Correlates of the Affective Impact of Auditory Hallucinations in Psychotic Disorders

Transcription

1 Correlates of the Affective Impact of Auditory Hallucinations in Psychotic Disorders by David L. Copolov, Andrew Mackinnon, and Tom Trauer Abstract While many who hear auditory hallucinations (AHs) experience them as unpleasant, some do not Little is known about the correlates of AHs that are not unpleasant, or of the characteristics of those who hear them. To better understand this symptom, we used a comprehensive structured interview schedule to study 199 subjects who had experienced AHs. Subjects' responses to AHs were combined into two indexes: one assessing total affective impact and the other assessing the affective direction (positive or negative). Subjects who had grandiose delusions experienced their AHs more positively. AHs that were more frequent, lasted longer, and were louder were experienced more negatively. AHs heard in the second person and those related to people with whom the subjects had personal relationships were more positive than those heard in the third person. Many other aspects of AHs were unrelated to total affective impact or direction. It is argued that the positive evaluation of voices by subjects requires greater attention than it has received previously. Implications for assessment, clinical practice, and research are discussed. Keywords: Auditory hallucinations, psychosis, affect Schizophrenia Bulletin, 30(1): ,2004. Auditory hallucinations are a frequent and prominent component of psychotic disorders (Lowe 1973; Slade and Bentall 1988) and are uncommon but not rare in community samples (Tien 1991). Whether or not AHs occur in persons diagnosed with a mental illness, the dominant professional view is that the AH is a psychological abnormality and that the "symptoms of an illness are usually assumed to be undesirable" (Miller et al. 1993, p. 587). While the majority of patients experiencing AHs describe them in negative terms, in some cases the reactions are neutral or even pleasant (Hamilton 1976, p. 49). Of the 22 patients studied by Silva and Lopez de Silva (1976), about one-third described their hallucinations as pleasant. Sixteen of O'Sullivan's (1994) 40 patients in treatment sometimes experienced their voices as pleasant (three patients always so), and five never experienced them as unpleasant. Just over half of the patients studied by Miller et al. (1993) reported some positive effect of hallucinations. Similar findings were reported by Oulis et al. (1995): 53 percent of the 60 patients in this study described their voices as hostile and 15 percent as friendly. Nayani and David (1996) reported their sample as being evenly split between those with more and those with less distress associated with AHs. Of the 30 persons with AHs studied by Close and Garety (1998), none reported their voices as having only positive content, but 17 reported both positive and negative content. The studies of the emotional or evaluative impact of AHs suggest the following points. First, while in most studies most respondents describe their voices as unpleasant, there are sizable minorities for whom the voices are either not unpleasant or pleasant. Authors have tended to summarize their results in terms of the majority findings; for example, Oulis et al. concluded that "usually their [voices'] content is hostile to the patient" (1995, p. 100) on the basis of 13 of their 25 patients. Second, in many studies the emotional impact of AHs on their hearers is assessed in a very simple fashion, often by means of a single question (e.g., Lowe 1973). Typically, little regard is paid to the complexity of emotional response or to the possibility of mixed reactions. Third, most studies have had modest numbers of subjects, typically 60 or less, with only Nayani and David (1996), Oulis et al. (1997), and Carter et al. (1996) having as many as 100. Inferences and generalizations based on small samples are weakened by the instability of the results. Send reprint requests to Prof. D. Copolov, Mental Health Research Institute of Victoria, Locked Bag 11, Parkville Victoria 3052, Australia. dlc@mhri.edu.au. 163

