A systematic review examining factors predicting favourable outcome in cognitive behavioural interventions for psychosis

SCHRES-07050; No of Pages 9 Schizophrenia Research xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Schizophrenia Research journal homepage: www.elsevier.com/locate/schres

A systematic review examining factors predicting favourable outcome in cognitive behavioural interventions for psychosis O'Keeffe J ⁎, Conway R, McGuire B National University of Ireland, Galway, Ireland

a r t i c l e

i n f o

Article history: Received 12 September 2016 Received in revised form 12 November 2016 Accepted 14 November 2016 Available online xxxx Keywords: CBT Psychosis Factors predicting success

a b s t r a c t Psychosis is a debilitating mental health condition affecting approximately 4 persons per 1000. Cognitive behavioural therapy for psychosis (CBTp) has been shown to be an effective treatment for psychosis and is recommended by several national guidelines. CBTp does not work equally well with everyone, however, with some 50% of clients receiving little benefit. This review sets out to systematically assess the literature and methodological quality of a number of studies, which examine factors predicting successful outcome in CBTp. The databases CINAHL, Cochrane, EBSCO, EMBASE, ISI Web of Science, MEDLINE (Ovid), PsycARTICLES, PsycINFO, PubMed, and Scopus were electronically searched. English language articles in peer reviewed journals were reviewed. Search terms “psychosis”, “psychotic disorder”, “cognitive behavioural therapy”, “cognitive therapy”, “randomised controlled trial”, “predictor”, and “treatment outcome” in various combinations were used as needed. Only randomised controlled trials (RCTs) were included. Results suggest that female gender, older age, and higher clinical insight at baseline, each predicted better outcome in CBT interventions with psychotic patients, as did a shorter duration of the illness, and higher educational attainment. Several other factors, such as higher symptom severity at baseline, were suggestive of predictive capacity but further research to clarify was indicated. Providers of mental healthcare should consider these findings when offering CBTp. The onus is also on healthcare providers to better equip non-responders to CBTp. Further investigation into a limited number of predictive factors, with an agreed set of outcome measures, would allow future researchers more direct comparisons between studies. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Psychotic disorders are a group of illnesses characterized by the experience of a range of unusual, often distressing, mental and emotional events such as delusions and hallucinations. Psychosis can be experienced by individuals having diagnoses of schizophrenia, bipolar disorder, and psychotic depression, or the psychosis may be drug-induced. Traditionally, clinicians and researchers have categorised psychotic symptomology into positive symptoms and negative symptoms. Positive symptoms are characterized by prominent delusions, hallucinations, positive formal thought disorder, and persistently bizarre behaviour; negative symptoms, by affective flattening, avolition, and attentional impairment (Andreasen and Olsen, 1982). A recent systematic review conducted in the United Kingdom (Kirkbride et al., 2012) found the prevalence of psychosis in the population at 4 per 1000, with the annual incidence rate at 32 cases per 100,000. Cognitive behavioural therapy (CBT) has been shown to be an effective, evidence-based, treatment for many debilitating

⁎ Corresponding author. E-mail address: [email protected] (J. O'Keeffe).

psychological difficulties such as depression and anxiety (Butler et al., 2006). In the past two decades growing evidence has demonstrated the effectiveness of using CBT with a population experiencing psychosis. A recent review of the efficacy of CBT for psychosis (CBTp) (Turkington et al., 2013) concluded that CBTp can show robust effect and strong patient acceptability (Morrison et al., 2012a). Several national guidelines e.g. Morrison et al. (2004) have recommended CBTp as a first-defence treatment, alongside antipsychotic medication, for both first-episode psychosis and drug-resistant psychosis. In Ireland, the Health Service Executive (HSE), in a National Clinical Care Programme, fully adopted the NICE (2009) guidelines and recommended all patients experiencing first-episode psychosis be offered CBT for psychosis, alongside medical treatment. A nationwide training programme was implemented in 2013 to equip frontline staff with CBTp skills (HSE, 2013). Despite its demonstrated effectiveness, CBTp does not work equally well with everyone. In a recent review Lincoln et al. (2014a) report that, on average, 16% of patients discontinue CBTp, and of those who continue in therapy, approximately half do not show reliable symptom improvement (Wykes et al., 2008). Several studies have set out to examine which factors are predictive of positive outcome in CBTp with a view to optimising therapy provision.

http://dx.doi.org/10.1016/j.schres.2016.11.021 0920-9964/© 2016 Elsevier B.V. All rights reserved.

