Insight, psychopathology, explanatory models and outcome of schizophrenia in India: a prospective 5-year cohort study

This study was carried out in the Department of Psychiatry, Christian Medical College, Vellore, Tamil Nadu, south India. The total area of Vellore district is 4314.29 km². The 122-bed hospital provides short-term care psychiatric care for the town, the district and a much wider rural area beyond. The department treats a variety of mental and behavioural disorders with a multi-disciplinary and eclectic approach using standard pharmacological and psychological therapies. The hospital has a daily outpatient clinic in which about 350–400 patients are seen. Patients requiring hospitalization are admitted for a period of 2–6 weeks.

The hospital follows standard protocols for the treatment of patients with schizophrenia. Antipsychotic medication (e.g. risperidone, olanzapine, chlorpromazine) are drugs of choice and employed in adequate doses (e.g. 4–8 mg of risperidone, 15–25 mg of olanzapine and 400–1000 mg of chlorpromazine) and for a period of about 2 years for people with first episode of psychosis. Clozapine is reserved for patients who do not respond to at least 2 antipsychotic drugs given in adequate dose for at least 6–12 weeks. All patients admitted to the hospital also receive psychoeducation, supportive psychotherapy and occupational therapy. Patients with specific problems related to marriage, sex, interpersonal and vocational issues are helped by psychological interventions (e.g. cognitive and behavioural therapy).

The details of the cohort are described in other publications [8, 18, 19]. The study group consisted of patients with schizophrenia having their first contact with mental health services and living within a 100 km radius of the study site. Patients were carefully screened for a DSM IV diagnosis of schizophrenia [28]. The details of the study were explained and written informed consent was obtained. The following instruments were employed for assessment at baseline:

We used Tamil versions of the instruments. The cohort was followed up at 6 and 12 months as part of the initial study [8, 18, 19]. The patients continued to receive treatment at the psychiatric hospital. We then attempted to follow up all the 131 patients at 60 months. We assessed patients who regularly attended the hospital. We visited and interviewed all other patients in their homes.

We used the following instruments to assess outcomes at 5 years: (i) Schedule for Assessment of Insight Expanded (SAI-E), [32, 33] (ii) the Brief Psychiatric Rating Scale (BPRS), [36] (iii) Short Explanatory Model Interview (SEMI), [20, 21]. We also employed the following instrument to assess outcome:

We defined remission as being less than mild on all items P1, P2, P3, N1, N4, N6, G5, G9 on PANSS [46] using standard criteria.

We used descriptive statistics to describe continuous variables and frequency distributions for categorical variables. We calculated odds ratios and their 95% confidence intervals. We employed multiple linear and logistic regression for multivariable analysis. We employed Generalized Estimating Equations (GEE) analysis to analyse longitudinal, repeated and correlated data. We used the SPSS version 16 for analysis.

The Institutional Review Board of the Christian Medical College, Vellore, approved the study protocols.

Table 1 records details of course and outcome over the 5-year period. The majority of subjects met remission criteria while a minority reached their pre-morbid level of functioning, were regular with medication and with follow up at the hospital. A little over half the subjects admitted that they believed that the disease model was their best explanation for their illness, while the rest preferred non-medical explanatory models.

Insight scores at 5-year follow up was negatively correlated with psychopathology [BPRS total score (Pearson's Corr. coeff. -0.57; p = 0.000), PANSS total score (Pearson's Corr. coeff. -0.66; p = 0.000)] and disability (Pearson’s Corr. coeff. -0.53; p = 0.000) and positively with the total number of non-medical models (Pearson’s Corr. coeff. 0.37; p = 0.000). BPRS total score (β = −0.55; t = −6.13; p = 0.000), PANSS total score (β = −0.65; t = −0.77; p = 0.000), disability (β = −0.51; t = −05.6; p = 0.000) and the number of nonmedical explanatory models (β = 0.356 t = −0.32; p = 0.001) remained statistically significantly associated with insight after adjustment for age, sex, literacy and residence using multiple linear regression.

