- Email: [email protected]
Intrusive thoughts and auditory hallucinations: a comparative study of intrusions in psychosis
Behaviour Research and Therapy 38 (2000) 1097±1106
www.elsevier.com/locate/brat
Intrusive thoughts and auditory hallucinations: a comparative study of intrusions in psychosis Anthony P. Morrison a, b,*, Caroline A. Baker a a
Psychology Services, Mental Health Services of Salford, Bury New Road, Manchester M25 3BL, UK b Department of Clinical Psychology, University of Liverpool, Liverpool, UK
Abstract Several theories of auditory hallucinations implicate the involvement of intrusive thoughts and other theories suggest that the interpretation of voices determines the distress associated with them. This study tested the hypotheses that patients who experience auditory hallucinations will experience more intrusive thoughts and be more distressed by them and interpret them as more uncontrollable and unacceptable than the control groups. It also examines whether the interpretation of hallucinations is associated with the distress caused by them and whether there are dierences in the way that patients respond to and interpret their thoughts and voices. A questionnaire examining the frequency of intrusive thoughts and the reactions to them was administered to a group of patients with a diagnosis of schizophrenia who experienced auditory hallucinations, a psychiatric control group and a non-patient control group. In addition, the patients in the ®rst group completed a similar questionnaire in relation to their voices. Analyses of covariance showed that patients who experienced auditory hallucinations had more intrusive thoughts than the control groups and that they found their intrusive thoughts more distressing, uncontrollable and unacceptable than the control groups. Correlational analyses revealed that patients' interpretations of their voices were associated with the measures of distress in relation to them. Repeated measures analyses of covariance found no dierences between thoughts and voices on the dimensions assessed. The theoretical and clinical implications of these ®ndings are discussed. 7 2000 Elsevier Science Ltd. All rights reserved.
* Corresponding author. 0005-7967/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 0 0 5 - 7 9 6 7 ( 9 9 ) 0 0 1 4 3 - 6
1098
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
1. Introduction There has been extensive research conducted in recent years examining cognitive intrusions and their role in psychopathology. Intrusive thoughts were originally de®ned by Rachman (1978) as being repetitive thoughts, images or impulses that are unacceptable or unwanted; subsequently Rachman (1981) added that they are usually accompanied by subjective discomfort and must interrupt ongoing activity. It has been found that normal intrusive thoughts are a common experience (Rachman & De Silva, 1978; Salkovskis & Harrison, 1984) and it has been suggested that many everyday thoughts could be de®ned as being intrusive (Rachman & Hodgson, 1980). Several of the current theories regarding the development and maintenance of auditory hallucinations explicitly involve some notion of intrusions and others are certainly compatible with such notions. Homan (1986) has suggested that auditory hallucinations are the result of `parasitic memories' which disrupt language production processes and that the unintendedness of verbal images is a key component of the phenomenology of voices. In addition, Hemsley's (1993) cognitive model of schizophrenia suggests that the ``intrusion of unexpected/unintended material from long-term memory'' is a cognitive abnormality associated with schizophrenia. Morrison, Haddock and Tarrier (1995) presented an heuristic model which suggests that auditory hallucinations may be experienced when intrusive thoughts are attributed to an external source, in order to reduce cognitive dissonance. They speculate that this dissonance is caused by the incompatibility of certain intrusive thoughts and metacognitive beliefs (in particular, beliefs about controllability). Bentall (1990) has also implicated metacognitive beliefs as a top-down factor that may in¯uence the occurrence of auditory hallucinations. Baker and Morrison (1998) found that patients experiencing auditory hallucinations scored higher on metacognitive beliefs concerning both positive beliefs about worry and negative beliefs about uncontrollability and danger. Wells and Matthews' (1994) self-referent executive function (S-REF) model of emotional dysfunction would also suggest that the occurrence of hallucinations may be in¯uenced by such metacognitive beliefs, as hallucinations would be conceptualised within their model as low-level intrusions mediated by self-beliefs. Such beliefs are likely to be associated with dysfunctional attempts at control which would be expected to increase the frequency of intrusions. These theories would also suggest that patients experiencing auditory hallucinations would interpret intrusive thoughts as being uncontrollable and dangerous and be more upset by such intrusions. Morrison (1998), in a cognitive analysis of the maintenance of auditory hallucinations, suggested that an internal or external trigger results in a normal auditory hallucination that is then misinterpreted as threatening the physical or psychological integrity of the individual (such as ``I must be mad'', ``The Devil is talking to me'' and ``If I do not obey the voices they will hurt me''). These misinterpretations produce an increase in negative mood and physiological arousal which produce more hallucinations leading to a vicious circle. Simultaneously, the misinterpretation of the hallucination elicits safety seeking behaviours (including hypervigilance) designed to prevent the feared outcome (e.g. madness, possession or obedience) which can both increase the occurrence of auditory hallucinations and prevent the discon®rmation of the misinterpretation (therefore maintaining it). There is considerable evidence that suggests that the interpretation of intrusions is central to the understanding of
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
1099
auditory hallucinations. Kingdon and Turkington (1993, p. 78) state that ``the meaning invested in hallucinations may also be of importance Ð whether a person says to himself, `The devil is talking to me' or `I must be going crazy', or dismissively; `That was a strange sensation, I must have been overtired'``. Tarrier (1987) has noted that appraisals of positive symptoms elicited such responses, and Chadwick and Birchwood (1994) conclude that coping strategies chosen by patients experiencing auditory hallucinations (particularly engagement and resistance) appear to be driven by underlying beliefs about voices, and assert that ``aective, cognitive and behavioural responses evolve together and are always meaningfully related'' (p. 200). In a study utilising the beliefs about voices questionnaire (BAVQ; Chadwick & Birchwood, 1995) they found a strong positive relationship between appraisals of malevolence and resistance of the voices and between appraisals of benevolence and engagement with the voices. Therefore, it is likely that interpretations of hallucinatory phenomena will determine the aective responses. This study tests the hypothesis that intrusive thoughts in patients who experience auditory hallucinations will be experienced more frequently and be rated as more distressing emotionally and interpreted as more uncontrollable and unacceptable, in comparison with the psychiatric and non-patient control groups. It is also predicted that interpretation of hallucinations will be associated with the distress caused by them. It is also hypothesised that there will be dierences in the way that patients who experience auditory hallucinations will respond to and interpret their thoughts and voices.
