Behavioural and Cognitive Psychotherapy, 2012, 40, 117–123
First published online 9 September 2011 doi:10.1017/S1352465811000348
Translating CBT for Voices into a Program for Carers:
A Pilot Study
La Trobe University, Melbourne, Australia
La Trobe University and North Western Mental Health, Melbourne, Australia
La Trobe University, Melbourne, Australia
Background: The treatment of persisting psychotic symptoms with cognitive behavioural
therapy (CBTp) is now established as an evidence-based treatment; however its availability
remains limited. We piloted a novel CBTp skills-based group training program for carers.
The aim was to reduce service users’ auditory hallucinations severity through carers’ use of
basic CBTp methods in their regular interactions. Method: Eight carer-service user dyads
and one carer-carer-service user triad participated. Carers attended the 10-week (25-hour)
program, and completed measures of subjective burden and expressed emotion (EE) over
a 30-week period that included a baseline phase prior to the training and implementation
phase. Service users completed weekly interviews assessing voice symptomatology. Analyses
were conducted for each dyad using time-series methods. Results: The training program was
rated highly. Improvements in symptoms and carer burden ranged from none to clinically
signiﬁcant across different dyads. Carer implementation of strategies was related to reduced
symptoms in one dyad; reductions in EE were related to symptom improvements in two
dyads. Conclusions: There may be beneﬁts in training carers in behavioural and cognitive
management of persisting hallucinations. Possible mechanisms for improvements in voice
symptomatology include reduction in carers’ EE, and carer implementation of strategies
Keywords: Cognitive behavioural therapy, psychosis, carers, training, auditory hallucinations.
Reprint requests to John Farhall, School of Psychological Science, La Trobe University, VIC 3086, Australia.
E-mail: email@example.com An extended version is also available online in the table of contents for this issue:
© British Association for Behavioural and Cognitive Psychotherapies 2011
118 J. Maxwell et al.
Treatment of psychotic symptoms with cognitive behavioural therapy (CBTp) has established
efﬁcacy); however, the availability of such treatment remains restricted outside of the
United Kingdom. A case study of delivery of informal cognitive therapy to an individual
with schizophrenia by a family carer (George, Bandopadhay and Cowan, 2005) raises the
possibility that dissemination of CBTp may be possible through training carers - a method that
may be particularly useful for difﬁcult-to-engage service users. The aim of this study was to
evaluate a course that aimed to skill carers of individuals who hear voices to better assist their
ill relative to cope with, and possibly reduce, persisting voices. The skills were based on CBTp
components but were to be implemented through carers’ daily interactions with their relative,
rather than through formal therapy sessions and “therapeutic” relationship. A valuable by-
product of such a course may be an improvement in carer-service user relationships, resulting
from more informed and helpful communication about persisting symptoms. The study
addressed the following research questions: Can the delivery of skills training in psychological
management strategies for voices delivered to carers reduce the severity of service users’
voices? Furthermore (while not a target of intervention), does the delivery of such a training
program result in changes in carer expressed emotion (EE) and subjective burden? Finally; if
changes are observed in service user symptomatology, are these associated with carers’ use of
psychological management strategies or altered carer expressed emotion?
Approvals were obtained from the relevant La Trobe University and Melbourne Health Ethics
Committees. Both members of each carer-service user dyad gave informed consent.
Eligibility criteria were: a DSM-IV diagnosis of schizophrenia or schizoaffective disorder and
6 months persistence of hallucinations whilst on stable medication. Eligible carers were family
members who had ﬁve or more contacts per week with their relative. Nine carers aged between
34 and 65 years, and 8 voice-hearing relatives aged between 19 and 40 years participated in the
study (8 carer-service user dyads and one carer-carer-service user triad). Duration of illness
ranged from 2 to 7 years, and number of hospitalizations from 2 to 10.
The Psychotic Symptom Rating Scale (PSYRATS; Haddock, McCarron, Tarrier and Faragher,
1999) was used to measure hallucinations severity. The Family Questionnaire (FQ) was used
to assess Carers’ Expressed Emotion (Wiedemann, Rayki, Feinstein and Hahlweg, 2002) and
the Involvement Evaluation Questionnaire (IEQ; van Wijngaarden et al., 2000) to measure
carer subjective burden. The number and frequency of cognitive and behavioural strategies
implemented by carers was recorded by a self-report checklist completed on a daily basis
during the 10 weeks training. Carer feedback was recorded in the ﬁnal session via a course
CBT for voices and a program for carers 119
Design and procedure
A within-subjects, multiple-baseline, repeated-measures design was selected to accommodate
small numbers and allow for the assessment of individual differences in efﬁcacy. The study
had two phases: a 10-week baseline phase followed by a 20-week training and implementation
phase (comprising 10, weekly, 21
/2hour interactive sessions, and a further 10 weeks, in which
carers were encouraged to continue using strategies learned). Voice phenomenology measures
were obtained from service users via brief weekly interviews for the 30 weeks of the study.
