Schizophrenia relapse in Kuala Lumpur, Malaysia: Do relatives expressed emotion and personality traits matter?

Similar documents
Anti-psychotics started widely used in 1950 s

CHAPTER-V SUMMARY, CONCLUSION AND RECOMMENDATION

COPING STRATEGIES OF THE RELATIVES OF SCHIZOPHRENIC PATIENTS

Procedia - Social and Behavioral Sciences 140 ( 2014 ) PSYSOC 2013

OUTCOMES OF DICHOTOMIZING A CONTINUOUS VARIABLE IN THE PSYCHIATRIC EARLY READMISSION PREDICTION MODEL. Ng CG

Daniel Boduszek University of Huddersfield

COMPARISON OF FAMILY ENVIRONMENTAL SCALE (FES) SUBSCALES BETWEEN MALAYSIAN SETTING WITH THE ORIGINAL DIMENSION OF FES

S P O U S A L R ES E M B L A N C E I N PSYCHOPATHOLOGY: A C O M PA R I SO N O F PA R E N T S O F C H I LD R E N W I T H A N D WITHOUT PSYCHOPATHOLOGY

THE RELATIONSHIP BETWEEN LIFE SATISFACTION AND ATTACHMENT STYLES WITH PSYCHOLOGICAL WELL-BEING OF CHILDREN ON COVERED BY BEHZISTEY IN TEHRAN

Extraversion. The Extraversion factor reliability is 0.90 and the trait scale reliabilities range from 0.70 to 0.81.

Work Personality Index Factorial Similarity Across 4 Countries

THE PREVALENCE OF DEPRESSIVE SYMPTOMS AND POTENTIAL RISK FACTORS THAT MAY CAUSE DEPRESSION AMONG ADULT WOMEN IN SELANGOR

Exploring the Relationship Between Substance Abuse and Dependence Disorders and Discharge Status: Results and Implications

Journal of American Science 2010;6(10) Age and gender differences and construct of the children s emotional intelligence

BEHAVIORAL ASSESSMENT OF PAIN MEDICAL STABILITY QUICK SCREEN. Test Manual

Neurotic Styles and the Five Factor Model of Personality

VARIABLES AND MEASUREMENT

University of Warwick institutional repository:

Personality. Unit 3: Developmental Psychology

Family-centered Stress Management for Childhood Cancer: A Multimodal Intervention for Children Newly Diagnosed with Cancer and their Families

THE EFFECTS OF OWNING A PET ON SELF-ESTEEM AND SELF-EFFICACY OF MALAYSIAN PET OWNERS

Wisconsin Dementia Care Guiding Principles

Supplementary Online Content

The Bengali Adaptation of Edinburgh Postnatal Depression Scale

Questionnaire on Anticipated Discrimination (QUAD)(1): is a self-complete measure comprising 14 items

DESIGN TYPE AND LEVEL OF EVIDENCE: Randomized controlled trial, Level I

Assessing the Validity and Reliability of the Teacher Keys Effectiveness. System (TKES) and the Leader Keys Effectiveness System (LKES)

Everyday Problem Solving and Instrumental Activities of Daily Living: Support for Domain Specificity

Bringing hope and lasting recovery to individuals and families since 1993.

Personality measures under focus: The NEO-PI-R and the MBTI

Reliability and Validity of the Pediatric Quality of Life Inventory Generic Core Scales, Multidimensional Fatigue Scale, and Cancer Module

Managing Depression as a Chronic Condition. D. Green MD TOH/Bruyere Shared Care Program

CHAPTER 3 RESEARCH METHODOLOGY. In this chapter, research design, data collection, sampling frame and analysis

Validation of the Chinese Juvenile Victimisation Questionnaire

Mood Disorders Society of Canada Mental Health Care System Study Summary Report

PSYCHOLOGIST-PATIENT SERVICES

Collecting & Making Sense of

Impacts of Instant Messaging on Communications and Relationships among Youths in Malaysia

CHAPTER-5. Family Disorganization & Woman Desertion by Socioeconomic Background

Women s Health Development Unit, School of Medical Science, Health Campus, Universiti Sains Malaysia b

Calculating clinically significant change: Applications of the Clinical Global Impressions (CGI) Scale to evaluate client outcomes in private practice

Glasgow, 1055 Great Western Road, Glasgow G12 0XH, UK

Autobiographical memory as a dynamic process: Autobiographical memory mediates basic tendencies and characteristic adaptations

Procedia - Social and Behavioral Sciences 116 ( 2014 )

User Evaluation of an Electronic Malaysian Sign Language Dictionary: e-sign Dictionary

Lecture Outline Signs and symptoms in psychiatry Adjustment Disorders Other conditions that may be a focus of clinical attention

CHAPTER 2 Personality Traits: A Good Theory

Circumstances Surrounding Accidental Poisoning in Children

Multiple Act criterion:

ALABAMA SELF-ASSESSMENT INDEX PILOT PROGRAM SUMMARY REPORT

The Perinatal Mental Health Project (PMHP)

The Relationship between Personality Traits and Reading Proficiency

Personality: Definitions

Validation of the Russian version of the Quality of Life-Rheumatoid Arthritis Scale (QOL-RA Scale)

Opinion of Primary Care Giver to Psychotherapy Regarding Neurosis Patients

SUMMARY AND DISCUSSION

Author's response to reviews

Examining the Psychometric Properties of The McQuaig Occupational Test

Media, Discussion and Attitudes Technical Appendix. 6 October 2015 BBC Media Action Andrea Scavo and Hana Rohan

Help-seeking behaviour and its impact on patients attending a psychiatry clinic at National Hospital of Sri Lanka

Critical Thinking Assessment at MCC. How are we doing?