2. Method 2.1. Participants 2.1.1. Hallucinating group This group consisted of 15 participants who met DSM-IV (APA, 1994) criteria for schizophrenia, and who were currently experiencing auditory hallucinations. Ages of participants in this group ranged from 28 to 65, with a mean of 43.93 years S:D: 9:90), and there were 11 males and four females. All participants were being treated with neuroleptic medication. 2.1.2. Non-hallucinating psychiatric control group Fifteen participants who met DSM-IV criteria for schizophrenia, but had not heard voices for at least three years were recruited into this group. Ages ranged from 25 to 64, the mean age being 42.93 years S:D: 10:53), and there were 12 males and three females. All participants were currently receiving neuroleptic medication. 2.1.3. Non-psychiatric control group The non-psychiatric control group consisted of 15 participants, who reported that they had not experienced any psychiatric illness in the previous three years and had never experienced
1100
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
auditory hallucinations. Participants were aged between 25 and 65 years, and the mean age was 38.73 years S:D: 14:85). There were 10 males and ®ve females. There were no signi®cant dierences between the groups in terms of age and scores on a measure of IQ (National Adult Reading Test; Nelson, 1982). 2.2. Materials The following measures were administered to participants (in addition to those described in the procedure). 2.2.1. Distressing thoughts questionnaire (DTQ: Clark & de Silva, 1985) This measure consists of 12 distressing thought statements (e.g. ``Thoughts or images that I am a failure'', ``Thoughts or images of the death of a close friend or family member''). For each statement, a total of ®ve questions are used to assess frequency (`How often does thought or image enter your mind?), sadness (`How sad or unhappy does this thought or image make you feel?'), worry (`How worried does this thought or image make you feel?), removal (How dicult is it for you to remove this thought from your mind?') and disapproval (How much do you disapprove of having this thought or image enter your mind?'). All items are rated on a nine point likert scale. Frequency is rated from `never' (1) to `daily' (9), through `at least twice per month' (5). The other four factors are rated from `not at all' (1) to `extremely' (9), with 5 representing `moderately'. Clark and de Silva (1985) used the DTQ to assess the distressing thoughts of a large non-psychiatric sample (303 undergraduate students), and found the scale to be suciently reliable. 2.2.2. Distressing voices questionnaire (DVQ) This questionnaire was designed by the authors for the purposes of this study to measure the dimensions of auditory hallucinations on the same subscales as the DTQ. It was hoped that by so doing, directly comparable data would be generated. The DVQ was constructed by changing the DTQ thought-referent statements to voice-referent statements by substituting the word `voice' for `thought' e.g. `Voice(s) saying that I am a failure'. Each voice statement was assessed using the same factors used in the DTQ, and scored in the same way. 2.3. Procedure This study was part of a larger procedure which is described more fully in Baker and Morrison (1998). All patients were assessed using the hallucinations, delusions and the behavioural observations subscales of a structured clinical interview (modi®ed KGV) which was conducted by the second author; these are scored on a ®ve point scale (0±4). All subjects were then given the Hospital Anxiety and Depression Scale (HAD; Zigmond & Snaith, 1983) which is a 14 item self-report scale and the Meta-Cognitions Questionnaire (MCQ: CartwrightHatton & Wells, 1997) which measures metacognitive beliefs using 65 items, followed by the administration of the National Adult Reading Test (NART; Nelson, 1982). Each subject then
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
1101
completed the DTQ and those subjects who experienced auditory hallucinations ®nally completed the DVQ. The data was analysed using parametric statistics as it met assumptions of normality, based upon an inspection of skewness and measures of central tendency.
3. Results 3.1. Subject characteristics All descriptive data showing subject characteristics are shown in Table 1. One-way analyses of variance indicated that there were no signi®cant dierences between the groups on age F 2, 42 0:80, n.s.), estimated IQ F 2, 42 1:67, n.s.), anxiety F 2, 42 1:89, n.s.) or depression F 2, 42 2:26, n.s.). The sex ratios appeared similar for the three groups. Independent t-tests indicated that the only signi®cant dierence between groups 1 and 2 on the KGV-R was the `hallucinations' score t 17:0, d:f: 28, p < 0.01). The group means were identical for `delusions' scores, and for ratings of negative symptoms. Before analysing the data, the internal reliability of the two scales was measured using Cronbach's alpha tests. These indicated that the internal reliability of the DTQ was very good (Cronbach a 0:98 and the DVQ was also found to have high internal reliability (Cronbach a 0:97). 3.2. Intrusive thoughts An analysis of covariance using frequency of depressing thoughts as the dependent variable showed a signi®cant main eect of GROUP F 2, 44 14:45, p < 0.001) and depression was a signi®cant covariate t 3:13, p < 0.005). Inspection of parameter estimates revealed that this was accounted for by the hallucinators scoring signi®cantly higher than the two control groups. A similar analysis of covariance using frequency of anxious thoughts as the dependent variable and anxiety as a covariate showed a signi®cant main eect for group
Table 1 Subject characteristics Variable
Hallucinators
Psychiatric controls
Non-patient controls
Age Sex ratio (M:F) Anxiety Depression NART I.Q. KGV-R sub-scales Hallucinations Delusions Behavioural observations
43.93 (9.90) 11:4 10.27 (4.64) 7.6 (5.40) 107.20 (5.81)
42.93 (10.53) 12:3 8.20 (4.28) 5.27 (4.71) 105.20 (7.90)
38.73 (14.85) 10:5 7.33 (3.79) 4.0 (3.89) 110.27 (8.92)
3.87 (0.35) 2.53 (1.59) 2.87 (2.32)
0.33 (0.72) 2.53 (1.25) 2.87 (2.32)
± ± ±
1102
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