The IEQ and FQ were completed by carers weekly.
The 10-session program combined education and experiential exercises on the experience of
voice-hearing, how to communicate effectively about voices, and behavioural and cognitive
strategies for coping with voices (including coping strategy enhancement, belief change,
generating alternative explanations). A program manual is available upon request.
Statistical analyses used SPSS 14 for Windows. For primary outcome variables, scores were
converted into a percentage of the maximum score. The time-series data for each participant
were studied in baseline and intervention phases, and models (with conﬁdence intervals) ﬁtted
for each phase. Outcomes were measured by statistically (α<0.05), and clinically (greater
than 10% improvement), signiﬁcant differences between baseline and intervention phases.
Due to the small number of time-series data points, conﬁdence intervals were calculated
conservatively, using standard error values multiplied by critical values of the t-distribution
(based on 0.05 level of signiﬁcance on a two-tailed test). Linear regressions were run to
investigate the relationship between carer strategy use and voice severity.
A subsample of the PSYRATS data was co-rated by an independent provisional psychologist,
experienced in PSYRATS assessment and blind to study phase, yielding an almost perfect
correlation (PSYRATS total score r=0.998, t=55.89, p=.00).
Acceptability of the intervention
Carers’ feedback indicated high levels of acceptability (mean satisfaction rating 9.78/
10, ±0.35), but wide individual variation in rating ease of implementation of strategies
(from 5–10/10, μ=7, ±1.34). Qualitative feedback suggested that cognitive strategies
(e.g. discussing beliefs about voices) were more difﬁcult to implement than behavioural
120 J. Maxwell et al.
Carer strategy use
Carers reported their implementation of strategies on a daily checklist. Behavioural
strategies taught were: encouraging use of distraction, sub-vocalization, competing auditory
stimuli, self-instruction activity; encouraging changes in level of social contact; minimizing
stressors/triggers; and altering the environment. Cognitive strategies were: encouraging
distancing from voices; identifying beliefs about voices with relative; exploring evidence for
or against beliefs about voices; testing a belief by behavioural experiment; and identifying
alternative explanations. All carers reported using at least six of the eight behavioural
strategies, and at least two of the ﬁve cognitive strategies at some time during the training
and implementation phase.
Reduction in voice severity was associated with greater carer strategy use, for 5 out of 7
dyads; however, with the exception of one dyad – Service User 6 (t=−3.54, p=.01) whose
voice severity changed substantially – these relationships were not statistically signiﬁcant. A
moderate negative correlation approaching statistical signiﬁcance was identiﬁed for Service
User 2 (t=−2.03, p=.08).
Service User 1 reported voices at too few points for time-series analysis (3/10 weeks of
baseline; reducing to 3/20 weeks in training and follow-up phase). The mean differences in
voice severity (PSYRATS) between the baseline phase and the training and implementation
phase for service users 2–7, based on individual time series analyses, are presented in
Figure 1. Clinically signiﬁcant improvements in voice severity were identiﬁed for Service
User 7 (t=4.217, p=.00), and Service User 6 (t=2.90, p=.009). Service users 2, 3, 4
and 5 did not show clinically or statistically signiﬁcant improvements in voice severity in the
training and implementation phase.
Carers 6 and 7 showed clinically signiﬁcant reductions in EE in the training and
implementation phase of the study (t=4.53, p=.00; t=12.52, p=.00, respectively), with
Carers 4 a and 4b showing statistically signiﬁcant reductions (t=9.11, p=.00; t=8.29, p=
.00 respectively). Carers 2 and 5 showed a statistically, but not clinically signiﬁcant reduction
in expressed emotion in the training and implementation phase of the study (t=2.43, p=.02;
t=3.11, p=.00 respectively), with Carer 1 and Carer 8 showing no statistically signiﬁcant
change. Model ﬁtting was not possible for Carer 3 due to ﬂuctuating scores.
Five of the eight carers showed a statistically signiﬁcant reduction in burden between the
baseline phase and the training and implementation phase. Carer 7 showed a clinically
signiﬁcant reduction in Subjective burden with a mean reduction of 25.6% in the training and
implementation phase (t=12.94, p=.00). Carers 4 a and 4b and carers 1 and 6 all showed
statistically signiﬁcant reductions (t=10.22, p=.00; t=10.11, p=.00; t=3.07, p=.005;
t=2.90, p=.008, respectively), but only the latter two approached clinical signiﬁcance.