Mental Health in Workplaces in Taipei

Preliminary Conclusion

The Counselors Psychological Well Being

Factor Analytic Theories. Chapter 11

Aiming for recovery for patients with severe or persistent depression a view from secondary care. Chrisvan Koen

PLS 506 Mark T. Imperial, Ph.D. Lecture Notes: Reliability & Validity

Methodology Introduction of the study Statement of Problem Objective Hypothesis Method

PREVALENCE AND DETERMINANTS OF APPROPRIATE HEALTH SEEKING BEHAVIOUR AMONG KNOWN DIABETICS: RESULTS FROM A COMMUNITY- BASED SURVEY

MJP Online Early

Statistics Anxiety among Postgraduate Students

Factor Analytic Theories

The happy personality: Mediational role of trait emotional intelligence

Impulsivity, negative expectancies, and marijuana use: A test of the acquired preparedness model

Title: The Theory of Planned Behavior (TPB) and Texting While Driving Behavior in College Students MS # Manuscript ID GCPI

The relationship of five personality factors with the feeling of happiness among students

Sense-making Approach in Determining Health Situation, Information Seeking and Usage

CALIFORNIA COUNTIES TREATMENT RECORD REQUIREMENTS

Risk-Assessment Instruments for Pain Populations

Modeling Health Related Quality of Life among Cancer Patients Using an Integrated Inference System and Linear Regression

The Impact of Relative Standards on the Propensity to Disclose. Alessandro Acquisti, Leslie K. John, George Loewenstein WEB APPENDIX

ORIGINAL ARTICLE Validation of the Hospital Anxiety and Depression Scale and the psychological disorder among premature ejaculation subjects

DVI Pre-Post: Standardization Study

RJSS. The Relationship between Personality Dimensions and Social Capital of Athletic Trainers in Azerbaijan Gharbi Province

Sex Differences in Depression in Patients with Multiple Sclerosis

Basic concepts and principles of classical test theory

Depression, Anxiety and Locus of Control among Elderly with Dementia

PERSONALITY OF POLISH SOLDIERS AND THEIR WAY OF STRESS COPING DURING BOSNIA PEACEKEEPING MISSION

The Prevalence of Depression Among the Elderly in Sepang, Selangor

Stigma in Patients Using Mental Health Services

RATING MENTAL WHOLE PERSON IMPAIRMENT UNDER THE NEW SABS: New Methods, New Challenges. CSME/CAPDA Conference, April 1, 2017

The Impact of Emotional Intelligence towards Relationship of Personality and Self-Esteem at Workplace

CHAPTER 3: METHODOLOGY

Screening for depressive symptoms: Validation of the CES-D scale in a multi-ethnic group of patients with diabetes in Singapore

Setting Non-profit psychiatric hospital. The economic analysis was carried out in the USA.

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

Analysis of Confidence Rating Pilot Data: Executive Summary for the UKCAT Board

Adult Patient Information and Consent Form

The Relationship between Personality Styles and Academic Self- Efficacy among Jordanian EFL Learners

Transcription:

Available online at www.sciencedirect.com ScienceDirect Comprehensive Psychiatry 55 (2014) 188 198 www.elsevier.com/locate/comppsych Schizophrenia relapse in Kuala Lumpur, Malaysia: Do relatives expressed emotion and personality traits matter? AbRahman Roseliza-Murni a,, Tian P.S. Oei b, Yusooff Fatimah a, Desa Asmawati a a School of Psychology & Human Development, Faculty of Social Sciences & Humanities, Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor, Malaysia b School of Psychology, The University of Queensland & CBT Unit, Toowong Private Hospital, Qld 4072, Australia Abstract Background: Schizophrenia which is perhaps the most disabling and puzzling form of a mental disorder is often conceptualized as gila (lunacy or madness) by the Malaysian society. The debilitating nature of the disorder and recurrent relapse of its psychotic episodes have often been misunderstood and lead to confusion among the family members, who play the role as primary caregivers. While expressed emotion (EE) has been widely studied in the Western world, it is not well understood in Malaysia. Objectives: This study aimed at investigating the predictive role of relatives EE index and components as well as personality traits in relapse among schizophrenia patients in Kuala Lumpur and Klang Valley, Malaysia. Methods: A total of 160 subjects consisting of 80 patients diagnosed with chronic schizophrenia and 80 relatives were recruited to participate in Phase 1 of the study. Only patient-participants were followed-up 6 months later for Phase 2 of the study in order to check for possible relapse. The Family Questionnaire (FQ) and the short scale Eysenck Personality Questionnaire-Revised (EPQ-R) were administered to measure relatives EE and personality traits, respectively. Patients were considered to relapse if they were readmitted into psychiatric wards 6 months post-hospital discharge. Results: The findings revealed that the odds for patients to relapse were increased by 8 times when relatives demonstrated high-ee level. Relatives critical comments (CC) turned out to be the strongest predictor with the odds to relapse increased by 12% when they demonstrated an increase in CC level. Relatives personality traits particularly the extraversion trait also turned out to be the significant direct predictor to patients relapse. Our results showed that a unit decrease in extraversion trait score predicted the odds for patients to relapse by 23%. Discussions: Our findings supported the Western findings on the significant role of relatives high-ee level particularly the CC scale on the course of schizophrenia. The novelty of the current finding was demonstrated in the significant role of relatives extraversion trait that directly predicted patients relapse. Results also indicated the feasibility of EE in predicting relapse among schizophrenia patients in Klang Valley, Malaysia. 2014 Elsevier Inc. All rights reserved. 1. Introduction People suffering from schizophrenia have often been reported to experience recurrent relapse of its psychotic episodes. In fact many people who improved after an episode of schizophrenia tended to experience the psychotic symptoms again later on in their lives. Past research found that 78% of Disclosures: The authors declare that they have no conflict of interest to disclose concerning the subject of this article. Funding supports are detailed in the Acknowledgement section. Additional contributions: All the authors have collaborated in this study and have approved of the final manuscript. Corresponding author. Tel.: +60 3 8921 3602; fax: +60 3 8921 3541. E-mail address: rmurni@ukm.my (A. Roseliza-Murni). those who experienced one psychotic episode tended to experience recurrent relapse with different degrees of impairment in between the psychotic episodes [1]. According to Shepherd et al. [2] nearly half of the 78% schizophrenia patients who experienced repeated relapse will experience a progressive deterioration in functioning following each successive relapse. Another study also reported that more than 50% of the patients with schizophrenia have poor outcome which among others include repeated hospitalizations and symptom exacerbation [3]. In Malaysia, attempts to collect information about patients with mental disorders have only started in 2003 by the National Mental Health Registry (NMHR) and the first report was published in 2006 [4]. This pioneering project was targeted at gathering data on new cases of schizophrenia that sought treatment from government-based 0010-440X/$ see front matter 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.comppsych.2012.12.026