F 2, 44 18:58, p < 0.001). Anxiety was a signi®cant covariate t 2:69, p < 0.01). Inspection of parameter estimates showed that the main eect was accounted for by the hallucinators scoring signi®cantly higher than the control groups. Multivariate analyses of covariance were conducted using emotional responses and interpretations of distressing thoughts as dependent variables and frequency of distressing thoughts as covariates. In relation to anxious thoughts the analysis revealed a signi®cant main eect for group F 3:48, p < 0.005) and frequency of anxious thoughts was found to be a signi®cant covariate for all of the dependent variables (t > 4.2, p < 0.001, for all variables). Univariate F-tests were used to investigate the source of this eect and revealed signi®cant group dierences for worry F 2, 40 3:83, p < 0.05) and disapproval F 2, 40 6:42, p < 0.005) both of which were accounted for by the hallucinators scoring signi®cantly higher than the control groups, and removal F 2, 40 9:84, p < 0.001) which was accounted for by the hallucinators scoring signi®cantly higher than the psychiatric controls who scored signi®cantly higher than the non-patient controls. A trend towards a signi®cant group eect for sadness was noted F 2, 40 3:00, p 0:061). In relation to depressing thoughts the analysis revealed a signi®cant main eect for group F 2:95, p < 0.01) and frequency of depressing thoughts was found to be a signi®cant covariate for all of the dependent variables (t > 3.1, p < 0.005, for all variables). Univariate F-tests were used to investigate the source of this eect and again revealed signi®cant group dierences for worry F 2, 40 6:36, p < 0.005) and disapproval F 2, 40 7:85, p < 0.001) both of which were accounted for by the hallucinators scoring signi®cantly higher than the control groups, and removal F 2, 40 9:54, p < 0.001) which was accounted for by the hallucinators scoring signi®cantly higher than the psychiatric controls who scored signi®cantly higher than the non-patient controls. A trend towards a signi®cant group eect for sadness was also noted F 2, 40 3:09, p = 0.056). The means and standard
Table 2 Means (and standard deviations) for dimensions of distressing thoughts and voicesa Variable Anxiety-related intrusions Frequency Sadness Worry Removal Disapproval Depression-related intrusions Frequency Sadness Worry Removal Disapproval a a,b
Non-patient controls
Psychiatric controls
Hallucincators (thoughts)
Hallucinators (voices)
21.87a (10.72) 20.93 (10.70) 16.60a (7.71) 15.53a (7.94) 16.93a (10.41)
26.73a (11.03) 24.73 (6.87) 22.33a (10.26) 21.67b (6.70) 21.47a (8.35)
42.67b (7.30) 42.20 (8.44) 41.00b (9.79) 40.80c (9.70) 42.00b (9.47)
29.29 31.79 31.57 32.43 33.07
(12.57) (12.67) (12.34) (14.50) (12.23)
19.33a (7.17) 21.00 (7.88) 17.53a (7.29) 13.60a (7.86) 14.00a (7.42)
25.40a (6.90) 27.00 (6.77) 27.00a (9.37) 19.87b (6.88) 22.53a (8.46)
36.80b (7.69) 40.67 (6.29) 39.80b (7.07) 39.53c (7.49) 40.36b (8.59)
34.50 35.15 35.29 37.00 37.79
(8.90) (10.77) (10.89) (12.17) (10.51)
indicate the location of signi®cant between groups dierences as revealed using post-hoc tests.
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
1103
deviations for frequency, sadness, worry, disapproval and removal for both depressing and anxious thoughts are shown in Table 2. 3.3. Intrusions: thoughts and hallucinations Paired t-tests revealed signi®cant dierences between the frequency of anxious thoughts and anxiety related voices t 2:49, p < 0.05) and between the frequency of depressed thoughts and depression-related voices t 2:82, p < 0.05). Inspections of the means showed that in both cases this was accounted for by voices being less frequent than thoughts. A series of paired t-tests revealed signi®cant dierences between the following dimensions of thoughts and voices (all signi®cance levels are two-tailed): worry t 2:25, p < 0.05), sadness t 2:84, p < 0.05) and removal t 2:24, p < 0.05) for anxiety-related intrusions and sadness t 2:53, p < 0.05) for depression-related intrusions. There were trends for disapproval of anxiety-related intrusions t 1:86, p 0:087 and worry for depression-related intrusions t 1:83, p 0:90). All of these dierences were accounted for by participants scoring higher for thoughts than voices. There were no dierences for removal or disapproval of depression-related intrusions. A series of repeated measures analyses of covariance using emotional response and interpretation of thoughts and voices as the dependent variables and frequency of thoughts and voices as a varying covariate were conducted to examine whether the above dierences were still present when frequency was controlled for. These analyses revealed no signi®cant dierences for any of the dimensions (all F values (1, 12) < 2.6, non-signi®cant for all analyses). The means and standard deviations for frequency, sadness, worry, disapproval and removal for both depression-related and anxious-related voices are also shown in Table 2. 3.4. Interpretations of hallucinations Correlational analyses between the emotional responses to the hallucinations and the interpretations of the hallucinations showed signi®cant associations for the composite measure of anxiety-related and depression-related voices (see Table 3 for the correlation matrices). To further explore these relationships, a multiple regression analysis was conducted using worry
Table 3 Correlation matrix for dimensions of voicesa Variable
2
3
4
5
(1) (2) (3) (4) (5)
0.82 ± ± ± ±
0.87 0.99 ± ± ±
0.81 0.89 0.94 ± ±
0.85 0.97 0.98 0.88 ±
a
Frequency Worry Sadness Removal Disapproval
All correlations signi®cant ( p < 0.001).
1104
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
about hallucinations as the dependent variable and frequency of voices was entered on step 1 and disapproval of the voices on step 2 as predictor variables. The correlation matrix and the results of the ®nal multiple regression equation can be seen in Tables 3 and 4, respectively. The multiple R was 0.81 and signi®cant F 6, 79 4:12, p < 0.002). The adjusted R 2 was 0.18, indicating that a moderate amount of the variance was accounted for by these predictor variables. An examination of the tolerances of the individual variables found them to be acceptably high, indicating that colinearity was not a problem. On step 1, with the mood variables entered, the multiple R was 0.81 and signi®cant F 1, 12 23:85, p < 0.001). On step 2, when the interpretation variable was entered, the multiple R was 0.97 and signi®cant, the adjusted R 2 was 0.93 and the increment in R 2 of 0.28 was signi®cant F 52:46, p < 0.0001).
