Effect of a structured educational intervention on explanatory models of relatives of patients with schizophrenia: Randomised controlled  trial

We examined the effect of a structured educational programmeon explanatory models of illness among the relatives of peoplewith schizophrenia, in a randomised controlled trial. Participantswere assessed at baseline (n=100) and after 2 weeks (n=75)using a vignette from the Short Explanatory Model Interview.There was a reduction in non-biomedical causal explanatorymodels at follow-up among those who had completed the structurededucational programme compared with the control group. Therewas no significant difference in non-biomedical treatment explanatorymodels between the two groups.

INTRODUCTION

TOP

INTRODUCTION

By eliciting the explanatory models of patients and their relativesin routine clinical practice, mental health professionals canbetter understand the subjective experience of illness (Kleinman, 1980).The past few years have witnessed an increase in theliterature on beliefs about causes of schizophrenia and itstreatment (Angermeyer & Matschinger, 1996; Banerjee & Roy, 1998;Kulhara et al, 2000), but few studies have evaluatedthe efficacy of interventions to change explanatory models.We used a randomised controlled design to examine the effectof a structured educational programme on explanatory modelsof illness among the relatives of people with schizophrenia.

METHOD

TOP

METHOD

Out-patients who attended the Department of Psychiatry, ChristianMedical College, Vellore, India for the first time and whosatisfied ICD–10 (World Health Organization, 1993) diagnosticcriteria for a research diagnosis of schizophrenia were invitedto participate in the study with their relatives (Fig. 1).Patients continued to be managed by their psychiatrists. Informedconsent was obtained from the relatives, who were then randomisedin blocks of four, using a computer program, to a group thatwould receive the structured educational intervention or toa control group. Progress through the trial is shown in Fig. 1.

View larger version (11K):
[in this window]
[in a new window]
[as a PowerPoint slide]
Fig. 1 CONSORT diagram.

The Short Explanatory Model Interview (SEMI; Lloyd et al, 1998) formed the basis of the evaluation. A vignette describing atypical patient with chronic psychosis was presented (Joel et al, 2003). This was followed by open-ended questions toelicit the relatives’ beliefs about the perceived causesand consequences of the condition, and help-seeking behaviour.A verbatim record of the responses was made, and these werelater grouped into categories using the procedure recommendedby the SEMI. The Tamil version of the instrument was employed(Joel et al, 2003). Basic demographic and clinical informationwas also recorded. The assessments were conducted on day 1(baseline) and after 2 weeks of a structured educational programmein the intervention group. The research nurse who performedthe evaluations was masked to the intervention status of theparticipants.

We developed a structured educational programme which discussedthe different explanatory causal and treatment models prevalentin the region, and we also presented the biomedical perspectivewithout dismissing or directly challenging local beliefs. Thisintervention had three components: exploring explanatory models;psychoeducation, aimed at teaching the relatives about the illness,symptoms, treatment and prognosis; and strategies to reducethe risk of relapse. The education package covered the followingtopics in two sessions: symptoms, beliefs about causation,psychosocial influences, prevalence, biomedical model, diagnosis,treatments (including medication and adherence), and copingstrategies for families.

Sample size was calculated assuming that half of those who receive education and one-fifth of those who do not would considerthe illness to have a medical cause by the end of the trial.For a power of 80% and 95% confidence the minimum sample requiredwas 72 (36 in each arm). To compensate for possible loss tofollow-up we recruited 100 participants.

RESULTS

TOP

RESULTS

A total of 100 patients and 100 relatives (one first-degreerelative per patient) were contacted, all of whom agreed totake part in the study. The majority of the participants weremale (56%), married (91%) and literate (80%), with a mean ageof 45.3 years (s.d.=15.7). Most of the participants (60%) regardedthe condition of the patient described in the vignette as a disease. Nearly all of the relatives who were interviewed (94%)felt that help should be sought from a doctor or hospital.However, a significant proportion of the relatives also attributedthe patient’s condition to previous deeds (43%) or topunishment by God (43%), and felt that visiting temples and places of worship could solve the problem (33%). A minorityof participants attributed the patient’s condition toblack magic (33%) or evil spirits (10%), and felt that helpshould be sought from a traditional healer (5%) or a shaman(9%). The majority of participants held more than one explanatorymodel of illness, and many believed that there was at leastone non-biomedical explanation for the patient’s psychosis.The differences between the baseline socio-demographic variablesand the explanatory models of the two groups were not statisticallysignificant.