A. Roseliza-Murni et al. / Comprehensive Psychiatry 55 (2014) 188 198 189 facilities, university hospitals and private health services in order to assist the Ministry of Health (MOH) Malaysia, healthcare providers and non-governmental organizations towards planning and evaluating the mental health services [5,6]. Even though NMHR was established to collect data on first-episode schizophrenia, some of the newly registered cases also included those of chronic schizophrenia patients who were being treated elsewhere [6]. Since this pioneer project only targeted data on the first-episode schizophrenia [6], the exact data regarding schizophrenia relapse cases in Malaysia are yet to be ascertained. A few factors have been implicated in schizophrenia relapse. Kingdon and Turkington [7] for instance, suggested that relapse can be linked to life events and circumstances such as low-key but enduring type of stress or due to specific factors that trigger anxiety to which the person is vulnerable. One of the main contributory factors that have been consistently found to relate to relapse is the emotional climate of household environment demonstrated by family members toward people with schizophrenia known as the expressed emotion (EE) [8 13]. EE has been considered as athermometerof affective family environment [14, p.80] and it has been suggested to reflect family members general attitudes toward patients with schizophrenia in their daily interactions [15]. Generally, EE measures the emotional climate within the home environment [16] indexed by the presence of critical attitudes/ critical comments (CC), hostile behavior and emotionally over-involved (EOI)/intrusive attitudes when relatives speak about the patients in an interview conducted during patients psychiatric hospitalization [17]. The concept of EE which was based on Brown and colleagues seminal works on the role of family factors in influencing the course of schizophrenia [18 21] has indeed generated much research interest within the Western world. More recent studies have also repeatedly demonstrated that patients with schizophrenia who returned to live with their families that demonstrated high-ee level tended to relapse twice as likely within 6 to 9 months post-hospital discharge compared to their counterparts who returned to low-ee households [10,13,22]. An aggregate analysis of 25 studies by Bebbington and Kuipers [23] found that 50% of the patients in high-ee household relapsed compared to 21% of those in low-ee household who relapse within the same specified period. In sum, this suggests the robust role of EE in predicting schizophrenia relapse among the Western samples. Nevertheless, EE's role in influencing the course of schizophrenia among Malaysian populations has not received much attention. Up to this point there is only one published study on EE among relatives of schizophrenia patients conducted in Malaysia which specifically targeted patients and relatives who lived in Kelantan and Terengganu. This study investigated the prevalence of high-ee attitudes among 83 relatives of schizophrenia patients and found that majority of them (72.3%) demonstrated low-ee level. The authors did not specifically conduct any analysis with regard to the predictive role of EE on schizophrenia relapse however, they concluded that the EE levels among their study sample did not predict schizophrenia relapse in the patients [24]. These authors postulated that the most likely explanation to their findings was cultural differences between schizophrenia patients and their relatives in Kelantan and Terengganu compared to those in other countries [24], however the authors did not provide further explanations to their claims. Another possible explanation to Azhar and Varma's [24] finding is the validity of the 30-item questionnaire used to measure EE, as it was personally drawn and rated in the form of a Yes/No/Infinite response style with no psychometric properties reported. Thus, it may contribute to differences in their findings compared to the worldwide findings. Furthermore, the floor effect of 72.3% among low-ee relatives may have reduced the variance for a significant result. One area that has been neglected in schizophrenia relapse research is the role of relatives personality traits that may directly impact on the course of schizophrenia. To date, the role of relatives personality traits have been investigated as one of the predictors of EE in relatives and indeed it has been found that relatives EE level was related to some aspects of their personality traits [15,25]. It has been discovered that high-ee relatives tended to score significantly higher on norm-favoring subscale but scored significantly lower than their counterparts in low-ee group on the tolerance, flexibility, empathy and selfrealization subscales of the California Psychological Inventory (CPI). Of these subscales, flexibility turned out to be the single most relevant trait that contributed towards relatives EE level [25]. Meanwhile, another study demonstrated that high-ee mothers scored significantly higher on the conscientiousness and significantly lower on the neuroticism scale of the Revised NEO-Personality Inventory (NEO-PI) than the low-ee mothers [15]. Taken together, results of these studies provided clinically useful information regarding the role of relatives personality traits that may be associated with their EE levels. However, the direct impacts of relatives personality traits on the course of schizophrenia have not been investigated. The principal aims of the present study were to investigate the predictive value of relatives EE index (high and low) and its components (CC and EOI) on schizophrenia relapse and the predictive value of relatives personality traits on relapse among schizophrenia patients. It was hypothesized that: 1. Relatives high EE level would be a significant predictor of schizophrenia relapse. 2. Relatives EE components (CC and EOI) were both significant predictors of schizophrenia relapse. 3. Relatives personality traits (extraversion, neuroticism and psychoticism) were significant predictors of schizophrenia relapse. 2. Methods 2.1. Participants Patients with the primary diagnosis of schizophrenia who were receiving inpatient psychiatric treatments during