The total number of non-medical explanatory models at 5^th year follow up was significantly correlated negatively with psychopathology [BPRS score (Pearson’s Corr. coeff. -0.36; p = 0.000), PANSS total score (Pearson’s Corr. coeff. -0.42; p = 0.000)] and disability (Pearson’s Corr. coeff. -0.30; p = 0.003) and positively with insight scores (Pearson’s Corr. coeff.0.37; p = 0.000). BPRS total score (β = −0.32; t = −3.33; p = 0.001), PANSS total score (β = −0.39; t = −4.2; p = 0.000), disability (β = −0.26; t = −2.7; p = 0.000) and insight score (β = 0.31; t = −3.3; p = 0.001) remained statistically significantly associated with the total number of non-medical models after adjustment for age, sex, literacy and residence using multiple linear regression.

Table 2 documents the factors associated with remission at 5 years on bivariate analysis. Table 3 and Table 4 record the relationships between remission on one hand and clinical variables and explanatory model variables respectively on the other, using multivariate procedures. Variables at baseline, which were statistically significant on bivariate analysis, and those variables significantly related at 0, 6, 12 and 60 months were employed for the multivariate statistical models. Clinical remission at 5 years was associated with pre-baseline factors (e.g. urban residence, fluctuating course of initial illness), improved functioning and shorter duration of total psychotic illness (e.g. better GAF scores at 6 months, lower psychopathology at 12 months). Insight scores and the presence of at least one non-medical explanatory model were only associated with remission later in the course of illness. The disease model was associated with remission while individual non-medical models (e.g. beliefs in evil spirit, black magic) were associated with poor outcome. These relationships remained statistically significant when logistic regression was used to adjust for factors associated with belief systems (e.g. age, sex, residence, education) and significant baseline variables. There was a significant interaction effect between disease explanation and non-medical explanatory models illness and remission. The relationship between remission and the different explanatory model variables, at 60 months, remained statistically significant even after adjusting for regularity of follow up over the 5-year period. Remission, insight scores and number of non-medical explanatory models of illness held were not significantly associated with number of hospital visits, the number of missed appointments or regularity of follow-up.

The association between remission and changes in SAIE and BPRS scores between baseline and 6 and 12 months were tested for statistical significance. The only significant variable associated with remission were reduction in BPRS score between baseline and 12 months (t = −3.16; df = 87; p = 0.002), which remained significant (OR = 0.91; 95% CI 0.85-0.98; p = 0.01) after logistic regression analysis adjusting for other clinically significant variables. Changes in insight scores during the first year of treatment were not associated with outcome at 5 years.

Figures 1 documents the serial insight and psychosis ratings and Figure 2 records the numbers of explanatory models of illness over the 5 year period. BPRS scores showed consistent decline while there was a steady increase in insight scores, number of patients who held disease model and those who simultaneously subscribed to disease and at least one non-medical model. The number of subjects who held at least one non-medical model was high at baseline, decreased with reduction in psychosis and treatment response and then seemed to increase with the stabilization of the course and outcome of illness.

The data was analysed using Generalized Estimating Equations (GEE) to assess the impact of serial assessments of SAI-E scores and the number of non-medical models on serial BPRS scores. Baseline variables significantly associated with remission and the number of visits to the hospital were entered as fixed variables in the model to adjust for their effects. Women (B = −2.65; SE = 1.03; p = 0.01), rural residence (B = − 2.06; SE = 1.00; p = 0.038), SAI-E scores (B = − 0.65; SE = 0.04; p = 0.000) and number of non-medical models (B = 5.31; SE = 0.91; p = 0.000) were significantly associated with BPRS scores. The non-significant variables in the model were: number of meals per day, duration of untreated psychosis, fluctuating initial course, and number of visits to hospital.

Many investigations have recorded the simultaneous presence of multiple and contradictory models of illness in people [8, 18‐23]. This is considered the norm in low and middle income countries but has also been recorded in western populations [20, 24]. Previous studies have documented high insight scores in people who subscribe to disease models [8, 9, 19]. In addition, investigations who have examined non-medical models have demonstrated their negative relationship with insight scores [8, 19]. Nevertheless, such correlations are natural considering the fact the instruments, which assess insight focus on disease explanations, attributions and actions, and concentrate on the recognition of mental illness attribution and seeking medical and psychiatric treatment. These instruments do not consider locally and culturally relevant attributions and help seeking.