4. Discussion The results of this study suggest that patients who experience auditory hallucinations experience more anxiety-related and depression-related cognitive intrusions than both psychiatric and non-patient controls. For both anxiety-related and depression-related intrusive thoughts it was found that patients experiencing auditory hallucinations found such intrusions more worrying and more dicult to remove and disapproved of such intrusions more than both control groups; in addition there was a strong trend for these patients to feel more sad as a result of their intrusions. These ®ndings support the suggestion of Morrison et al. (1995) that intrusive thoughts are involved in the development of auditory hallucinations and are consistent with the view that such intrusions are disapproved of by patients experiencing auditory hallucinations. The ®nding that there were dierences in some of the responses to hallucinations and thoughts and the way that some were interpreted is consistent with theories of hallucination that suggest some bene®t is associated with the misattribution of thoughts as voices, such as anxiety reduction (Bentall, 1990) or reduction of cognitive dissonance (Morrison et al., 1995). However, the fact that these dierences were not found when frequency of thoughts and voices were controlled for suggests that they may have been a function of frequency; further work with a larger sample size is probably required before a de®nitive conclusion can be reached. The ®nding that the interpretations of voices were associated with the emotional reactions to voices supports the cognitive models of Chadwick and Birchwood (1994) and Morrison (1998) which suggest that beliefs about voices or interpretations of voices determine aective
Table 4 Summary statistics for regression analysis Variable
Beta
Partial r
F
Sig. of F
Step 1: Frequency of voices Step 2: Disapproval of voices
ÿ0.07 1.03
ÿ0.14 0.91
0.22 52.47
n.s. 0.0001
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
1105
responses. In addition, the ®nding that the interpretation of voices (how much patients disapproved of them) was the only signi®cant predictor of emotional reaction to the voices (how much patients worried about them) adds further support to such theories. The interpretation of these ®ndings should be quali®ed by the consideration of a number of methodological weaknesses. The comparative analyses of the experiences of intrusive thoughts across the groups may have been aided by the inclusion of non-psychotic anxious and depressed control groups. Also, the small number n 13 of patients experiencing hallucinations that provided data on both their thoughts and voices makes the within subjects analyses more prone to type II errors (accepting the null hypothesis mistakenly). However, these results do oer some support for the involvement of intrusive thoughts in the experiencing of auditory hallucinations and the suggestion that patients who do experience auditory hallucinations ®nd intrusive thoughts more distressing and more dicult to remove (hence uncontrollable) and exhibit greater disapproval of them. This study also supports the suggestion that interpretations of voices are involved in determining the distress associated with them. The clinical implications of such ®ndings include conducting a detailed assessment of patients experiences of intrusive thoughts and voices and their metacognitive beliefs in relation to both. A cognitive behavioural formulation incorporating this information could then be used to guide the selection of interventions that may include challenging metacognitive beliefs using verbal and behavioural methods of reattribution, the provision of normalising information regarding intrusive thoughts and speci®cally targeting appraisals of voices that seem linked to maintaining the distress and disability associated with them.
References American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington DC: APA. Baker, C., & Morrison, A. P. (1998). Metacognition, intrusive thoughts and auditory hallucinations. Psychological Medicine, 28, 1199±1208. Bentall, R. P. (1990). The syndromes and symptoms of psychosis: or why you can't play 20 questions with the concept of schizophrenia and hope to win. In R. P. Bentall, Reconstructing schizophrenia. London: Routledge. Cartwright-Hatton, S., & Wells, A. (1997). Beliefs about worry and intrusions: the metacognitions questionnaire and its correlates. Journal of Anxiety Disorders, 11, 279±296. Chadwick, P., & Birchwood, M. (1994). The omnipotence of voices: a cognitive approach to auditory hallucinations. British Journal of Psychiatry, 164, 190±201. Chadwick, P. D., & Birchwood, M. (1995). The omnipotence of voices II: the beliefs about voices questionnaire (BAVQ). British Journal of Psychiatry, 166, 773±776. Clark, D., & de Silva, P. (1985). The nature of depressive and anxious depressive thoughts: distinct or uniform phenomena? Behaviour Research & Therapy, 23, 383±393. Hemsley, D. R. (1993). A simple (or simplistic?) cognitive model for schizophrenia. Behaviour Research & Therapy, 31, 633±645. Homan, R. E. (1986). Verbal hallucinations and language production processes in schizophrenia. Behavioural & Brain Sciences, 9, 503±548. Kingdon, D. G., & Turkington, D. (1993). Cognitive behavioural therapy of schizophrenia. New York: Guilford. Morrison, A. P. (1998). A cognitive analysis of auditory hallucinations: are voices to schizophrenia what bodily sensations are to panic? Behavioural and Cognitive Psychotherapy, 26, 289±302.
1106
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
Morrison, A. P., Haddock, G., & Tarrier, N. (1995). Intrusive thoughts and auditory hallucinations: a cognitive approach. Behavioural and Cognitive Psychotherapy, 23, 265±280. Nelson, H. E. (1982). National adult reading test. Windsor, Berks: NFER-Nelson. Rachman, S. J. (1978). An anatomy of obsessions. Behaviour Analysis & Modi®cation, 2, 235±278. Rachman, S. J. (1981). Unwanted intrusive cognitions. Advances in Behaviour Research & Therapy, 3, 89±99. Rachman, S. J., & De Silva, P. (1978). Abnormal and normal obsessions. Behaviour Research & Therapy, 16, 233± 238. Rachman, S. J., & Hodgson, R. (1980). Obsessions and compulsions. NJ: Prentice Hall. Salkovskis, P. M., & Harrison, J. (1984). Abnormal and normal obsessions: a replication. Behaviour Research & Therapy, 22, 549±552. Tarrier, N. (1987). An investigation of residual psychotic symptoms in discharged schizophrenic patients. British Journal of Clinical Psychology, 26, 141±143. Wells, A., & Matthews, G. (1994). Attention and emotion: a clinical perspective. Hillsdale, NJ: Laurence Erlbaum Associates. Zigmond, A. S., & Snaith, R. P. (1983). The hospital anxiety and depression scale. Acta Psychiatrica Scandinavia, 67, 361±370.