190 A. Roseliza-Murni et al. / Comprehensive Psychiatry 55 (2014) 188 198 October 2009 through February 2010 (Phase 1) at Hospital Kuala Lumpur (HKL) and University Malaya Medical Center (UMMC), together with their key relatives were invited to participate in this study. Both hospitals are located at the densely populated areas of Kuala Lumpur and served to treat patients within a 25-km radius. This study employed purposive sampling method whereby all patients who fulfilled the study requirement according to the inclusion and exclusion criteria were approached. A total of 160 participants (80 patients and 80 relatives) participated. The following inclusion criteria were used for patient selection: 1. diagnosed with schizophrenia by the treating psychiatrists based on the diagnostic criteria of DSM-IV [26] or DSM-IV-TR [27]; 2. received inpatient psychiatric treatments and were identified by their hospital records or referrals from the psychiatrists or psychiatric nurses; 3. had at least a prior history of psychiatric inpatient due to schizophrenia; 4. lived with relatives at least 3 months prior to onset of the recent psychotic episode; 5. would be living with relatives post-hospital discharge; 6. were granted informed consents by their next of kin or the responsible relatives to participate in the study; and 7. could speak in either Bahasa Melayu or English. Patients were excluded from the study if they suffered comorbid substance abuse or had major disabilities, organic brain disorder or mental retardation that could compromise their functioning during periods of remission. In cases where patients were able to speak in Bahasa Melayu or English but relatives have difficulties in conversing and understanding the language, these patients were also excluded from the study. Each patient admitted to the psychiatric ward had details about one other person who was responsible for his or her well-being and these relatives declared themselves as the next of kin. Details about the next of kin were available in patients hospital records. Thus, the person who declared him/ herself as the next of kin was approached and invited to participate as relative-participants. Relatives were recruited when they expressed willingness and gave written consents to participate in the study. Relative-participants were also required to be able to converse in either Bahasa Melayu or English. 2.2. Procedures Prior to starting the study, approvals, consents and ethical clearances were obtained from the Ministry of Health Research and Ethics Committee (MREC) and Medical Ethics Committee (MEC) of UMMC. Once the approvals, consents and clearances were obtained, an online registration with the National Medical Research Registry (NMRR) was conducted and the current research was assigned with ID2640. The NMRR is a web-based tool designed to support the implementation of the National Institute of Health guidelines which are applicable to all researches that used the MOH facilities in order to ensure transparency and to increase public trust in the conduct of medical related research. The following procedures were implemented for data collection. Patients were initially assessed by the psychiatrists at both hospitals for the diagnosis of schizophrenia. Once the diagnosis was confirmed, patients hospital records were reviewed to check for those who met the inclusion and exclusion criteria of this study. Their key relatives were then contacted via telephone by the first author (ARRM) to explain about the study and to get their informed consents to include the patients as well as to invite these relatives to participate in this research. Once the relatives gave voluntary informed consents for patients to participate and they also agreed to become the participant, an appointment for interview with relatives was arranged. Patients were also approached for participating in this research and they were explained about the nature of this study. A total of 89 relatives whose patients met the inclusion and exclusion criteria were approached and only 80 of them participated (5 relatives declined to participate while another 4 who initially agreed to participate in the study did not show up on the appointment day and did not pick up the telephone for further appointments). The final sample involved 80 patients and 80 relatives. Interviews with the relatives were conducted during visiting hours either in the afternoon or in late evenings in a consultation room within the psychiatric ward. During this interview relatives were asked about patients illness history and they were required to fill in the questionnaires designed to assess their personality traits, EE-index and its components. In cases where the relatives were either illiterate or have difficulties reading the questionnaires (some were due to short-sightedness), the experimenter would read the questions to them and mark their answers according to their responses. The length of time taken for the interview session was about 30 min. Subsequent to the interview, the relatives would meet up with the patients for social visits. The procedures for assessing relatives EE index and components during patients psychiatric hospitalization and that their attitudes during that time were predictive of patients relapse 6 to 9 months later were consistent with the procedures of past research on relatives EE and schizophrenia relapse [8 14,18 23]. Patients conditions were followed-up 6 months later to check for possible relapse. The 6-month follow-up period was based on the suggestions that an episode of schizophrenia lasted for 6 months and for patients on medications such as the current sample an episode of schizophrenia that is successfully treated may not last longer than 1 month [26,27]. The 6-month follow-up period was also considered as the conservative relapse criteria by past researches [28,29]. 2.3. Measure for relatives EE The Family Questionnaire (FQ) [30] a 20-item selfreported instrument was administered to measure relatives

A. Roseliza-Murni et al. / Comprehensive Psychiatry 55 (2014) 188 198 191 EE level. It measures two distinct constructs of EE, the CC and EOI in which 10 items were designed to measure each construct. The FQ is scored on a 4-point scale that range from Never/Very Rarely to Very Often. Items of each scale are added up to represent the score for that particular scale. The cut-off scores for the CC and EOI scales are 23 and 27, respectively. Thus, relatives are classified as high-ee if they scored either 23 or greater on the CC scale or 27 or greater on the EOI scale. They are considered as high EE if their scores on at least one of the two scales exceeded the cut-off point [30]. The cut-off scores for both EE index and components remained as suggested by Weidemann et al. [30] since reliability analysis yielded high values for both components. The Selection of FQ over other instruments such as the Camberwell Family Interview (CFI) or the Five Minute Speech Sample (FMSS) was also due to its internal consistency particularly the FQ-EOI scale, which was higher than the EOI scale of the FMSS and CFI. Evaluation of correct classification rate between the FQ and CFI suggested high agreement between the two instruments in classifying relatives EE level. The FQ correctly classified 74% of high- EE relatives that had been rated by the CFI [30]. It had been suggested that the FQ was so far the only paper-and-pencil measure that closely associated with both the CC and EOI scales of the CFI and correctly classified relatives in CC and EOI groups 78% and 71%, respectively [31]. Since Bahasa Melayu is the national language and English is the second language in Malaysia, the FQ was presented in both languages. Back translation was employed for translating the FQ into Bahasa Melayu. To address any difficulties with regard to linguistic and content equivalence between Bahasa Melayu and English, two Western trained psychologists who were well-versed in both Bahasa Melayu and English as well as an English university lecturer were involved in the translation process. Reliability analysis for the current FQ- Bahasa Melayu version revealed that Cronbach's alphas for the CC and EOI were 0.90 and 0.88, respectively. 2.4. Measure for relatives personality traits The short scale Eysenck Personality Questionnaire-Revised (EPQ-R) [32] is a well-established instrument that has been extensively used in many studies on personality traits. The EPQ-R is a self-reported 48-item dichotomous response questionnaire that measures three distinct personality traits in addition to a social desirability scale. The four subscales of the EPQ-R are the extraversion, neuroticism, psychoticism and lie scales which measure different facets of personality via 12 items on each scale [32]. Just like the FQ, the EPQ-R was also presented in both Bahasa Melayu and English versions. Two Western trained psychologists and an English university lecturer as described in the translation of FQ were also employed for translating the EPQ-R into Bahasa Melayu. Reliability analysis for three personality traits of EPQ-R Bahasa Melayu version revealed that Cronbach's alphas were as follows: extraversion = 0.77, neuroticism = 0.81 and psychoticism = 0.17. The low reliability level of the psychoticism scale was consistent with the argument of Eysenck et al. [32] that it suffered from several psychometric shortcomings such as low reliability, low range of scoring and grossly skewed distribution. In addition, it had also been argued that the psychoticism scale taps several different facets of personality such as lack of empathy, hostility, cruelty and non-conformism, which may influence its low reliability level when compared with the extraversion scale which largely measures the sociability and activity levels [33]. Thus, the low reliability of the psychoticism scale in the present study may reflect the above postulations. 2.5. Measure for patient relapse Relapse is operationally defined as psychiatric rehospitalization that resulted from the return or worsening of schizophrenia symptoms as assessed by the treating psychiatrists. The key measurement of relapse is rehospitalization, and it is based on the suggestion that hospitalization often indicates that clinical symptoms have reached a level of severity that cannot be tolerated or safely managed in an outpatient department [34, p.408]. Therefore, patients were considered to relapse if they have been readmitted into the psychiatric wards 6 months post-hospital discharge. All patients were considered to have equal chance of receiving inpatient psychiatric treatment for their relapse and since most of them were registered with the Department of Social Welfare, their treatment charges were free. This thus should rule out the possibility of not admitting the patients for psychiatric treatment due to economic reasons. 2.6. Patients and relatives characteristics Schizophrenia patients consisted of 56 males and 24 females in which 55% of them were Malays, 25% Chinese, 16.25% Indians and 3.75% of other races. The 70% maleand 30% female-patients reflected the demographic profile of the registered cases provided by the NMHR report with a total of 2:1 ratio between male and female patients. The ethnic group distribution also resembled the data provided by NMHR report whereby 55% of the registered cases were Malays, followed by 28% Chinese and 9% Indians, which were reflective of the Malaysian population [35]. Patients marital status indicated that 62.5% of them were single, 23.8% were married, while the remaining 13.8% were either widowed or divorced. Again these data comply with NMHR statistics whereby 68% of the patients in the registry reported to be single, 23% married and 8% were either widowed, divorced or separated [35]. Patients age at the first onset of schizophrenia ranged from 12 to 42 years (Mean=22.15, Standard Deviation [SD]=5.76) and their past psychiatric admissions ranged from one to 20 times (Mean=5.37; SD=4.51). For majority of the patients (97.5%), the mean number of days admitted to the psychiatric ward was 25 days, while two patients (2.5%) were hospitalized for 45 days and 60 days, respectively. All patients were