CBT for voices and a program for carers 121
Figure 1. Service user voice severity: mean differences from baseline to training and implementation
phase (with conﬁdence intervals)
This study evaluated the feasibility and impact of a course that taught CBT skills to carers of
voice hearers. The program was clearly acceptable: carers reported high levels of satisfaction,
an ability to understand the concepts and strategies, and an intention to continue using
these after the program. The feasibility of carers implementing naturalistic behavioural and
cognitive intervention strategies was also supported: all carers implemented at least 9 of the
13 strategies taught.
The statistically and clinically signiﬁcant improvements for two of the six service users
with time series data support our hypothesis that voice severity can be impacted by an
intervention that works through carers, rather than directly with the service users. The
proportion improving, if replicated in larger studies, is within the range of outcomes for
therapist-implemented CBTp (Gaudiano, 2005). Rather than the training adding to the burden
of carers, a majority experienced reduced subjective burden. Given that reductions in burden
have been rarely reported in the family intervention for psychosis literature, this raises the
possibility that the explanatory models and speciﬁc skills of the intervention gave carers
conﬁdence to interact more effectively with service users about voices.
The results support more than one model of change in voice severity. The primary model
of the study, that service user coping would improve, received some support: there was
an association between an increase in carers’ encouragement of psychological management
strategies and reduced voice severity, particularly for one dyad. It is possible that the analysis
undertaken (correlations between weekly ratings of voice severity and extent of carer strategy
122 J. Maxwell et al.
use that week) may have been insensitive by not allowing for any time lag between the carers’
suggestion of a strategy and the service users’ application(s) of it.
An alternative explanation for reductions in service user voice severity is via reduction
in carers’ EE. Despite not being a direct target of the intervention, clinically signiﬁcant
improvements in EE were identiﬁed for two carers, and in each case their voice hearing
relative showed clinically signiﬁcant improvements in voice severity. Reductions in EE may
have occurred through increased carer conﬁdence from skills acquisition or via a shift in
illness attributions (Barrowclough, Johnston and Tarrier, 1994) resulting from improved
understanding about the nature of psychotic experiences. If supported by future studies,
these results would strengthen the case for the importance of symptom-speciﬁc psycho-
education in family interventions. These results also raise a question regarding the speciﬁc
effects of behavioural and cognitive strategies over and above any contribution from improved
relationships and symptom-speciﬁc education.
Recruitment of a small, self-selected sample limits the generalizability of the results; however,
the difﬁculty in recruiting consenting pairs of service users and carers, and the evaluation
demands on participants precluded a larger scale pilot. Nonetheless, the study took an
important and realistic ﬁrst step in establishing a treatment effect when carers are trained to
informally use behavioural and cognitive change interventions with voice hearers. Inclusion
in future studies of speciﬁc measures of carer conﬁdence, relationships and attributions, and
the tracking of service user implementation of strategies would enable further assessment of
these possible mechanisms of action.
The present pilot study provides promising results indicating that training family carers in
basic principles and strategies of CBT for psychosis is possible, highly acceptable, and
can result in clinically signiﬁcant improvements in service user voice symptomatology, and
reduction in carer burden. The course thus has the potential to redress, in a naturalistic way,
two of the practical challenges to dissemination of clinician-delivered CBTp - difﬁculties with
service user engagement, and access to therapy.
Barrowclough, C., Johnston, M. and Tarrier, N. (1994). Attributions, expressed emotion, and patient
relapse: an attributional model of relatives’ response to schizophrenic illness. Behavior Therapy, 25,
Gaudiano, B. A. (2005). Cognitive behavior therapies for psychotic disorders: current empirical status
and future directions. Clinical Psychology: Science and Practice, 12, 33–50. doi:10.1093/clipsy/
George, S., Bandopadhay, S. and Cowan, C. (2005). A case report of informal cognitive therapy
of delusions by a carer: a novel treatment paradigm? European Psychiatry, 20, 80–81. doi:10.1016/
CBT for voices and a program for carers 123
Haddock, G., McCarron, J., Tarrier, N. and Faragher, E. B. (1999). Scales to measure dimensions
of hallucinations and delusions: the Psychotic Symptom Rating Scales (PSYRATS). Psychological
Medicine, 29, 879–889. doi:10.1017/S0033291799008661.
van Wijngaarden, B., Schene, A. H., Koeter, M., Vazquez-Barquero, J. L., Knudsen, H. C.,
Lasalvia, A., McCrone, P. and Group, E. S. (2000). Caregiving in schizophrenia: development,
internal consistency and reliability of the Involvement Evaluation Questionnaire-European Version:
EPSILON Study 4. British Journal of Psychiatry, 177, s21–s27. doi:10.1192/bjp.177.39.s21.
Wiedemann, G., Rayki, O., Feinstein, E. and Hahlweg, K. (2002). The Family Questionniare:
development and validation of a new self-report scale for assessing expressed emotion. Psychiatry
Research, 109, 265–279.