2 Schizophrenia Bulletin, Vol. 30, No. 1, 2004 D.L. Copolov et al. The current study examined the emotional impact of AHs. This is important for several reasons. First, as noted by Miller et al. (1993, p. 588), "Routinely asking patients whether their hallucinations are helpful... might identify fruitful areas for psychosocial intervention, so that the needs met by hallucinations could be met by other means." A related and second reason is that treatment and management approaches for patients who do not find their AHs disturbing will presumably be different from approaches for those who do; psychological therapies for psychotic symptoms must be tailored to the individual needs of the patient, just as pharmacotherapies are (Shergill et al. 1998). Third, as Miller et al. (1993) also noted, lack of distress or even positive attitudes toward AHs might reduce the patient's incentive to comply with medication. Using a sample of patients who responded to a comprehensive interview about their AHs (the Mental Health Research Institute Unusual Perceptions Schedule [MUPS], Carter et al. 1995), we developed indexes of AHs' emotional impact on the individual. The structural and other correlates of emotional impact were explored. Method Patients. One hundred and ninety-nine patients were interviewed using the MUPS; 134 (67.3%) were male, and the mean age was 32.7 years (standard deviation [SD] = 10.67, range 15-63). Most (69.3%) had never been married (the marital status of two was unknown), and most (84.9%) were bom in an English-speaking country. Of the 194 whose educational background was known, 8 (4.1%) were educated to primary school level or less, 137 (70.6%) were educated to secondary school level, 39 (20.1%) had tertiary education experience, and 10 (5.2%) had obtained a trade or technical qualification. Seventy-three (36.7%) lived in their parental family home, 64 (32.2%) lived in their own home, 17 (8.5%) lived in a boarding house or a private hotel, 14 (7.0%) lived in a group home, 9 (4.5%) lived in sheltered accommodation, and the accommodation of a further 22 (11.1%) was other or not known. Of the 198 for whom the information was available, 169 (85.4%) were unemployed or living on a disability pension. Nine (4.5%) were in professional or paraprofessional employment, 14 (7.1%) were tradespersons or clerks, and 5 (2.5%) were laborers or did related work. The main recruitment sources were the inpatient wards of a psychiatric hospital (126 [63.6%]), a clubhouse for the mentally ill (17 [8.6%]), private psychiatrists (15 [7.6%]), or community mental health services (15 [7.6%]). Diagnoses were made by administering the relevant sections of the Structured Clinical Interview for DSM-III-R (Spitzer et al. 1990); thus, the diagnoses relate to the time of the MUPS interview, whereas the AHs being reported could be from the past. Schizophrenia was the most common principal psychiatric diagnosis (161 [80.9%]). A further 27 (13.6%) had an affective psychosis, 6 (3.0%) had some other nonorganic psychosis, and 5 had a borderline personality disorder (BPD). For the 191 subjects for whom information was available, 16 (8.4%) were not prescribed antipsychotic medication. For the remaining 175 (91.6%), the mean daily dose of antipsychotic medication, in chlorpromazine equivalents, was 474 mg (SD = 286.6, range 25-1,500). The mean age of onset of AHs was 23.2 years, or 9.5 years prior to the MUPS interview. MUPS. The MUPS is a semistructured interview that investigates many aspects of AHs. Its 365 questions are organized into seven main sections: physical characteristics, personal characteristics, relationship and emotive aspects, form and content, cognitive processes, personal perceptions, and psychosocial issues. Provision is made for the interviewer to record relevant judgments of the patient and the interview process. Response formats (i.e., fixed or free) vary according to the question being asked. The content, reliability, and acceptability of the MUPS has been reported previously (Carter et al. 1995). Carter et al. (1995) also provide a more detailed description of the areas covered by the MUPS. Once informed consent was obtained, the full MUPS was administered by a professional who had been trained in the use of the instrument. Where necessary, the interview was administered over a number of sessions to avoid fatigue. Results Development of Indexes of Total Affectivity and Affective Direction. To study the emotional impact of AHs, it was necessary to develop suitable indexes from the MUPS interview data. Three sections of the interview protocol are relevant to this: tone, content, and feelings. In each of these sections, the patient is presented with a number of adjectives on cards and asked to select however many apply. The tone section comprised 16 adjectives, the content section 13, and the feelings section 21 (table 1). These adjective items were subjected to factor analysis using Mplus 2.01 (Muthdn and Muthdn 1998). This program is suitable for the factor analysis of dichotomous (yes/no) items. A number of goodness of fit indexes were calculated. These included a chi-square test, the nonnormed fit index (NNFI; Bentler and Bonnett 1980), the comparative fit index (CFI; Bentler 1990), and the root mean square error of approximation (RMSEA; Browne and Cudeck 1992). Values close to 1.00 are desirable on 164

