The National Service Framework for Mental Health emphasises inequities in care across ethnic groups in the UK. This, coupled with the Royal College of Psychiatrists' resolve to address concerns about institutionalised racism, promises a fundamental change in existing clinical practice. Practitioners are undecided about the best way to achieve such change. It is over a decade since socio-anthropological methods were recommended as one way of improving the cultural capabilities of practitioners (Reference KleinmanKleinman, 1988). People explain their distress in a multitude of ways, often blaming social circumstances, relationship problems, witchcraft or sorcery, or a broken taboo (Reference WeissWeiss, 1997). Kleinman recommended that a patient's explanatory models of illness should be elicited using a mini-ethnographic approach that explored their concerns: ‘Why me?’ ‘Why now?’ ‘What is wrong?’ ‘How long will it last?’ ‘How serious is it?’ ‘Who can intervene or treat the condition?’ The clinician can gather a better understanding of the subjective experience of illness, and so promote collaboration and improve clinical outcomes and patient satisfaction. Lloyd et al (Reference Lloyd, Jacob and Patel1998) and Weiss (Reference Weiss1997) have developed instruments to elicit explanatory models. The domains within these are summarised in Table 1, alongside those found in the related Illness Perception Questionnaire, which focuses on physical illness.

Table 1 Explanatory models and illness perception questionnaires

IPQ (Reference Weinman, Petrie and Moss-MorrisWeinman et al, 1996)	SEMI (Reference Lloyd, Jacob and PatelLloyd et al, 1998)	EMIC (Reference WeissWeiss, 1997)
Identity	Naming the condition	Patterns of distress
Causes	What causes it?	Perceived causes
	Is it an illness?
Consequences		Disease-specific queries
Controllability	Who do you see about it?	Seeking help and treatment
	What you can do about it?
	What your doctor can do about it
Time line		General illness beliefs

EXPLANATORY MODELS IN PRACTICE

Kleinman's original approach involved asking questions through an exploratory process of qualitative enquiry. This leads to complex and multi-layered responses which carry with them information about social rituals, symbols in communication, forms of knowledge and illness narratives. The patients' rich view of the world and of their illness within that world gives rise to a better understanding of their illness, including its meaning to them and their expected recovery process. Empirical evidence suggests that patients are most satisfied where their psychiatrist shares their model of understanding distress and treatment (Reference Callan and LittlewoodCallan & Littlewood, 1998). The proposed mechanism is for patient and practitioner to share information, a feature that distinguishes traditional healing systems from Western biomedicine, which thrives on esoteric knowledge being held by the professional. Despite the appeal of investigating explanatory models, such findings have not been influential on routine psychiatric practice, albeit there is now a greater emphasis on consumer views and satisfaction. Mental health professionals are not commonly equipped with a social science background, although social psychiatry is an established clinical and research speciality. It may be that the explanatory paradigm of psychiatrists, which has a predominantly biomedical orientation, explains the hesitation with which social science methods are absorbed into routine psychiatric practice. The ‘process’ of enquiry is crucial to social scientists and should be of prime important in clinical psychiatric practice. Yet rarely does clinical practice systematically apply the process rules to elicit explanatory models and to maximise collaboration and communication between patient and professional.

The difficulty may also lie in the task of the clinician. When explanatory models are investigated, the socio-anthropological framework of participant observation and open-ended conversation embraces the authentic view of the patient's world. This is lost if the questions focus on making a diagnosis and introducing a treatment. Aspects of the explanatory model that address these are given higher priority, leading to the neglect of the patient's total experience of the illness. A study of first-time presenters to mental health services showed that explanatory models did not consist of a coherent set of beliefs, but a variety of explanations that are either held simultaneously or taken up and dismissed rapidly (Reference Williams and HealyWilliams & Healy, 2001). Faced with such transient beliefs, it is difficult to distil a single set of causal explanations that might relate to behaviour, diagnosis or adherence to medication treatment. This may be the difficulty facing clinicians. Indeed, Williams & Healy recommend the term ‘explanatory map’ rather than ‘model’, as this reflects the diversity and complexity found within systems of health beliefs.