192 A. Roseliza-Murni et al. / Comprehensive Psychiatry 55 (2014) 188 198 discharged when their symptoms were stabilized. With regard to pharmacotherapy, 71.5% of the patients were prescribed with typical antipsychotic medication while 28.5% received atypical antipsychotic medication. At 6-month post-hospital discharge 63.8% of the patients were reported to be taking medication regularly, 20% irregularly while the remaining 16.2% defaulted on treatment. All of these patients were living with at least one family member prior to their psychiatric hospitalization and at post-hospital discharge. Relative-participants consisted of 48.8% males and 51.2% females. These relatives were the primary caregivers and their relationships with the patients included mothers (31.2%), fathers (16.2%), spouses (15%), siblings (27.5%), adult offspring (5%), aunts/uncles (3.75%) and a grandmother (1.25%). The length of time these relatives were living with the patients ranged from 1 to 65 years (Mean = 23.46, SD = 12.68), while the mean number of family members living in the same household with the patients was 4.2 (SD = 2.13). 2.7. Analysis Preliminary data analyses showed that the data did not violate the statistical assumptions. These data were analyzed using the IBM SPSS Statistics 20. The two-way contingency table analysis using crosstabs was employed to evaluate the statistical relationship between categorical variables. Meanwhile, t-tests were used to assess the differences in variables between patients who relapsed and those who did not relapse. Logistic regression was administered to predict the contribution of relatives EE index, EE components and personality traits on patients relapse. Logistic regression was used as it is able to predict the discrete outcome from a set of predictors that may be continuous, discrete, or dichotomous in nature, or even a mixture of these variables [36]. 3. Results 3.1. The distribution of patients relapse and relatives EE At 6-month post-hospital discharge 52.5% (42) of the patients had relapsed and were readmitted into the psychiatric ward (Phase 2). The distribution of relapse based on ethnic groups revealed that 61.4% of Malay patients relapsed, 50% of Chinese patients relapsed, 30.8% of Indian patients relapsed and 33.3% of patients from other races relapsed. The difference of relapse rate between ethnic groups was not significantly different (Pearson χ 2 = 4.34, p = 0.227). Relatives EE index and components measured during patients initial psychiatric hospitalization (Phase 1) demonstrated that 53.8% of the relatives were classified into the low-ee category, while the remaining 46.2% demonstrated high-ee level. Eighty percent of the relatives were categorized in the low-eoi group, while the remaining 20% were classified as high-eoi. Meanwhile, 65% of the relatives scored in the low-cc category and the remaining 35% were classified as high-cc. Comparison of index EE among ethnic groups revealed that 47.7% of Malay relatives demonstrated high-ee, followed by 50% of Chinese relatives, 38.5% of Indian relatives and 33.3% of relatives from other ethnic group. There was no significant difference in high-ee index between ethnic groups, Pearson χ 2 =0.67, p=0.88. Further comparison between the three ethnic groups also revealed that 34.1% of the Malay relatives were categorized into high-cc, 45% of the Chinese relatives were categorized into high-cc, followed by 23.1% Indians and 33.3% relatives of other ethnic group. The difference in high-cc between ethnic groups was not statistically significant, Pearson χ 2 = 1.71, p = 0.63. Likewise, comparison between ethnic groups with regard to EOI component also did not yield any statistically significant finding, χ 2 =3.15,p= 0.37. Table 1 summarizes the distribution of patients relapse and relatives EE index and components based on ethnic groups. In addition, assessment of patients symptom severity conducted during their index hospitalization using the Structured Clinical Interview for Positive and Negative Syndrome Scale (SCI-PANSS) [37] also revealed no significant difference between those who lived with high-ee and low-ee relatives, t = 1.39, p = 0.17. Table 2 summarizes the correlation matrix between all variables scored by the relatives on EPQ-R and FQ. As can be seen from Table 2 both of relatives neuroticism and extraversion traits had significant correlations with their CC scores. Relatives neuroticism trait was also significantly negatively correlated with their extraversion trait. However, these correlations were at the lower ends. 3.2. The predictive value of relatives EE index on schizophrenia relapse A logistic regression analysis was performed to assess the prediction of patients relapse 6 months post-hospital discharge on the basis of relatives EE index (high and low-ee). The model with EE index yielded a likelihood ratio χ 2 (1) = 19.37, p = 0.000, Nagelkerke R 2 = 0.29, suggesting that the model fits significantly better than the constant-only model. This model also yielded an improvement in the overall correct classification value of 73.8% compared to the constant-only model (52.5%). Among the correctly classified cases, 78% was for the prediction of non-relapse cases and 69% for the prediction of relapse cases. Results indicated that relatives high-ee level contributed significantly to the prediction of schizophrenia relapse, Wald χ 2 = 16.73, p = 0.000. For every one unit increase in relatives high-ee level, the odds for patients to relapse 6 months later were increased by a multiplicative factor of 8.37 (95% CI: 3.02 23.15). Table 3 summarizes the contribution of relatives high-ee level to the prediction of patients relapse. 3.3. The predictive value of relatives CC and EOI on schizophrenia relapse To avoid multicollinearity between index EE (high and low-ee level) and its components (CC and EOI), a separate