3 Correlates of the Affective Impact of Auditory Hallucinations Schizophrenia Bulletin, Vol. 30, No. 1, 2004 Table 1. Tone, content, and feeling descriptors and loadings on factors Tone Content Feeling Positive Negative Adjective Loading Adjective Loading Gentle Loving Kind Friendly Quiet Helpful Guiding Affirming Changeable Inspiring Comforted Not alone anymore Reassured Excited Unconcerned Inspired Happy Harsh Angry Crackly Authoritative Bossy Malicious/nasty Muted Muffled Indistinct/fuzzy Sharp Menacing Persecutory Abusive/insulting Obscene Intrusive Derogatory ("put-down") Accusatory Threatening Critical Terrified Irritated Sad Helpless Hopeless Angry Anxious Agitated Depressed Intruded upon Overwhelmed Frightened Out of control Confused the NNFI and the CFI. Browne and Cudeck (1992) suggest that values of the RMSEA below 0.08 indicate acceptable model fit. Following exploratory analyses of the correlation matrix, a second order factor model was fitted to the data. Separate factors were defined for the positive and negative items in each domain of responses (tone, content, feelings). Factors relating to negative responses were defined as loading on a second order general negative factor, while the positive response factors loaded on a second order general positive factor. Overall, this model provided an excellent fit to the data (x 2 = , df = 117, x 2 /df = 1.69, NNFI = 0.95, CFI = 0.93, RMSEA = 0.059). All loadings were statistically significant. With the exception of three items on the negative tone factor (muted, muffled, and indistinct/fuzzy), all first order loadings were substantial (mean = 0.74, range ). Loadings of each domain on the general positive and negative factors were extremely high. A model in which correlations within domains (e.g., positive tone with negative tone) were permitted yielded only negligible, nonsignificant values. The correlation between the two second order factors (r = 0.05) was negligible in magnitude and nonsignificant. Loadings of adjectives on first order factors, and of first order factors on the two second order factors, are presented in table 2. This pattern of results suggested that the structure of the items could be conceptualized as two independent dimensions with no differentiation due to the domain of the response. Indeed, an orthogonal two-factor model fitted the data very well (x 2 = , df= 83, x 2 ldf= 1.88, NNFI = 0.94, CFI = 0.93, RMSEA = 0.067). This sup- 165

4 Schizophrenia Bulletin, Vol. 30, No. 1, 2004 D.L. Copolov et al. Table 2. Loadings of First Order Factors on Second Order Factors First order factor Tone Content Feelings Second Order Factor Positive Negative ported the construction of two scales: a positive response scale (PRS) comprising the 17 positive items across the domains of tone, content, and feeling; and a negative response scale (NRS) comprising 30 negative items. The three items with loadings below 0.4 were omitted. The modest loadings of these items are consistent with the relatively neutral valence of the items. The PRS and NRS raw scores were divided by 17 and 30, respectively, to place them on the same scale, and multiplied by 100 to express them as percentages. Reflecting the factor structure of the items, the two scales had high reliability (Cronbach's alpha was 0.90 and 0.91 for the PRS and NRS, respectively). These two scales were combined to produce two indexes: one of the direction or valency of affect, which represents the degree of positivity or negativity, and the other of the strength of affective response. This was informed by the approach taken in the scoring of the Bradburn Affect Balance Scale (Bradbum 1969), where affect balance is represented by the difference between positive and negative mood states. A psychometric justification for adding or subtracting scores on orthogonal factors is presented by Van Schuur and Kruijtbosch (1995). An affective direction index (AFFD) was derived by subtracting the NRS score from the PRS score, and a total affectivity index (TAFF) was derived by adding the two response scale scores and dividing by two. These indexes could be computed for 196 patients with complete data. The mean of the AFFD scores was (SD = 38.5, range ), indicating that on average the affective evaluation of the AHs was negative. The mean TAFF score was 44.8 (SD = 20.2, range ), indicating that on average about half the adjectival items were endorsed, but with a wide degree of variation. The distributions of both AFFD and TAFF were bell shaped, and for both the measures of skewness and kurtosis were nonsignificant at the 0.05 level. The correlation of 0.11 between AFFD and TAFF was nonsignificant (p ). As part of the MUPS interview, patients were asked, "Do your voices/the sounds upset (worry) you?" Response choices were as follows: no, a bit, moderately, and a lot. Of 194 patients answering this question, the number giving each of these answers was 33, 12, 34, and 115. The mean TAFF scores of these four groups were not significantly different (F = 2.1, df = 3, 190, p = 0.09), but the TAFF scores of those reporting no worry was significantly lower than of those reporting some degree of worry (/ = 2.05, df = 192, p = 0.04). The mean AFFD scores for the four response categories were 27.0, 8.8, -20.5, and -33.5, which are highly significantly different (F = 35.9, df= 3, 190, p < ) and show a clear trend for increasing negativity with increasing worry (the correlation between AFFD scores and the four response choices was 0.60). Post hoc testing of pairs of means (Scheff6's tests) showed that all differences except between "no worry" and "a bit," and between "moderately" and "a lot," were significant at the 0.05 level or beyond. The close correspondence between the TAFF and AFFD scores and the responses to the question about how upsetting or worrying the AHs were supports their validity. Relationship Between AFFD and TAFF Scores and Demographic Characteristics. The relationships between mean AFFD and TAFF scores and the patient's age, sex, marital status, and level of education; whether the patient was bom in an English-speaking country; and whether the patient was working, unemployed, or on a disability pension were all not statistically significant. Relationship Between Emotional Impact Scores and Diagnostic and Treatment Characteristics. All 196 patients could be allocated to one of four diagnostic groups: schizophrenia 158, affective disorder 27, other nonorganic psychosis 6, and BPD 5. The differences in mean TAFF scores were very small and nonsignificant. While the mean AFFD scores of all groups were negative, indicating an overall negative evaluation of their AHs, those of subjects with BPD were much lower. Post hoc tests revealed that the differences between the BPD and schizophrenia and affective disorders groups were significant (table 3). Interviewers identified the presence of delusions and classified them according to a list (e.g., persecutory, guilt, catastrophe). Almost all patients (95.0%) were identified as having delusions, and for many of the delusion types, there was no significant association with TAFF or AFFD scores, using a Bonferroni-adjusted alpha level of However, significant differences were found between those with and without delusions of the grandiose (special mission type) on AFFD and TAFF (tables 3 and 4). Similar differences were found for delusions of the grandiose (special powers type) and were significant at the level. Significantly higher TAFF scores were obtained by patients rated as having delusions of the passivity type (Schneider 1959). Specifically, TAFF scores 166

5 Correlates of the Affective Impact of Auditory Hallucinations Schizophrenia Bulletin, Vol. 30, No. 1, 2004 Table 3. Analyses of affect balance scores (AFFD) as a function of diagnosis and hallucination characteristics Comparison Diagnosis or symptom Borderline personality disorder vs. schizophrenia Borderline personality disorder vs. affective disorder Grandiose delusion (special mission) vs. not Grandiose delusion (special powers) vs. not Hallucination characteristic Frequency Duration Loudness AHs in the second person vs. not AHs conversing about themselves in relation to patient AH message/content linked to someone influential AH possibly own voice or thoughts Note. AH = auditory hallucination. Test statistic Scheffe's test Scheffe's test /(182) =3.87 /(187) = 2.51 F(3,192) = 3.17 F(4,182)=4.97 F(2,144) = ) = 2.46 /(191) = ) = ) = Table 4. Analyses of total affect scores (TAFF) as a function of diagnosis and hallucination characteristics Comparison Diagnosis or symptom Grandiose delusion (special mission) vs. not Grandiose delusion (special powers) vs. not Delusion of being controlled Mind reading Thought broadcast Thought insertion Thought withdrawal Hallucination characteristic Frequency AHs in the first person vs. not AHs in the second person vs. not AHs in the third person vs. not Note. AH = auditory