The reduction of psychotic symptoms and improved functioning during the first year of treatment is associated with an increase in the use of disease explanations and a marked reduction of causal non-medical beliefs. However, the number of people subscribing to disease explanations tends to plateau with only about half the subjects preferring it at 5 years. The pattern of change in the number of indigenous models of illness employed argues that they help coping. Their manifest reduction at 6- and 12-month follow-up coincides with the reduction in psychosis for all patients. Their marked increase at 60 months suggests their role as a coping strategy when the pattern of course and outcome of illness becomes established. Previous studies have documented the inverse relationship between psychopathology and insight [8, 9, 19, 47‐49]. However, many of these studies focused on cross-sectional and early illness data or had failed to adjust for many illness characteristics.

The fact that pretreatment illness variables, improvement in functioning and reduction in psychopathology determined outcome at points in time before insight scores and explanatory models were significant suggest their major role in the process. The bivariate statistical association during the early course of illness between insight, explanatory models and outcome, lost their statistical significance when baseline and clinical variables were included in the multivariable analysis arguing that such relationships are confounded by illness characteristics. The independent predictor status of insight and explanatory models for clinical remission only becomes significant late in the course of illness. The positive relationship between insight and the number of non-medical explanatory models of illness, their negative relationship with psychopathology argues their impact on outcome is possibly secondary to the inherent nature of the disease. The relationships persisted after adjusting for the number of hospital visits.

These findings challenge the direction of the relationship between insight and psychopathology arguing that insight may be secondary to psychopathology. The association of insight and non-medical explanatory models to long-term outcome later in the course of illness suggests that explanatory models are coping mechanisms rather than being causally related to outcome.

The inverse correlation between insight scores and psychopathology in earlier reports [19, 47‐49] did not examine its impact on long-term outcome and take into consideration the time course of the illness. Although many studies have documented good insight as predictor of outcome, this relationship may be confounded by illness variables, which are usually not systematically excluded. The fact that pre-treatment variables and current psychopathology confound the relationship between explanatory models of illness at all points of time makes it difficult to conclusively argue that insight predicts longer-term outcome. Our data suggests the reverse, that psychopathology and illness characteristics predict insight, explanatory models and outcome in schizophrenia.

The heterogeneity of clinical features, course, outcome and treatment response of schizophrenia probably reflects different diseases, illnesses and trajectories. The fact that non-medical beliefs held during the later part of the illness were associated with non-remission and treatment non-response suggests that patients with a worse course and outcome and poorer response to treatment may select such explanatory models to cope with the devastating impact of their illness.

Pluralistic societies employ multiple approaches to health and illness. The simultaneous use of contradictory models suggests their use in coping with different aspects of the condition [27]. The fact that people with non-medical beliefs were willing to take anti-psychotic medication argues for the complexity of the issues demanding a nuanced understanding of explanations and coping mechanisms. The persistence of the association between poor outcome and non-medical causal models, after adjusting for regularity of hospital follow-up, suggests that they are coping mechanisms for poor outcome conditions.

Mental illness and their labels are stigmatizing across cultures leading to much prejudice and discrimination. This is particularly significant in people with residual and poor outcome schizophrenia whose limited response to medication, adverse effects and continued symptoms demands the need for explanations, which go beyond simple disease. Non-medical, supernatural and external explanatory models seem to be preferred to disease explanations in order to cope with the devastation of the disease in people who have not recovered with treatment. They seem to be culturally acceptable mechanism to cope with complex and incapacitating outcomes. The persistence of multiple explanatory models of across cultures for chronic illnesses may suggest an evolutionary advantage.