A. Roseliza-Murni et al. / Comprehensive Psychiatry 55 (2014) 188 198 193 Table 1 Distribution of patients relapse and relatives EE index and components based on ethnic groups. Malays Chinese Indians Others N total (%) Patients status at 6-month post-hospital discharge Relapsers 27 (61.4%) 10 (50%) 4 (30.8%) 1 (33.3%) 42 (52.5%) Non-Relapsers 17 (38.6%) 10 (50%) 9 (69.2%) 2 (66.7%) 38 (47.5%) Relatives EE at patients initial psychiatric hospitalization High-EE 21 (47.7%) 10 (50%) 5 (38.5%) 1 (33.3%) 37 (46.2%) Low-EE 23 (52.3%) 10 (50%) 8 (61.5%) 2 (66.7%) 43 (53.8%) High-CC 15 (34.1%) 9 (45%) 3 (23.1%) 1 (33.3%) 28 (35%) Low-CC 29 (65.9%) 11 (55%) 10 (76.9%) 2 (66.7%) 52 (65%) High-EOI 10 (22.7%) 2 (10%) 4 (30.8%) 0 (0%) 16 (20%) Low-EOI 34 (77.3%) 18 (90%) 9 (69.2%) 3 (100%) 64 (80%) logistic regression analysis was performed to assess the predictive value of relatives CC and EOI on patients relapse processes. The model with CC and EOI scales yielded the likelihood ratio χ 2 (2) = 9.32, p = 0.009, Nagelkerke R 2 = 0.15, suggesting that the model with relatives CC and EOI components fits significantly better than the constant-only model. The addition of CC and EOI into the model also indicated an improvement in model fit, Hosmer and Lemeshow χ 2 (8) = 6.24, p = 0.62, suggesting a model with good predictive value. This model also produced an overall correct classification value of 62.5%, which was an improvement from the constant-only model (52.5%). Of the correctly classified cases, 57.9% was for the prediction of non-relapse cases and 66.7% for the prediction of relapse cases. The contribution of CC and EOI in predicting patients relapse is shown in Table 4. Of the two EE components, relatives CC level turned out to be the significant contributor to patients relapse, Wald χ 2 = 7.69, p = 0.006. For every one unit increase in relatives CC level, the odds for patients to relapse were increased by a factor of 1.12 (95% CI: 1.03 1.18). In contrast, relatives EOI level did not significantly predict patients relapse. 3.4. The predictive value of relatives personality traits on schizophrenia relapse Another logistic regression was also performed to predict patients relapse as a function of relatives personality traits. The model with three relatives personality traits yielded a likelihood ratio χ 2 (4) = 9.90, p = 0.019, Nagelkerke R 2 = 0.27, suggesting that the model as a whole fits significantly better than the constant-only model. The addition of three personality traits into the model also indicated an improvement in model fit, Hosmer and Lemeshow χ 2 (8) = 10.47, p = 0.23, which suggested a model with good predictive value. Of the three relatives personality traits, extraversion turned out to be the significant predictor to patients relapse, Wald χ 2 = 6.99, p = 0.008. The inverse relationship between extraversion and relapse indicated that for every one unit decrease in relatives extraversion trait the odds for patients to relapse were increased by 23% (95% CI: 0.64 0.94). Due to poor alpha value of the psychoticism trait, another logistic regression was performed with psychoticism score excluded from the analysis. The model with extraversion and neuroticism traits yielded a slightly better likelihood ratio of χ 2 (4) = 9.37, p = 0.009, Nagelkerke R 2 = 0.15, when compared to the earlier one. This also suggested that the model as a whole fits significantly better than the constant-only model. Again, the inclusion of only two personality traits into the model indicated an improvement in model fit, Hosmer and Lemeshow χ 2 (8) = 10.71, p = 0.22, which suggested a model with good predictive value. Results demonstrated a slight improvement in the strength of relatives extraversion trait in predicting patients relapse Wald χ 2 = 7.19, p = 0.007. Again, the inverse relationship between the two variables suggested that for every one unit decrease in relatives extraversion trait the odds for patients to relapse were increased by 23% (95% CI: 0.64 0.93). Table 5 summarizes the contribution of extraversion and neuroticism traits in predicting patients relapse. 4. Discussions 4.1. Main findings The present study revealed novel and interesting findings with regard to the predictors of relapse among schizophrenia patients in Kuala Lumpur, Malaysia. Results demonstrated that the odds for patients to relapse 6 months post-hospital discharge were increased by 8.37 times when their relatives demonstrated an increase in high-ee level. In addition, the Table 2 Correlation matrix between all variables scored by relatives on EPQ-R and FQ (n=80). Relatives N P E CC EOI Neuroticism (N) 1 0.213 0.28 0.27 0.063 Psychoticism (P) 1 0.008 0.049 0.018 Extraversion (E) 1 0.28 0.022 CC 1 0.21 EOI 1 pb0.05.