hallucination. were significantly higher for those with delusions of being controlled, mind reading, thought broadcast, thought insertion, and thought withdrawal (table 4). AFFD scores did not differ significantly for those with and without these forms of delusions. No association was found between TAFF and AFFD scores and antipsychotic medication dose, expressed in chlorpromazine equivalents. The correlation between duration of illness and AFFD was close to zero, but the correlation of 0.15 with TAFF was significant (p = 0.04). Relationship Between TAFF and AFFD Scores and Frequency, Duration, and Loudness. In response to the question "During the last episode with the voices/sounds, how frequently were they with you?" 19 (9.7%) said rarely, 30 (15.3%) said occasionally, 51 (26.0%) said often, and 96 (49.0%) said constantly. The mean TAFF Test statistic /(182) = ) = ) = ) = ) = ) = ) = 3.35 F(3,192) = ) = ) = ) = and AFFD scores were significantly different (tables 3 and 4). Specifically, those who hear their voices constantly feel more strongly about them, and those who hear them often to constantly feel more negatively about them. Patients were asked to indicate the duration of their AHs. Because the distribution of durations was highly skewed and noncontinuous, they have been grouped thus: up to 1 minute, 2 to 10 minutes, 15 to 63 minutes, 90 minutes to 8 hours, and 12 hours or more. Of the 190 patients with valid responses, 26 (13.7%) fell into the first group, 40 (21.0%) into the second, 22 (11.6%) into the third, 63 (33.2%) into the fourth, and 39 (20.5%) into the fifth. The respective TAFF scores were not significantly different, but the AFFD scores were (table 3). Thus, longer durations of AHs were associated with more negative, but not stronger, feelings about them. Sixty-three patients (32.1%) heard their AHs as soft 167

6 Schizophrenia Bulletin, Vol. 30, No. 1, 2004 D.L. Copolov et al. (too faint, whisper, or soft), 69 (32.5%) heard them at normal loudness, and 55 (28.1%) heard them as loud (loud or yelling/screaming); the remainder (4.6%) either did not respond to this question or gave other responses. The mean TAFF scores of these three groups did not differ significantly, but the AFFD scores did (table 3). Thus, louder AHs were associated with more negative, but not stronger, feelings about them. Relationship Between TAFF and AFFD Scores and Patients' Relationships to the AHs. Patients indicated whether their voices addressed them in the first, second, or third person, or impersonally. Given that many patients heard several voices, patients could say yes or no to each of these four forms of address. Patients obtained significantly higher TAFF scores for AHs that addressed them in the first, second, and third person but only had more negative AFFD scores in relation to AHs that addressed them in the second person. Voices talking about themselves in relation to the patient (e.g., We are normal, you are mad) were associated with more negative AFFD scores (table 3). Thus, two questions relating to different forms of address in the second person show that such address is perceived more negatively. To the question "Is it possible that the idea behind the message/content of the voices is linked or connected to someone who is or was influential in your life?' 64 (32.7%) said no and 109 (55.6%) said yes; the rest (11.7%) gave intermediate or no replies. The mean AFFD score of those saying no was significantly lower than that of those saying yes (table 3). Patients were asked "Is it possible that it was actually your own voice and thoughts you heard?"; 99 said no and 67 said yes (30 gave other or no responses). Differences between these two groups on TAFF were small and nonsignificant, but the mean AFFD scores of those answering yes were higher than those of subjects answering no, the difference being nearly significant (table 3), suggesting a tendency for these kinds of voices to be perceived more positively. Lack of Relationship Between TAFF and AFFD Scores and Other Aspects of AHs. There were no statistically significant relationships or associations between TAFF and AFFD scores and the following aspects of AHs: time of day; whether the voices were those of adults or children; handedness; hearing AHs inside or outside the head; hearing AHs on the left or on the right; how real the voices seemed; similarity of the mood of the voice to the patient's own mood at the time when heard; or interviewer ratings of insight, attitude, or rapport Discussion Several evaluative aspects of AHs (content, tone, and feelings evoked) showed a strong, consistent, and