www.elsevier.com/locate/brat
Intrusive thoughts and auditory hallucinations: a comparative study of intrusions in psychosis Anthony P. Morrison a, b,*, Caroline A. Baker a a
Psychology Services, Mental Health Services of Salford, Bury New Road, Manchester M25 3BL, UK b Department of Clinical Psychology, University of Liverpool, Liverpool, UK
Abstract Several theories of auditory hallucinations implicate the involvement of intrusive thoughts and other theories suggest that the interpretation of voices determines the distress associated with them. This study tested the hypotheses that patients who experience auditory hallucinations will experience more intrusive thoughts and be more distressed by them and interpret them as more uncontrollable and unacceptable than the control groups. It also examines whether the interpretation of hallucinations is associated with the distress caused by them and whether there are dierences in the way that patients respond to and interpret their thoughts and voices. A questionnaire examining the frequency of intrusive thoughts and the reactions to them was administered to a group of patients with a diagnosis of schizophrenia who experienced auditory hallucinations, a psychiatric control group and a non-patient control group. In addition, the patients in the ®rst group completed a similar questionnaire in relation to their voices. Analyses of covariance showed that patients who experienced auditory hallucinations had more intrusive thoughts than the control groups and that they found their intrusive thoughts more distressing, uncontrollable and unacceptable than the control groups. Correlational analyses revealed that patients' interpretations of their voices were associated with the measures of distress in relation to them. Repeated measures analyses of covariance found no dierences between thoughts and voices on the dimensions assessed. The theoretical and clinical implications of these ®ndings are discussed. 7 2000 Elsevier Science Ltd. All rights reserved.
* Corresponding author. 0005-7967/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 0 0 5 - 7 9 6 7 ( 9 9 ) 0 0 1 4 3 - 6
1098
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
1. Introduction There has been extensive research conducted in recent years examining cognitive intrusions and their role in psychopathology. Intrusive thoughts were originally de®ned by Rachman (1978) as being repetitive thoughts, images or impulses that are unacceptable or unwanted; subsequently Rachman (1981) added that they are usually accompanied by subjective discomfort and must interrupt ongoing activity. It has been found that normal intrusive thoughts are a common experience (Rachman & De Silva, 1978; Salkovskis & Harrison, 1984) and it has been suggested that many everyday thoughts could be de®ned as being intrusive (Rachman & Hodgson, 1980). Several of the current theories regarding the development and maintenance of auditory hallucinations explicitly involve some notion of intrusions and others are certainly compatible with such notions. Homan (1986) has suggested that auditory hallucinations are the result of `parasitic memories' which disrupt language production processes and that the unintendedness of verbal images is a key component of the phenomenology of voices. In addition, Hemsley's (1993) cognitive model of schizophrenia suggests that the ``intrusion of unexpected/unintended material from long-term memory'' is a cognitive abnormality associated with schizophrenia. Morrison, Haddock and Tarrier (1995) presented an heuristic model which suggests that auditory hallucinations may be experienced when intrusive thoughts are attributed to an external source, in order to reduce cognitive dissonance. They speculate that this dissonance is caused by the incompatibility of certain intrusive thoughts and metacognitive beliefs (in particular, beliefs about controllability). Bentall (1990) has also implicated metacognitive beliefs as a top-down factor that may in¯uence the occurrence of auditory hallucinations. Baker and Morrison (1998) found that patients experiencing auditory hallucinations scored higher on metacognitive beliefs concerning both positive beliefs about worry and negative beliefs about uncontrollability and danger. Wells and Matthews' (1994) self-referent executive function (S-REF) model of emotional dysfunction would also suggest that the occurrence of hallucinations may be in¯uenced by such metacognitive beliefs, as hallucinations would be conceptualised within their model as low-level intrusions mediated by self-beliefs. Such beliefs are likely to be associated with dysfunctional attempts at control which would be expected to increase the frequency of intrusions. These theories would also suggest that patients experiencing auditory hallucinations would interpret intrusive thoughts as being uncontrollable and dangerous and be more upset by such intrusions. Morrison (1998), in a cognitive analysis of the maintenance of auditory hallucinations, suggested that an internal or external trigger results in a normal auditory hallucination that is then misinterpreted as threatening the physical or psychological integrity of the individual (such as ``I must be mad'', ``The Devil is talking to me'' and ``If I do not obey the voices they will hurt me''). These misinterpretations produce an increase in negative mood and physiological arousal which produce more hallucinations leading to a vicious circle. Simultaneously, the misinterpretation of the hallucination elicits safety seeking behaviours (including hypervigilance) designed to prevent the feared outcome (e.g. madness, possession or obedience) which can both increase the occurrence of auditory hallucinations and prevent the discon®rmation of the misinterpretation (therefore maintaining it). There is considerable evidence that suggests that the interpretation of intrusions is central to the understanding of
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
1099
auditory hallucinations. Kingdon and Turkington (1993, p. 78) state that ``the meaning invested in hallucinations may also be of importance Ð whether a person says to himself, `The devil is talking to me' or `I must be going crazy', or dismissively; `That was a strange sensation, I must have been overtired'``. Tarrier (1987) has noted that appraisals of positive symptoms elicited such responses, and Chadwick and Birchwood (1994) conclude that coping strategies chosen by patients experiencing auditory hallucinations (particularly engagement and resistance) appear to be driven by underlying beliefs about voices, and assert that ``aective, cognitive and behavioural responses evolve together and are always meaningfully related'' (p. 200). In a study utilising the beliefs about voices questionnaire (BAVQ; Chadwick & Birchwood, 1995) they found a strong positive relationship between appraisals of malevolence and resistance of the voices and between appraisals of benevolence and engagement with the voices. Therefore, it is likely that interpretations of hallucinatory phenomena will determine the aective responses. This study tests the hypothesis that intrusive thoughts in patients who experience auditory hallucinations will be experienced more frequently and be rated as more distressing emotionally and interpreted as more uncontrollable and unacceptable, in comparison with the psychiatric and non-patient control groups. It is also predicted that interpretation of hallucinations will be associated with the distress caused by them. It is also hypothesised that there will be dierences in the way that patients who experience auditory hallucinations will respond to and interpret their thoughts and voices.
