarnoJenkins & Karno, 1992). Culture determines whether any type of EE is a prominent part of the emotional response among family members, the specific content/target of various EE components, and factors influencing their levels. It follows that the study of EE should have both etic and emic characteristics. For example, the frequency and intensity of individual types of EE varies widely across cultures, with the proportion of high EE among relatives of individuals with schizophrenia defined by the Western operational criteria ranging from 8% in rural India (Reference Wig, Menon and BediWig et al, 1987b ) to 67% in urban USA (Reference Vaughn, Snyder and JonesVaughn et al, 1984).

EE AS ILLNESS BEHAVIOUR ACROSS CULTURES

The finding that a very low proportion (4%) of the relatives of people with schizophrenia in the Indian study was rated high in emotional overinvolvement was attributed to issues of interrater reliability resulting from training problems (Reference Wig, Menon and BediWig et al, 1987a ). A corresponding figure of 17.5% was reported in China, which fell to 7.0% when the cut-off point of measurement of emotional overinvolvement was raised from 3 to 4 (Reference Phillips and XiongPhillips & Xiong, 1995). The authors commented that the Western criteria for high emotional overinvolvement were inappropriate in Chinese culture, where most unmarried persons with schizophrenia stay with their parents. This may bring about much more contact between patients and their parents, leading to greater exposure to EE in Chinese and other non-Western societies than in their Western counterparts. The implications of such differential exposure to EE may deserve further enquiry.

Previous work in psychological anthropology has observed that Chinese families are parent-child dominated, rather than husband-wife dominated as observed in Western societies (Reference HsuHsu, 1973). Unlike the Western family, in which the autonomy and independence of children are stressed, in the Chinese family children's obedience and acceptance of parents' criticism, and interdependence between generations, are encouraged. The sick role played by a Chinese person with schizophrenia and the family's response has certain emic characteristics (Reference Lin, Lin, Kleinman and LinLin & Lin, 1980). It is therefore likely that the norms of various components of EE among Chinese relatives of patients with schizophrenia, and perhaps also among relatives in other cultures dominated by extended families (Reference El-IslamEl-Islam, 1982), are considerably different from those among their Western counterparts. The evidence does indicate that the distribution and mean scores of criticism, hostility and emotional involvement are significantly different across cultures (Reference Leff and VaughnLeff & Vaughn, 1985). Possibly other components of EE, in addition to criticism, hostility and emotional overinvolvement, merit examination and comparison between different cultures.

MEASUREMENT OF EE ACROSS CULTURES

The measurement of EE has commonly applied the 90-min Camberwell Family Interview (CFI) (Reference Brown and RutterBrown & Rutter, 1966) for individual relatives, and the degree of EE has been rated according to operational criteria (Reference Vaughn and LeffVaughn & Leff, 1976).

Since the evidence has indicated a cross-cultural difference regarding the level of tolerance among family members towards the person with schizophrenia, it is likely that patients' perception of and coping with different levels of EE differ cross-culturally. Therefore, EE assessment has to be re-calibrated in different cultures. Furthermore, raters assessing EE in different cultures might be influenced by their own non-Western cultural backgrounds. These issues require cross-cultural validity and interrater reliability studies, in which researchers from different cultures together might observe, using CFI interviews, and rate EE of relatives from different cultures. This could be followed by a thorough discussion of any differences in content, meaning, expression and cultural norms of any EE observed by any rater. In so doing, emic bias in the interviewer or the interviewee could be optimally resolved, including the different expressiveness of patients and family members across cultures. For example, relatives in some cultures might mask their negative attitude and emotional responses in front of interviewers (Reference Phillips and XiongPhillips & Xiong, 1995). Such a strategy, which requires adequate financial support and staffing, has recently been applied in the assessment of psychopathology using the Schedules for Clinical Assessment in Neuropsychiatry (SCAN), with considerable progress (Reference Cheng, Tien and ChangCheng et al, 2001). Similarly, the possible influence on EE assessment of differences in the frequency of contact between patients and relatives, as well as types of relatives, should be taken into account.

CONCLUSIONS

Research in EE and mental disorder could be more useful if studies of the cross-cultural validity, the reliability and the general methodological issues of EE assessment were carried out longitudinally across cultures.