Medicine and psychiatry, despite their attempts at a universal understanding of mental disorders, cannot do justice to patients without the recognition of the contexts and the sharing of personal narratives. Psychiatry with its focus on objective behaviours, symptom checklists and diagnostic criteria has reduced the importance of patient experience and narratives [50]. Illness narratives contextualise the patient. They capture the individual’s suffering in an everyday context, in contrast to the medical narratives that reflect the needs of psychiatry. Patient experiences are particularly important for psychiatrists and mental health professionals who work every day with people with chronic and disabling mental illnesses but have never actually walked in their shoes. Illness narratives refocus the doctor-patient interaction and provide a window into the patient’s reality and his/her ways of coping [51].

The lack of laboratory tests to diagnose mental disorders tends to raise doubts about their diagnostic certainty making many patients and their families uncertain and uncomfortable about psychiatric labels [52]. In addition, negative attitude related to mental illness abound in most societies, result in stigma and lead to discrimination. The acceptance of such diagnostic labels can reduce self-esteem among patients. Patients, consequently, refuse to accept biomedical models and explanations for their experience and conditions.

The subjective dimension of insight has been conceptualized as a particular form of narrative production, called narrative insight [53]. Explanatory models of patients and their families are narratives to make sense of and convey experiences of illness, control them, maintain or improve quality of life [52]. These narratives can increase empathic understanding of illness experiences and life events, reduce their threat, provide some degree of control over these phenomena and incidents and improve their quality of life. They may often be effective aids to coping, particularly for treatment-resistant symptoms and deficits and incapacitating medication side effects.

There is a need to reconcile personal narratives with the biomedical model so that many personal strategies employed to coping with mental illness are not devalued or dismissed. Such experiential knowledge, although personal, will have to be recast in each case. It will help translate experience, provide legitimate frameworks and create a language and interface for improved communication [50]. The reframing of such narratives will impact psychiatric care, practice, research and health care delivery systems. Patient experience and perspectives, devalued and delegitimized by canonical authority, needs to be reemphasized and integrated into clinical practice. There is a need to foreground patient experience in order to impact the mainstream psychiatric discourse.

It has been argued that insight is shaped by psychology (i.e. motivation and denial) and constraints of biology (i.e. cognitive impairment and anosognosia) and is influenced by social constructions of illness (e.g. culturally specific explanatory models) [6, 7]. Its comparison to anosognosia, as in brain disease and damage, may not be very useful as its response to antipsychotic medication and change over time and limit such comparison. Similarly, its attribution to cognitive impairment is less than definitive in view of its response to antipsychotic medication, which is generally not very useful in improving the many cognitive deficits in schizophrenia. Its reciprocal relationship with psychopathology during the initial course of schizophrenia suggests its association with delusional thinking and beliefs. Severe delusional illnesses, by their definition, preclude alternative explanations especially at the height of the illness and in those with severe disease.

The association of disease models of illness in those with good response to treatment and the correlation between non-medical illness explanations and poor response suggests their role in coping. The complex interaction between disease and indigenous models and the impact on remission argues for a nuanced understanding of the relationships between explanatory models and outcome. The complexity of schizophrenia, its heterogeneity in course and outcome, its varied response to antipsychotic medication and its unpredictable impact on diverse aspects of life seem to elicit a multifaceted response from people who suffer from it, their relatives and from society.

Insight is not just the possession of discrete facts about the nature of mental illness. Nor is it only an acknowledgement of a particular experience as abnormal. It is also not an acceptance of a singular truth about the person and his/her life. An awareness of a mental illness is a narrative act in which people make personal sense of the many challenges they face. Such challenges include persistent psychotic symptoms, negative syndrome, cognitive deficits, impaired social relations, perpetual instability and livelihood issues. The patient’s narrative of the illness would then vary according complexity of the problem and context. The insight narrative would also vary according to its role in coping and adapting to the new challenges and demands. The cultural context, acceptable cultural explanations and the prevalent social stigma related to mental illness would interact with the person’s illness to produce a particular understanding. The complexity of issues calls for a nuanced framing of insight.