2. Method 2.1. Participants 2.1.1. Hallucinating group This group consisted of 15 participants who met DSM-IV (APA, 1994) criteria for schizophrenia, and who were currently experiencing auditory hallucinations. Ages of participants in this group ranged from 28 to 65, with a mean of 43.93 years S:D: 9:90), and there were 11 males and four females. All participants were being treated with neuroleptic medication. 2.1.2. Non-hallucinating psychiatric control group Fifteen participants who met DSM-IV criteria for schizophrenia, but had not heard voices for at least three years were recruited into this group. Ages ranged from 25 to 64, the mean age being 42.93 years S:D: 10:53), and there were 12 males and three females. All participants were currently receiving neuroleptic medication. 2.1.3. Non-psychiatric control group The non-psychiatric control group consisted of 15 participants, who reported that they had not experienced any psychiatric illness in the previous three years and had never experienced
1100
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
auditory hallucinations. Participants were aged between 25 and 65 years, and the mean age was 38.73 years S:D: 14:85). There were 10 males and ®ve females. There were no signi®cant dierences between the groups in terms of age and scores on a measure of IQ (National Adult Reading Test; Nelson, 1982). 2.2. Materials The following measures were administered to participants (in addition to those described in the procedure). 2.2.1. Distressing thoughts questionnaire (DTQ: Clark & de Silva, 1985) This measure consists of 12 distressing thought statements (e.g. ``Thoughts or images that I am a failure'', ``Thoughts or images of the death of a close friend or family member''). For each statement, a total of ®ve questions are used to assess frequency (`How often does thought or image enter your mind?), sadness (`How sad or unhappy does this thought or image make you feel?'), worry (`How worried does this thought or image make you feel?), removal (How dicult is it for you to remove this thought from your mind?') and disapproval (How much do you disapprove of having this thought or image enter your mind?'). All items are rated on a nine point likert scale. Frequency is rated from `never' (1) to `daily' (9), through `at least twice per month' (5). The other four factors are rated from `not at all' (1) to `extremely' (9), with 5 representing `moderately'. Clark and de Silva (1985) used the DTQ to assess the distressing thoughts of a large non-psychiatric sample (303 undergraduate students), and found the scale to be suciently reliable. 2.2.2. Distressing voices questionnaire (DVQ) This questionnaire was designed by the authors for the purposes of this study to measure the dimensions of auditory hallucinations on the same subscales as the DTQ. It was hoped that by so doing, directly comparable data would be generated. The DVQ was constructed by changing the DTQ thought-referent statements to voice-referent statements by substituting the word `voice' for `thought' e.g. `Voice(s) saying that I am a failure'. Each voice statement was assessed using the same factors used in the DTQ, and scored in the same way. 2.3. Procedure This study was part of a larger procedure which is described more fully in Baker and Morrison (1998). All patients were assessed using the hallucinations, delusions and the behavioural observations subscales of a structured clinical interview (modi®ed KGV) which was conducted by the second author; these are scored on a ®ve point scale (0±4). All subjects were then given the Hospital Anxiety and Depression Scale (HAD; Zigmond & Snaith, 1983) which is a 14 item self-report scale and the Meta-Cognitions Questionnaire (MCQ: CartwrightHatton & Wells, 1997) which measures metacognitive beliefs using 65 items, followed by the administration of the National Adult Reading Test (NART; Nelson, 1982). Each subject then
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
1101
completed the DTQ and those subjects who experienced auditory hallucinations ®nally completed the DVQ. The data was analysed using parametric statistics as it met assumptions of normality, based upon an inspection of skewness and measures of central tendency.
3. Results 3.1. Subject characteristics All descriptive data showing subject characteristics are shown in Table 1. One-way analyses of variance indicated that there were no signi®cant dierences between the groups on age F 2, 42 0:80, n.s.), estimated IQ F 2, 42 1:67, n.s.), anxiety F 2, 42 1:89, n.s.) or depression F 2, 42 2:26, n.s.). The sex ratios appeared similar for the three groups. Independent t-tests indicated that the only signi®cant dierence between groups 1 and 2 on the KGV-R was the `hallucinations' score t 17:0, d:f: 28, p < 0.01). The group means were identical for `delusions' scores, and for ratings of negative symptoms. Before analysing the data, the internal reliability of the two scales was measured using Cronbach's alpha tests. These indicated that the internal reliability of the DTQ was very good (Cronbach a 0:98 and the DVQ was also found to have high internal reliability (Cronbach a 0:97). 3.2. Intrusive thoughts An analysis of covariance using frequency of depressing thoughts as the dependent variable showed a signi®cant main eect of GROUP F 2, 44 14:45, p < 0.001) and depression was a signi®cant covariate t 3:13, p < 0.005). Inspection of parameter estimates revealed that this was accounted for by the hallucinators scoring signi®cantly higher than the two control groups. A similar analysis of covariance using frequency of anxious thoughts as the dependent variable and anxiety as a covariate showed a signi®cant main eect for group
Table 1 Subject characteristics Variable
Hallucinators
Psychiatric controls
Non-patient controls
Age Sex ratio (M:F) Anxiety Depression NART I.Q. KGV-R sub-scales Hallucinations Delusions Behavioural observations
43.93 (9.90) 11:4 10.27 (4.64) 7.6 (5.40) 107.20 (5.81)
42.93 (10.53) 12:3 8.20 (4.28) 5.27 (4.71) 105.20 (7.90)
38.73 (14.85) 10:5 7.33 (3.79) 4.0 (3.89) 110.27 (8.92)
3.87 (0.35) 2.53 (1.59) 2.87 (2.32)
0.33 (0.72) 2.53 (1.25) 2.87 (2.32)
± ± ±
1102
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