Illness is not best described in diagnostic terms (Reference EisenbergEisenberg, 1981). The transition from illness experience to disorder is determined by social decision points rather than biomedically determined levels of disorder. Can psychiatrists consider earlier stages in the experience of their patients when they are concerned with making a diagnosis that fits and determining the treatment for a diagnosed condition? An approach is to learn about indigenous systems of healing and explanatory models which are common to specific cultural groups so that an understanding of distress which is closer to the patient's experience can be achieved before attempts are made to map these to specific biomedical diagnostic categories and related care pathways (Reference PatelPatel, 1995). Any treatments can be based on a shared understanding, taking account of the best clinical evidence, while not neglecting or contravening beliefs that help people cope with misfortunes of all kinds.

RESEARCH

The research schedules in Table 1 are designed to elicit explanatory model data. The Explanatory Model Interview Catalogue (EMIC) and the Short Explanatory Model Interview (SEMI) can be used flexibly to bridge the gap between qualitative and quantitative methods of gathering health belief data (Reference WeissWeiss, 1997; Reference Lloyd, Jacob and PatelLloyd et al, 1998). The Illness Perception Questionnaire (IPQ) includes a fixed ranged of predetermined causal explanations from which patients can identify the one closest to their own views (Reference Weinman, Petrie and Moss-MorrisWeinman et al, 1996). This method assumes that the range of beliefs that are of interest are largely known, and that the constructs presented in the questions are familiar to people from different cultural and linguistic groups. The individual representation of illness sought by the IPQ contrasts with the aims of ‘explanatory model’ instruments. These seek to tap collective as well as individual cultural beliefs that include whole systems of knowledge and explanation about the ways in which the world works: explanation for misfortune, interpersonal conflict, kinship, communication, distress and the location of illness in this greater system of knowledge. These two different emphases explain the contrasting methods of data collection, the former using an instrument with closed questions and the latter using open-minded questions supported by participant observation and mini-ethnography. The EMIC, to its credit, has process rules and a system of presenting data, but it is a long interview and therefore expensive to use and unwieldy in large-scale surveys. The strength of the SEMI, on the other hand, is that it can be used in a semi-structured way to identify causal and other health beliefs, which can then be categorised for use in large-scale survey work. It allows discussion of the patient's problems, as well as exploring different ways of explaining distress by using vignette material. Although it is shorter, a great deal of qualitative information may be lost as — unlike the EMIC — there is no agreed system of managing and analysing the data. Including the explanatory model instruments in largescale surveys can also interfere with the qualitative ‘process’ of enquiry. A recent survey among Somali refugees in south London included quantitative measures of psychiatric symptoms and used vignettes from the SEMI to elicit explanatory models. Somali people found the switch to a different interview style difficult to negotiate, finding that they had little to say as they did not know the ‘answer’. Whichever instrument is used there is variation in data handling and analysis. The IPQ is the easiest to use because analysis simply involves entering the predetermined categories into univariate or multivariate analyses. The SEMI has no clear data-analytic steps or structure, other than a classification system into which responses can mapped. The absence of interview process rules can lead to researchers using it without understanding that it is not a technology but a framework within which explanatory models can be explored. Further dilemmas arise from the tension between qualitative and quantitative research methods (Reference Buston, Parry-Jones and LivingstonBuston et al, 1998). Qualitative methods might use a single interviewer, who derives health—belief conceptual categories by content analysis, grounded theory and/or ethnography. This is markedly different to the collection of data by a number of ‘trained’ interviewers in large surveys, where conceptual evolution during the survey is not emphasised. Finally, the process of deriving quantitative categories of health beliefs from verbatim transcripts is rarely described fully. Variation in data handling between studies may explain the heterogeneity of findings and may lead to a number of possible biases in the data analysis. Any research discipline considers variation of its own quality criteria and rules for analysis as a weakness. Yet flexibility and the desirable promotion of complementary quantitative and qualitative methods means that the divergent methodologies will be juxtaposed in analyses and potentially be subjected to analysis rules which are not suited to the data form or collection methods (Reference Canino, Lewis-Fernandez and BravoCanino et al, 1997).

CONCLUSION

Clinically relevant research is crucial. The impact of specific explanatory models on satisfaction with consultations, collaborative working, adherence to treatments, and clinical outcomes has not been systematically assessed. The methodological issues facing researchers need to be developed before the weight of research evidence is sufficiently compelling for clinicians routinely to explore their patients' explanatory models of distress. None the less, in the face of culturally complex clinical presentations, exploration of the patient's and clinician's explanatory model is valuable in developing culturally capable psychiatric practice.

DECLARATION OF INTEREST

None.