understandable relationship, such that they could be combined to produce indexes of amount and direction of affective or evaluative tone. Advantages of this approach are increased sensitivity and reliability over assessment by a single item, and incorporation of a wider range of evaluative aspects. Comparing these indexes with a simple four-point graded scale of distress, we found that those who indicated some degree of worry had stronger feelings about their AHs, and the greater the worry, the greater the negativity with which they were evaluated. About 75 percent of patients described being distressed "moderately" or "a lot" and, on average, they obtained clearly negative TAFF scores. These percentages are similar to those obtained by Romme et al. (1992) in the nonclinical community sample of persons experiencing AHs. The degree of correlation (0.6) between the negative evaluation of AHs and the degree of worry they produce, while substantial, indicates that it cannot be assumed that unpleasant AHs are necessarily associated with distress, and this should be reflected in the way that AHs are examined clinically. In addition, more allowance needs to be made in clinical and research inquiry for pleasant AHs. Most rating scales are heavily weighted to the negative, suggesting an expectation bias. The "distress" question offered the choices of no worry, "a bit," "moderately," and "a lot" A more balanced form of inquiry might encourage a wider range of description, including positive aspects. Strong and significant associations were found between affect balance (AFFD) scores and the frequency, duration, and volume of die AHs, suggesting some form of dose-response relationship. Given that most patients perceived their AHs negatively, we may say only that infrequent, brief, and quiet AHs tended to be less negative than those that were frequent, long, and loud. While the cross-sectional nature of the data does not allow us to draw firm causal inferences, it is likely that the increased salience of AHs contributes to their affective quality. Most therapeutic interventions aim to affect frequency directly; duration may be affected indirectly, and little attempt is made to influence volume. A number of treatment approaches have been shown to reduce frequency, at least temporarily (e.g., Slade 1974; Haddock et al. 1998), and a few have shown simultaneous or differential impact on frequency and duration. Allen et al. (1985) found that diversionary methods reduced frequency and removal methods reduced duration, while Mclnnis and Marks (1990) found that an audiotape-based treatment of a single case reduced duration but not frequency of persistent second-person AHs. Done et al. (1986), also using a single case study design, found that die wearing of an earplug in the dominant ear reduced both the frequency and volume of AHs. Shergill et al. (1998, p. 137), reviewing psycho- 168

7 Correlates of the Affective Impact of Auditory Hallucinations Schizophrenia Bulletin, Vol. 30, No. 1, 2004 logical approaches to AHs, concluded, "Almost all the strategies produced some benefit to some patients; treatment often improved AH-associated distress, rather than frequency of AH." The finding that the AHs of patients with grandiose delusions were more pleasant than those of patients without such delusions is consistent with the established association between grandiosity and mania (APA 1994). Patients with passivity delusions felt more strongly about their AHs but not consistently more positively or negatively. This last result is consistent with the findings of Blakemore et al. (2000), who found that subjects with AHs and/or passivity experiences did not rate self-produced tactile stimuli as less intense, tickly, or pleasant than externally produced stimuli, as persons without these symptoms do. This effect is generally understood as a failure in self-monitoring (Frith and Done 1989). While there were no consistent relationships between the affective indexes and demographic characteristics, there was one relationship with principal psychiatric diagnosis. On average, the quality of AHs experienced by patients diagnosed with BPD was significantly more negative than that of patients with other diagnoses. The prevalence of AHs in BPD is unclear: some reports (Clarke et al. 1995; Dowson et al. 2000) suggest that they are common, while others suggest that they are rare (Links et al. 1989; Nishizono-Maher et al. 1993). Our finding, based on only five cases, suggests that when they do occur, they are markedly negative. AHs that addressed the patient in the second person were significantly