F 2, 44 18:58, p < 0.001). Anxiety was a signi®cant covariate t 2:69, p < 0.01). Inspection of parameter estimates showed that the main eect was accounted for by the hallucinators scoring signi®cantly higher than the control groups. Multivariate analyses of covariance were conducted using emotional responses and interpretations of distressing thoughts as dependent variables and frequency of distressing thoughts as covariates. In relation to anxious thoughts the analysis revealed a signi®cant main eect for group F 3:48, p < 0.005) and frequency of anxious thoughts was found to be a signi®cant covariate for all of the dependent variables (t > 4.2, p < 0.001, for all variables). Univariate F-tests were used to investigate the source of this eect and revealed signi®cant group dierences for worry F 2, 40 3:83, p < 0.05) and disapproval F 2, 40 6:42, p < 0.005) both of which were accounted for by the hallucinators scoring signi®cantly higher than the control groups, and removal F 2, 40 9:84, p < 0.001) which was accounted for by the hallucinators scoring signi®cantly higher than the psychiatric controls who scored signi®cantly higher than the non-patient controls. A trend towards a signi®cant group eect for sadness was noted F 2, 40 3:00, p 0:061). In relation to depressing thoughts the analysis revealed a signi®cant main eect for group F 2:95, p < 0.01) and frequency of depressing thoughts was found to be a signi®cant covariate for all of the dependent variables (t > 3.1, p < 0.005, for all variables). Univariate F-tests were used to investigate the source of this eect and again revealed signi®cant group dierences for worry F 2, 40 6:36, p < 0.005) and disapproval F 2, 40 7:85, p < 0.001) both of which were accounted for by the hallucinators scoring signi®cantly higher than the control groups, and removal F 2, 40 9:54, p < 0.001) which was accounted for by the hallucinators scoring signi®cantly higher than the psychiatric controls who scored signi®cantly higher than the non-patient controls. A trend towards a signi®cant group eect for sadness was also noted F 2, 40 3:09, p = 0.056). The means and standard
Table 2 Means (and standard deviations) for dimensions of distressing thoughts and voicesa Variable Anxiety-related intrusions Frequency Sadness Worry Removal Disapproval Depression-related intrusions Frequency Sadness Worry Removal Disapproval a a,b
Non-patient controls
Psychiatric controls
Hallucincators (thoughts)
Hallucinators (voices)
21.87a (10.72) 20.93 (10.70) 16.60a (7.71) 15.53a (7.94) 16.93a (10.41)
26.73a (11.03) 24.73 (6.87) 22.33a (10.26) 21.67b (6.70) 21.47a (8.35)
42.67b (7.30) 42.20 (8.44) 41.00b (9.79) 40.80c (9.70) 42.00b (9.47)
29.29 31.79 31.57 32.43 33.07
(12.57) (12.67) (12.34) (14.50) (12.23)
19.33a (7.17) 21.00 (7.88) 17.53a (7.29) 13.60a (7.86) 14.00a (7.42)
25.40a (6.90) 27.00 (6.77) 27.00a (9.37) 19.87b (6.88) 22.53a (8.46)
36.80b (7.69) 40.67 (6.29) 39.80b (7.07) 39.53c (7.49) 40.36b (8.59)
34.50 35.15 35.29 37.00 37.79
(8.90) (10.77) (10.89) (12.17) (10.51)
indicate the location of signi®cant between groups dierences as revealed using post-hoc tests.
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
1103
deviations for frequency, sadness, worry, disapproval and removal for both depressing and anxious thoughts are shown in Table 2. 3.3. Intrusions: thoughts and hallucinations Paired t-tests revealed signi®cant dierences between the frequency of anxious thoughts and anxiety related voices t 2:49, p < 0.05) and between the frequency of depressed thoughts and depression-related voices t 2:82, p < 0.05). Inspections of the means showed that in both cases this was accounted for by voices being less frequent than thoughts. A series of paired t-tests revealed signi®cant dierences between the following dimensions of thoughts and voices (all signi®cance levels are two-tailed): worry t 2:25, p < 0.05), sadness t 2:84, p < 0.05) and removal t 2:24, p < 0.05) for anxiety-related intrusions and sadness t 2:53, p < 0.05) for depression-related intrusions. There were trends for disapproval of anxiety-related intrusions t 1:86, p 0:087 and worry for depression-related intrusions t 1:83, p 0:90). All of these dierences were accounted for by participants scoring higher for thoughts than voices. There were no dierences for removal or disapproval of depression-related intrusions. A series of repeated measures analyses of covariance using emotional response and interpretation of thoughts and voices as the dependent variables and frequency of thoughts and voices as a varying covariate were conducted to examine whether the above dierences were still present when frequency was controlled for. These analyses revealed no signi®cant dierences for any of the dimensions (all F values (1, 12) < 2.6, non-signi®cant for all analyses). The means and standard deviations for frequency, sadness, worry, disapproval and removal for both depression-related and anxious-related voices are also shown in Table 2. 3.4. Interpretations of hallucinations Correlational analyses between the emotional responses to the hallucinations and the interpretations of the hallucinations showed signi®cant associations for the composite measure of anxiety-related and depression-related voices (see Table 3 for the correlation matrices). To further explore these relationships, a multiple regression analysis was conducted using worry
Table 3 Correlation matrix for dimensions of voicesa Variable
2
3
4
5
(1) (2) (3) (4) (5)
0.82 ± ± ± ±
0.87 0.99 ± ± ±
0.81 0.89 0.94 ± ±
0.85 0.97 0.98 0.88 ±
a
Frequency Worry Sadness Removal Disapproval
All correlations signi®cant ( p < 0.001).
1104
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
about hallucinations as the dependent variable and frequency of voices was entered on step 1 and disapproval of the voices on step 2 as predictor variables. The correlation matrix and the results of the ®nal multiple regression equation can be seen in Tables 3 and 4, respectively. The multiple R was 0.81 and signi®cant F 6, 79 4:12, p < 0.002). The adjusted R 2 was 0.18, indicating that a moderate amount of the variance was accounted for by these predictor variables. An examination of the tolerances of the individual variables found them to be acceptably high, indicating that colinearity was not a problem. On step 1, with the mood variables entered, the multiple R was 0.81 and signi®cant F 1, 12 23:85, p < 0.001). On step 2, when the interpretation variable was entered, the multiple R was 0.97 and signi®cant, the adjusted R 2 was 0.93 and the increment in R 2 of 0.28 was signi®cant F 52:46, p < 0.0001).