194 A. Roseliza-Murni et al. / Comprehensive Psychiatry 55 (2014) 188 198 Table 3 Logistic regression analysis of patients relapse as a function of relatives EE index. Variables B Wald χ 2 test Odd ratio 95% CI for odd ratio Sig. Lower Upper High-EE 2.12 16.73 8.37 3.02 23.15.000 (Constant).84 6.34.43.012 pb.001. odds for patients to relapse were also found to increase by 12% when relatives demonstrated an increase in their CC level. These results indeed replicated the worldwide findings on the significant role of relatives high-ee level [9,11,13,18 23,38] particularly the CC component on schizophrenia relapse processes [21,39 41]. In fact, the eight-fold increase in relapse among schizophrenia patients in Malaysia exceeded the documented 3.7 times aggregated relapse rate as reported in the Western findings [13], but lower than the 10 times more frequent rate of relapse among the Belgrade sample [11] when relatives demonstrated high- EE attitudes. Additionally, the significant role of CC in contributing towards patients relapse processes also supported the postulation that critical comments account for the vast majority of what is really being measured by the construct of expressed emotion [39, p. 10]. Likewise, Hogarty et al., [42] also suggested that the most contribution of EE-relapse association comes from CC rather than EOI. Even though only 46.2% of the relatives demonstrated high-ee attitudes and 35% of them were categorized into the high-cc group, this study was able to demonstrate the predictive value of high-ee and CC component on schizophrenia relapse. The present finding was clearly in contrast to the suggestions made by the authors of the earlier Malaysian study, in which they claimed that relatives EE did not predict schizophrenia relapse among the patients in the states of Kelantan and Terengganu [24]. One of the explanations to the contrasting results was the fact that Azhar and Varma [23] did not actually conduct any analysis on the predictive role of relatives EE and schizophrenia relapse but only concluded such a relationship existed based on a large proportion of their study sample who demonstrated low-ee level. The difference in the rate of high-ee among relatives in the present study and that of Azhar and Varma [24] could possibly be explained by the role of urbanization and modernization in shaping relatives attitudes, values and patterns of communication. Modernization and the influence of western culture can be readily observed in the urban and metropolitan multi-racial areas of Kuala Lumpur and Klang Valley whereby people tended to be less cohesive and more individualistic. This condition is rarely observed in Kelantan and Terengganu, the two agrarian and less industrialized states within Malaysia which are predominantly populated by the traditional Malays. In general, people of Kelantan and Terengganu tended to adopt a more conservative traditional Malay culture that placed high values on social cohesion and collectivist community, which prioritized group over personal interests [44]. Research on traditional Malay communication styles by Lailawati [43] indicated that they are generally indirect when making a request as it might appear impolite to do so outright, hence the traditional Malays talk around what they intend to convey in the hope that their message is understood [43, p. 7]. The modern urban Malays on the other hand, are more explicit whereby their interaction styles seems to be more direct in order to achieve goals rather than to build relationships. [43, p.9]. The interaction patterns of the modern Malays are influenced more by objectivity rather than the old pattern of subjectivity, and they were said to prefer rationality over emotionality [43]. The difference in communication patterns among the modern urban residents to those of their counterparts in rural traditional areas could possibly explain the difference in EE patterns among the present sample and that of Azhar and Varma [24]. Parallel to the changes within Malay culture, Malaysia has also observed changes within its multi-racial group culture. The mixing, blending and inter-cultural marriages among the three main ethnic groups since the 19th century, have resulted in the existence of a new culture [43] known as the contemporary Malaysians. This new culture provided the ground for national unity and it could be commonly observed among the multi-racial residents of Kuala Lumpur and Klang Valley. It is possible that the Table 4 Logistic regression analysis of patients relapse as a function of relatives EE components (CC and EOI). Variables B Wald χ 2 test Odd ratio 95% CI for odd ratio Sig. Lower Upper CC.095 7.69 1.12 1.03 1.18.006 EOI.008.06 1.01.94 1.08.81 (Constant) 1.76 4.23.17.04 pb.05.

A. Roseliza-Murni et al. / Comprehensive Psychiatry 55 (2014) 188 198 195 Table 5 Logistic regression analysis of patients relapse as a function of two relatives personality traits. Variables B Wald χ 2 test Odd ratio 95% CI for odd ratio Sig. Lower Upper Extraversion.261 7.19.77.64.93.007 Neuroticism.035.211 1.04.90.89 1.20 (Constant) 2.11 4.51 8.28.034 pb.05. new culture may have some influence over the nonsignificant difference in the expression of EE among the three ethnic groups. Additionally, past study also suggested that when a society moves toward westernization, it is also expected that EE would play a larger role in determining the relapse processes [44]. When a society undergoes westernization it also embraces the living arrangement that comprises of a nuclear family unit, and this is commonly seen among residents of Kuala Lumpur and Klang Valley, whereby their living arrangements usually comprised of one or two parents and one or two children [45]. On the contrary, the people of Kelantan and Terengganu still practice and value the living arrangement that comprises of extended family or joint family units, whereby many relatives live nearby in the same village (kampung) or even share the same housing compound. It has been suggested that families consisting of small unit of members or single relative may experience greater burden than the extended families as those in the extended families can share their stresses and difficulties of dealing with the patients [46]. Another study also suggested that family members of the extended families tended to have more positive attitudes toward their mentally ill relatives than those from nuclear family units. The reason being, the more family members live together, the less likely the burden of care for an ill family member will fall on a small unit of people, such as in the nuclear family unit. This is then presumed to have a more positive family environment [47]. It is therefore possible that the stresses of living in urban metropolitan areas, which also provided limited family resources or supports in the form of extended/joint family members, influenced the difference in high-ee rates and the difference in the predictive role of EE in relapse processes among participants in the present study and those of Azhar and Varma's [24]. Another possible explanation to the differences in the current findings and those of Azhar and Varma's [24] was that their study used an unknown psychometric 30-item scale, whereas the present study used the FQ which had good reliability scores and had been suggested to have high agreement with CFI [30] the measure which is often used for EE assessment. The higher rate of high-ee index among the urban sample as opposed to the rural sample has also been observed in the study on an Indian sample. Wig et al. [48] conducted a study on relatives of schizophrenia patients in two different environment Chandigarh, a modern urban city and Ambala, a rural village. They found that 30% of the relatives living in the urban area demonstrated high-ee level while only 8% of the relatives in rural traditional area demonstrated high-ee attitudes (p b 0.01). Again, one possible explanation to this finding may be differences in life-style, in which relatives in rural area tended to live in the extended families while their counterparts in the urban area tended to be nuclear families. According to Wig et al. [48, p.165], it is conceivable that the emotions generated by family relationships become dispersed throughout the network of an extended family, instead of being concentrated in the much smaller group of a nuclear family. On the same note, Waxler [49] also argued that relatives living in rural areas tended to have lower expectations and higher tolerance towards schizophrenia patients than those relatives living in urban areas, thus explaining the lower expressions of negative emotions among relatives living in rural traditional environment. The insignificance of relatives high-eoi level in predicting schizophrenia relapse can be explained by a large proportion of relatives (80%) who were classified into the low-eoi category. While there were the remaining 20% of relatives classified into the high-eoi category, it still was not able to predict schizophrenia relapse. This finding is consistent with previous studies conducted in Chandigarh, Iran and Hong Kong [50,51,41]. Only 4% of the relatives in Chandigarh demonstrated high-eoi attitudes and this rate was considered too small to demonstrate any meaningful relationship with schizophrenia relapse [50]. Even though 50% of the relatives in Iran demonstrated high-eoi attitude, again this component was not able to predict schizophrenia relapse [51]. According to Kalafi and Torabi [52] within the context of Iran, mothers who are often the primary caregivers tended to demonstrate more tolerance, over-protective and self-sacrificing behavior towards the patients. These mothers may even encourage dependency and child-like attitudes in their adult offspring with schizophrenia [52]. Among the Chandigarh relatives Wig et al. [48] observed that they tended to be more tolerant towards the disturbance and dysfunctions that were related to schizophrenia. The inability of EOI component to predict schizophrenia relapse seemed to be in contrast to the findings in the Western culture [17 23]. According to Jenkins and Karno [39, p.16], culture, as a system of shared meanings and symbols, offers the most powerful explanation for the observed variations in expressed emotion in different