more unpleasant than those that did not; this is consistent with the finding by Oulis et al. (1997) that most patients' voices were both unpleasant and in the second person. Voice content that was linked with a significant person in the patient's life and voices that could have been the patient's own voice or thoughts were perceived as more pleasant. This group of findings suggests that the pleasantness or otherwise of AHs is related to how integrated they are with the patient's experience. It may be that the impersonality of second person address, and the unrelatedness of the experiences either to other people in one's life, or to one's own thinking, rendet them less pleasant (Morrison and Baker 2000). Chadwick and Birchwood (1994) found that, in many of their patients, beliefs about the malevolence or benevolence of voices were unrelated to the voice content, but Close and Garety (1998) found a strong relationship. Many approaches to reducing the distress caused by AHs involve recognizing and using their personal meaning (Romme and Escher 1993), and it has been suggested that the development of a stable relationship with AHs can serve an adaptive function (Benjamin 1989). It should be recognized that over half of our subjects were drawn from inpatient settings, and the average duration of hallucinatory experience was nearly 10 years. Thus, our sample could be considered to have serious and long-standing illnesses and caution needs to be exercised in generalizing our findings to patients with emerging, recent-onset, or transient disorders. The manner in which hearers evaluate their voices may well change and evolve with the progress of the illness. Indeed, investigation of this process should be a research priority, as early intervention may present an opportunity to temper patients' negative experience of their hallucinations. In conclusion, the analyses presented here demonstrate that two dimensions are required to characterize subjects' positive and negative experiences of and responses to AHs. These dimensions are uncorrelated. This is in contrast to much previous research that has assumed that a single dimension is adequate to record subjects' responses to their hallucinations. Even subjects who assess the tone, content, and feeling of their AHs as extremely negative may also rate part of their experience in positive terms. We may speculate that not wanting to lose positive AHs may be one reason underlying noncompliance with treatment, and inquiring into all aspects of AHs may increase empathy with the patient. The complexity of hallucinatory experience has yet to impact appropriately on research and clinical practice. References Allen, H.A.; Halperin, J.; and Friend, R. Removal and diversion tactics and the control of auditory hallucinations. Behaviour Research and Therapy, 23(5): , American Psychiatric Association. DSM-IV: Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: APA, Benjamin, L.S. Is chronicity a function of the relationship between the person and the auditory hallucination? Schizophrenia Bulletin, 15(2): , Bender, P.M. Comparative fit indexes in structural models. Psychological Bulletin, 107: , Bentler, P.M., and Bonnett, D.G. Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88: , Blakemore, S.-J.; Smith, J.; Steel, R.; Johnstone, E.C.; and Frith, CD. The perception of self-produced sensory stimuli in patients with auditory hallucinations and passivity experiences: Evidence for a breakdown in selfmonitoring. Psychological Medicine, 30: , Bradburn, N.M. The Structure of Psychological Well- 169

8 Schizophrenia Bulletin, Vol. 30, No. 1, 2004 D.L. Copolov et al. Being. Chicago, IL: Aldine, Browne, M.W., and Cudeck, R.C. Alternative ways of assessing model fit. Sociological Methods & Research, 21: , Carter, D.M.; Mackinnon, A.; and Copolov, D.L. Patients' strategies for coping with auditory hallucinations. Journal of Nervous and Mental Disease, 184: , Carter, D.M.; Mackinnon, A.; Howard, S.; Zeegers, T.; and Copolov, D.L. The development and reliability of the Mental Health Research Institute Unusual Perceptions Schedule (MUPS): An instrument to record auditory hallucinatory experience. Schizophrenia Research, 16: , Chadwick, P., and Birchwood, M. The omnipotence of voices: A cognitive approach to auditory hallucinations. British Journal of Psychiatry, 164: , Clarke, M.; Hafner, R.J.; and Holme, G. Borderline personality disorder: A challenge for mental health services. Australian and New Zealand Journal of Psychiatry, 29(3): , Close, H., and Garety, P. Cognitive assessment of voices: Further developments in understanding the