4. Discussion The results of this study suggest that patients who experience auditory hallucinations experience more anxiety-related and depression-related cognitive intrusions than both psychiatric and non-patient controls. For both anxiety-related and depression-related intrusive thoughts it was found that patients experiencing auditory hallucinations found such intrusions more worrying and more dicult to remove and disapproved of such intrusions more than both control groups; in addition there was a strong trend for these patients to feel more sad as a result of their intrusions. These ®ndings support the suggestion of Morrison et al. (1995) that intrusive thoughts are involved in the development of auditory hallucinations and are consistent with the view that such intrusions are disapproved of by patients experiencing auditory hallucinations. The ®nding that there were dierences in some of the responses to hallucinations and thoughts and the way that some were interpreted is consistent with theories of hallucination that suggest some bene®t is associated with the misattribution of thoughts as voices, such as anxiety reduction (Bentall, 1990) or reduction of cognitive dissonance (Morrison et al., 1995). However, the fact that these dierences were not found when frequency of thoughts and voices were controlled for suggests that they may have been a function of frequency; further work with a larger sample size is probably required before a de®nitive conclusion can be reached. The ®nding that the interpretations of voices were associated with the emotional reactions to voices supports the cognitive models of Chadwick and Birchwood (1994) and Morrison (1998) which suggest that beliefs about voices or interpretations of voices determine aective
Table 4 Summary statistics for regression analysis Variable
Beta
Partial r
F
Sig. of F
Step 1: Frequency of voices Step 2: Disapproval of voices
ÿ0.07 1.03
ÿ0.14 0.91
0.22 52.47
n.s. 0.0001
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
1105
responses. In addition, the ®nding that the interpretation of voices (how much patients disapproved of them) was the only signi®cant predictor of emotional reaction to the voices (how much patients worried about them) adds further support to such theories. The interpretation of these ®ndings should be quali®ed by the consideration of a number of methodological weaknesses. The comparative analyses of the experiences of intrusive thoughts across the groups may have been aided by the inclusion of non-psychotic anxious and depressed control groups. Also, the small number n 13 of patients experiencing hallucinations that provided data on both their thoughts and voices makes the within subjects analyses more prone to type II errors (accepting the null hypothesis mistakenly). However, these results do oer some support for the involvement of intrusive thoughts in the experiencing of auditory hallucinations and the suggestion that patients who do experience auditory hallucinations ®nd intrusive thoughts more distressing and more dicult to remove (hence uncontrollable) and exhibit greater disapproval of them. This study also supports the suggestion that interpretations of voices are involved in determining the distress associated with them. The clinical implications of such ®ndings include conducting a detailed assessment of patients experiences of intrusive thoughts and voices and their metacognitive beliefs in relation to both. A cognitive behavioural formulation incorporating this information could then be used to guide the selection of interventions that may include challenging metacognitive beliefs using verbal and behavioural methods of reattribution, the provision of normalising information regarding intrusive thoughts and speci®cally targeting appraisals of voices that seem linked to maintaining the distress and disability associated with them.
References American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington DC: APA. Baker, C., & Morrison, A. P. (1998). Metacognition, intrusive thoughts and auditory hallucinations. Psychological Medicine, 28, 1199±1208. Bentall, R. P. (1990). The syndromes and symptoms of psychosis: or why you can't play 20 questions with the concept of schizophrenia and hope to win. In R. P. Bentall, Reconstructing schizophrenia. London: Routledge. Cartwright-Hatton, S., & Wells, A. (1997). Beliefs about worry and intrusions: the metacognitions questionnaire and its correlates. Journal of Anxiety Disorders, 11, 279±296. Chadwick, P., & Birchwood, M. (1994). The omnipotence of voices: a cognitive approach to auditory hallucinations. British Journal of Psychiatry, 164, 190±201. Chadwick, P. D., & Birchwood, M. (1995). The omnipotence of voices II: the beliefs about voices questionnaire (BAVQ). British Journal of Psychiatry, 166, 773±776. Clark, D., & de Silva, P. (1985). The nature of depressive and anxious depressive thoughts: distinct or uniform phenomena? Behaviour Research & Therapy, 23, 383±393. Hemsley, D. R. (1993). A simple (or simplistic?) cognitive model for schizophrenia. Behaviour Research & Therapy, 31, 633±645. Homan, R. E. (1986). Verbal hallucinations and language production processes in schizophrenia. Behavioural & Brain Sciences, 9, 503±548. Kingdon, D. G., & Turkington, D. (1993). Cognitive behavioural therapy of schizophrenia. New York: Guilford. Morrison, A. P. (1998). A cognitive analysis of auditory hallucinations: are voices to schizophrenia what bodily sensations are to panic? Behavioural and Cognitive Psychotherapy, 26, 289±302.
1106
A.P. Morrison, C.A. Baker / Behaviour Research and Therapy 38 (2000) 1097±1106
Morrison, A. P., Haddock, G., & Tarrier, N. (1995). Intrusive thoughts and auditory hallucinations: a cognitive approach. Behavioural and Cognitive Psychotherapy, 23, 265±280. Nelson, H. E. (1982). National adult reading test. Windsor, Berks: NFER-Nelson. Rachman, S. J. (1978). An anatomy of obsessions. Behaviour Analysis & Modi®cation, 2, 235±278. Rachman, S. J. (1981). Unwanted intrusive cognitions. Advances in Behaviour Research & Therapy, 3, 89±99. Rachman, S. J., & De Silva, P. (1978). Abnormal and normal obsessions. Behaviour Research & Therapy, 16, 233± 238. Rachman, S. J., & Hodgson, R. (1980). Obsessions and compulsions. NJ: Prentice Hall. Salkovskis, P. M., & Harrison, J. (1984). Abnormal and normal obsessions: a replication. Behaviour Research & Therapy, 22, 549±552. Tarrier, N. (1987). An investigation of residual psychotic symptoms in discharged schizophrenic patients. British Journal of Clinical Psychology, 26, 141±143. Wells, A., & Matthews, G. (1994). Attention and emotion: a clinical perspective. Hillsdale, NJ: Laurence Erlbaum Associates. Zigmond, A. S., & Snaith, R. P. (1983). The hospital anxiety and depression scale. Acta Psychiatrica Scandinavia, 67, 361±370.