The 25 participants lost to follow-up (see Fig. 1) did notdiffer significantly from those assessed at 2 weeks with regardto socio-demographic variables, explanatory models at baseline,or treatment arm.

Intent-to-treat analysis of all participants (with the lastobservation carried forward for those who were lost to follow-up)showed that the intervention group had a statistically significantreduction in the total number of non-biomedical causal modelsof psychosis compared with the group of relatives who did notreceive additional education (mean=0.88 (s.d.=0.96) v. 1.32(1.15); P=0.08). The two groups also showed a significant differencewith regard to change in non-medical causal models from baselineafter adjusting for age, gender and literacy (mean=70.58 (s.d.=1.21) v. 0.14 (1.16); P=0.003). There were significant differences between the two groups at follow-up in the number attributingthe condition to black magic (8 v. 16; P=0.08) and the numberbelieving that visiting a place of worship would effect a cure(7 v. 14; P=0.04). However, there were no differences betweenthe two groups with regard to non-medical treatment modelsof illness (see data supplement to online version of this paper).Similar results were obtained when the data for participantswho completed the trial were analysed.

DISCUSSION

TOP

DISCUSSION

This study evaluated an educational intervention for psychosisand assessed its effect in the immediate follow-up period.Limitations included the drop-out rate of 25% and the lackof long-term assessment. However, the participants who werelost to follow-up did not differ significantly in terms ofbaseline variables from those who remained in the trial. Also,the results of an intent-to-treat analysis and analysis ofthose who completed the trial were similar.

The baseline data suggest that the relatives of patients withpsychosis have multiple, diverse and contradictory explanatorymodels of illness. Participants held simultaneous beliefs innaturalistic explanations (e.g. disease) and personalisticexplanations (e.g. supernatural causation, sin and punishment,karma). They also suggested that help could be sought from arange of different sources (e.g. doctor, temple or place ofworship, traditional healer). Other studies of explanatorymodels of psychosis have reported similar findings (Joel etal, 2003), and the issues surrounding the simultaneous holdingof multiple beliefs have been discussed in the literature (Saravananet al, 2004, 2005).

The educational intervention programme used in this study discussedthe local explanations for psychosis, and presented the biomedicalexplanatory model as an alternative. The indigenous beliefsof the participants were not challenged. The programme didnot claim exclusivity or superiority of biomedical beliefs,but discussed issues relating to symptoms, disease models, medicationand regular treatment. Although such issues are often raisedin routine clinical practice (and the control group may alsohave received such information), psychoeducation does not followa structured format, and psychiatrists tend to dismiss localexplanations and to favour biomedical concepts.

The relatives who received the educational intervention showedsome change in their explanatory models in the immediate follow-upperiod. However, many of the indigenous explanatory modelspersisted, especially those related to treatment. The resultsof this study suggest that although some explanatory modelscan be changed, others may be more resistant to modification.In the developing world, people with mental disorders oftenvisit places of worship, traditional healers and psychiatrichospitals in search of both relief from symptoms and cure (Jacob, 1999).However, it is acknowledged that holding non-biomedical beliefs about psychosis can delay the recognition of disease,prevent early institution of treatment with medication, andinterfere with adherence to treatment, resulting in a pooroutcome. Health education packages should discuss the advantagesof medication, but should not dismiss alternative explanationsof illness, as these may also help to restore mental health. Further research is needed to identify the components of theideal health education package and maximise its effectivenessin changing explanatory models and thereby preventing relapse.

View this table:
[in this window]
[in a new window]

Data supplement 1 Comparison of explanatory models of psychosis after 2 weeks: intent-to-treat analysis¹

REFERENCES

TOP