emotional impact of voices. British Journal of Clinical Psychology, 37: , Done, D.J.; Frith, CD.; and Owens, D.C. Reducing persistent auditory hallucinations by wearing an ear-plug. British Journal of Clinical Psychology, 25(2): , Dowson, J.H.; Sussams, P.; Grounds, A.T.; and Taylor, J. Association of self-reported past "psychotic" phenomena with features of personality disorders. Comprehensive Psychiatry, 41(1):42^8, Frith, CD., and Done, D.J. Experiences of alien control in schizophrenia reflect a disorder in the central monitoring of action. Psychological Medicine, 31: , Haddock, G.; Slade, P.D.; Bentall, R.P.; Reid, D.; and Faragher, E.B. A comparison of the long-term effectiveness of distraction and focusing in the treatment of auditory hallucinations. British Journal of Medical Psychology, 71(3): , Hamilton, M. Fish's Schizophrenia. Bristol, U.K.: John Wright and Sons, Links, P.S.; Steiner, M.; and Mitton, J. Characteristics of psychosis in borderline personality disorder. Psychopathology, 22(4): , Lowe, G.R. The phenomenology of hallucinations as an aid to differential diagnosis. British Journal of Psychiatry, 123: , Mclnnis, M., and Marks, I. Audiotape therapy for persistent auditory hallucinations. British Journal of Psychiatry, 157: , Miller, L.J.; O'Connor, E.; and DiPasquale, T. Patients' attitudes toward hallucinations. American Journal of Psychiatry, 150: , Morrison, A.P., and Baker, C.A. Intrusive thoughts and auditory hallucinations: A comparative study of intrusions in psychosis. Behaviour Research and Therapy, 38(11): , Muth6n, L.K., and Muth6n, B.O. Mplus User's Guide. Los Angeles, CA: Muth6n and Muth6n, Nayani, T.H., and David, A.S. The auditory hallucination: A phenomenological survey. Psychological Medicine, 26: ,1996. Nishizono-Maher, A.; Dcuta, N.; Ogiso, Y.; Moriya, N.; Miyake, Y.; and Minakawa, K. Psychotic symptoms in depression and borderline personality disorder. Journal of Affective Disorders, 28(4): , O'Sullivan, K. Dimensions of coping with auditory hallucinations. Journal of Mental Health, 3: , Oulis, P.; Mamounas, J.; Hatzimanolis, J.; and Christodoulou, G.N. A clinical study of auditory hallucinations. Psychiatriki, 8(3): , Oulis, P.G.; Mavreas, V.G.; Mamounas, J.M.; and Stefanis, C.N. Clinical characteristics of auditory hallucinations. Ada Psychiatrica Scandinavica, 92:97-102, Romme, M., and Escher, A. Accepting Voices. London, U.K.: MIND, Romme, M.A.J.; Honig, A.; Noorthoorn, E.O.; and Escher, A.D.M.A.C. Coping with hearing voices: An emancipatory approach. British Journal of Psychiatry, 161:99-103, Schneider, K. Clinical Psychopathology. London, U.K.: Grune and Stratton, Shergill, S.S.; Murray, R.M.; and McGuire, P.K. Auditory hallucinations: A review of psychological treatments. Schizophrenia Research, 32: , Silva, F., and Lopez de Silva, M.C. Hallucinations and behavior modification. Analisis y Modificacion de Conducta, 2(2):95-131, Slade, P., and Bentall, R., eds. Sensory Deception: A Scientific Analysis of Hallucinations. London, U.K.: Croom Helm, Slade, P.D. The external control of auditory hallucinations: An information theory analysis. British Journal of Social and Clinical Psychology, 13(l):73-79, Spitzer, R.L.; Williams, J.B.W.; Gibbon, M.; and First, M.B. Structured Clinical Interview for the DSM-III-R, Patient Edition (SCID-P, Version 1.0). Washington, DC: 170

9 Correlates of the Affective Impact of Auditory Hallucinations Schizophrenia Bulletin, Vol. 30, No. 1, 2004 American Psychiatric Press, Tien, A. Distributions of hallucinations in the population. Social Psychiatry & Psychiatric Epidemiology, 26: , Van Schuur, W.H., and Kruijtbosch, M. Measuring subjective well-being: Unfolding the Bradburn Affect Balance Scale. Social Indicators Research, 36:49-74, Acknowledgments This research was supported by grants from the National Health and Medical Research Council (grants and ). We are indebted to the many patients who participated in this study. We also thank Dorothy Carter, Eliza Sims, and Rosemary Thomas, who conducted the interviews. The Authors David L. Copolov, MBBS, Ph.D., DPM, MPM, FRACP, FRANZCP, is Director, Mental Health Research Institute of Victoria, Parkville, Victoria, Australia. Andrew Mackinnon, B.Sc. (Hons), Ph.D., is Deputy Director, Mental Health Research Institute of Victoria; and Professor, Department of Psychological Medicine, Monash University, Victoria, Australia. Tom Trauer, B.A. (Hons), Ph.D., ABPsS, MAPsS, is Honorary Principal Research Fellow, Mental Health Research Institute of Victoria; and Associate Professor, Department of Psychological Medicine, Monash University. 171

10