Alternative conceptualisations of insight have employed findings from investigations in the non-western cultures [14, 22‐24, 27, 54]. These findings include: (i) patient’s often provide non-medical explanations for the cause of illness (e.g. karma, evil spirits, black magic, sin, punishment by God, etc.), (ii) mental illness labels are very stigmatising across cultures and settings, (iii) many subjects simultaneously hold multiple, diverse and contradictory beliefs and reflect the pluralistic approach of patients, (iv) many patients simultaneously seek biomedical and non-biomedical interventions from a variety of healers and centres of cure and healing. These findings argue that the current multidimensional models are not culturally sensitive to assess insight. They suggest that it is difficult to have definable, objective or universal standard for insight. They argues that there is a need for universal conventions to compare individual explanations with sub-cultural perceptions [14, 54].

Alternative dimensions have been suggested for the assessment of insight (Table 5) [14, 54]. Insight should be assessed against local and cultural standards of providing non-delusional and culturally acceptable explanations and attributions and seeking locally acceptable and available interventions. The findings of this study also suggest that subscribing to multiple explanatory models of illness argues that it is a pragmatic response to the devastation of chronic and residual psychosis. The assessment of insight should be similar to the assessment of other clinical phenomena like delusions, which involve comparison with local and cultural yardsticks [14, 54].

The findings argue for the fact that the assessment of insight should be against the local cultural standards including disease models rather than universal yardsticks (e.g. of solely employing biomedical models). The assessment of insight should evaluate awareness, attribution and action. People with psychosis who are able to re-label their psychotic experience, offer non-delusional explanations for changes in themselves, which correspond to beliefs about illness held by the subculture, admit to the need for restitution, and seek locally available help, can be said to possess insight. The results recommend the use of universal conventions (of awareness, attribution to locally accepted non-delusional explanations and seeking available help) to assess insight in people with psychosis rather than the use of uniform criteria which focus solely on the biomedical model of disease.

Many people hold multiple causal and treatment models of their illness [55]. This is particularly true for chronic and disabling illnesses, which do not completely respond to medication and treatment. Attempts at elicitation of the locally relevant causal and treatment models of illness in addition to the biomedical model will provide clinicians with an understanding of residual problems and disability and the patient’s attempt at coping. While medication compliance is crucial to the reduction of psychotic symptoms and the maintenance of remission, other non-medical models also help in providing a pragmatic response to the persisting symptoms and problems. Non-medical beliefs should not be challenged or dismissed and the clinician should not claim exclusivity or superiority of the biomedical model. The use of diverse coping strategies for healing should be encouraged while emphasising medication compliance. Assessing insight in clinical practice will mandate the need to elicit all causal explanatory models and identify the diversity of help seeking. The use of a combination of different explanations including the biomedical model suggest attempts at coping employing many diverse culturally sanctioned strategies reflecting a greater awareness of the illness as well as the local context and environment (i.e. insight).

Research into insight in psychosis should also include non-medical belief models and move beyond pure biomedical explanations. There is a need to develop and refine instruments, which are able to capture diverse belief systems commonly employed by people with mental illness and also elicit local explanatory models of illness, their attribution and help seeking.

The findings of this study mandate replication. The sole use of biomedical perspectives in the evaluation of response to complex diseases needs to be replaced with a broad based approach and understanding of coping across cultures. People employ diverse approaches to maintain mental health. The partial solutions offered by individual systems of medicine force people to employ diverse and multiple strategies to cope with distressing symptoms and intractable problems. Consequently, there is a need for a non-judgemental approach to the explanatory models employed by people with mental illness. While antipsychotic medication has a powerful impact on outcome, the use of other strategies especially in chronic conditions to cope with the variable impairment, disability and handicap, even among those with clinical remission, suggest the need to allow patients to use multiple strategies to regain and maintain mental health.

Cultural knowledge, empathy and an understanding of its complex interaction, which impact diseases and shape the illness response, is crucial to providing culturally sensitive care. Presenting biomedical perspectives without dismissing patient beliefs and negotiating a shared treatment plan without claiming exclusivity are cardinal. This will allow mental health professionals to bridge the disease-illness and the healing-cure divides across all cultures and settings. The complexity of the issues demonstrated highlight the need for a nuanced understanding of issues related to insight, explanatory models of illness and culture in schizophrenia and for other chronic illness.