196 A. Roseliza-Murni et al. / Comprehensive Psychiatry 55 (2014) 188 198 population. Thus, it is possible that culture may play some important roles in the non-significant contribution of EOI on schizophrenia relapse processes among the present sample. It is possible that within the context of the Eastern culture which value inter-dependency among family members, the expression of overly-protective and selfsacrificing attitudes may reflect the concerns and caring dispositions of the relatives towards an ill family member [44]. Since there was evidence of cross-cultural difference with regard to the level of tolerance among relatives toward patients with schizophrenia [53], it is also possible for the different components of EE to have different impacts on patients relapse processes. Thus, high-eoi attitudes may not be stressful to the patients after all, instead it may have some comforting properties. More studies therefore, are needed to investigate the different roles of EE components on schizophrenia relapse. The novelty of the present study has been highlighted in the significant role of relatives personality traits as a direct predictor to patients relapse. Unlike previous studies that looked into the role of relatives personality traits as the contributory factor towards their EE levels [15,25], the present findings demonstrated the direct link between relatives personality traits and patients relapse processes. The significant Wald χ 2 of 7.19 (p = 0.007) demonstrated evidence of a strong relationship between relatives extraversion trait and patients relapse processes. The observed inverse relationship between these two variables suggested that the odds for patients to relapse were increased by 23% when relatives demonstrated a one unit decrease in their extraversion traits. On the contrary, this finding also denotes that relatives who score high on extraversion trait may contribute to a lesser rate of relapse among the patients. According to Rawlings [54] the extraversion scale measures different degrees of sociability with high scores indicating extraversion and low scores indicating introversion. The extravert has been described as being carefree, easy-going, optimistic, and likes to laugh and be merry. On the other hand, the typical introvert is described as reserved and distant except to intimate friends somewhat pessimistic, and places great value on ethical standards [55, p. 9]. Costa and McCrae [56] further suggested that the extraversion trait influenced a person's positive emotion and contributed greatly to their well-being. Likewise, Smith [57] suggested that happy people were those who were sociable and outgoing. Thus, it is possible that the characteristics of the extraverts influenced their daily interactions with the patients, which may subsequently act as a buffer against the relapse processes. On the contrary, the characteristics of the introverts may create a stressful household environment for patients with schizophrenia that subsequently contribute to their relapse processes. Eysenck and Eysenck's [58] hierarchical structure of personality traits, further suggested that apart from being sociable, the extraverts also tended to be lively, active, assertive, sensation-seekers, dominant, surgent and venturesome. Even though the extraverts may be talkative and tended to express their emotions to others more readily, their emotional expression may not necessarily be in the form of criticisms. In fact, the characteristics of the extraverts seemed to equate those of low-ee relatives which were described by Leff and Vaughn as characterized by the cool, easy going and flexible attitudes [17]. Hence, the inverse relationship between the extraversion trait and CC component. On the other hand, the no correlation between extraversion and EOI component was understandable as the extraverts did not seem to possess the characteristics of high-eoi relatives, in which they had been described as demonstrating exaggerated emotional responses, over-intrusive and self-sacrificing behaviors as well as over-identification with the patients [8]. In contrast to the extraversion, both psychoticism and neuroticism traits were not able to predict schizophrenia relapse. With regard to the psychoticism trait, it is possible that its low alpha value contributed to this finding. Furthermore, the scores for psychoticism trait between relatives of patients who relapsed and those who did not also indicated no significant difference (t[78] = 0.598, p = 0.55). Likewise, both groups of relatives also demonstrated a non-significant difference with regard to neuroticism trait (t[78] = 1.18, p = 0.24). It is therefore possible that the nonsignificant difference of the psychoticism and neuroticism traits between the two relative groups influence the nonsignificant contributions of these two personality traits on schizophrenia relapse. In sum, the present study demonstrated that relatives high-ee level particularly the CC scale contributed significantly to the prediction of schizophrenia relapse 6 months post-hospital discharge. It is also intriguing to note that relatives personality traits particularly the degree of extraversion-introversion traits play a significant role in the relapse processes among schizophrenia patients in Malaysia. The findings thus, demonstrated that relatives expressed emotion and personality traits indeed matter in predicting the course and outcome of schizophrenia, and therefore should be considered when designing intervention strategies. 4.2. Strengths and limitations A striking strength of this study is its contribution to the body of knowledge with regard to the significant direct role of relatives extraversion traits on patients relapse processes. To the best of our knowledge, the limited numbers of previous studies on relatives personality traits were mainly focused on its contribution to relatives EE index and components [15,25] but none has focused on patients relapse. The lack of study on this issue certainly calls for further investigations. Furthermore, the significant contributions of relatives high-ee attitudes particularly the CC components in the course of schizophrenia implied the serious need to incorporate the psychosocial-based intervention strategies into the treatment protocol. This is yet to be implemented as