Please cite this article as: O'Keeffe, J., et al., A systematic review examining factors predicting favourable outcome in cognitive behavioural interventions for psychosis, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.11.021

2

J. O'Keeffe et al. / Schizophrenia Research xxx (2016) xxx–xxx

The process of setting out to identify variables which are predictive of therapy success is well established in the research literature. For example, Ciao et al. (2015) examined factors predicting and moderating outcome in Family Based Treatment for bulimia, and found that older adolescents made more rapid advances in self-esteem in therapy. Button et al. (2015), in a related finding, demonstrated that older age of adult clients favourably predicted response to CBT for depression. In contrast, Torp et al. (2015) reviewed predictive factors in CBT for paediatric OCD, and found that children and adolescents who were older, and had more severe OCD had significantly poorer outcomes after 14 weeks of treatment. Studies of this kind seek to identify which demographic, personal, psychological or neuropsychological factors make it more likely for an individual to benefit from a particular therapy delivered under particular settings.

2. Method

1.1. Predictors for successful CBT for psychosis

2.3. Inclusion and exclusion criteria

Many studies have attempted to identify which factors predict successful outcome in CBTp (see Lincoln et al. (2014b) for a review). However, the picture emerging from these studies is far from consistent. Researchers have not agreed on a set of outcome measures which can be consistently applied, or which baseline predictors to validate. Considering socio-demographic factors, some studies have found that younger patients benefit more in terms of positive symptoms (Morrison et al., 2012a; Thomas et al., 2011) and that higher educational attainment was shown to predict better outcome in negative symptoms (Allott et al., 2011). The extent to which patients have insight into their symptoms, that they can attribute their symptoms to mental illness, and the influence of this insight on outcomes have also been examined. Higher insight at baseline predicted symptom improvement in two studies (Garety et al., 1997; Naeem et al., 2008) and Brabban et al. (2009) found that symptom reduction was associated with lower delusional conviction. Clinical factors have also shown to be predictive of therapy effectiveness. Lower baseline symptomology and, in particular, lower negative symptomology were shown to be related to more symptomatic improvement during CBTp (Tarrier Beckett et al., 1993). Other potential predictor variables identified include; longer duration of untreated psychosis (Drury et al., 1996), greater self-reflectiveness (Perivoliotis et al., 2010), lower occupational functioning (Allott et al., 2011), and higher neuropsychological functioning (Premkumar et al., 2011). The pattern of exploration of predictor variables of CBTp to date has largely been one of parallel, independent studies, each focusing on specific, often unique, predictors using similar but varying outcome measures. Study methodologies vary and contradictory findings are not uncommon. For example, Tarrier Beckett et al. (1993) found that lower baseline symptomology predicted greater outcome for CBTp, whereas Morrison et al. (2012b) found that more severe positive symptoms at baseline predicted better outcome. To date there has not been an attempt to synthesise these findings using a systematic review. The timing of this current systematic review is important. International guidelines consistently recommend CBTp for patients with psychosis, yet CBTp is not as beneficial to some as to others. The more that is known about predictive factors, the better providers can target interventions towards patients more likely to benefit from treatment. Furthermore, by highlighting a population for whom CBTp is less effective, an informed re-evaluation of CBTp treatment modules would be possible, so that later versions of CBTp manuals would be more effective for a greater proportion of the population experiencing psychosis. Thus, the primary research question was: what baseline variables, if any, predict successful outcome in CBT for persons experiencing psychosis?

Randomised controlled trials (RCTs) comparing at least 2 levels of one clinical, demographic or personal predictive factor (e.g. High Baseline Insight vs. Low Baseline Insight), and examining the factor's effect on treatment outcome with a cognitive behavioural intervention for psychosis, were included in the review (see Table 1). No limits were imposed on year of publication. Only articles published in full, with English text were included. In refining the search of articles, any RCT in psychosis with an intervention naming cognitive behavioural therapy as an active component was included. This meant that not all studies included employed an exclusively manualised CBTp protocol. Of the 10 studies included in the final analysis, 5 studies used CBTp as an intervention (Allott et al., 2011; Garety et al., 1997; Haddock et al., 2006; Lincoln et al., 2014a; Tarrier Yusupoff et al., 1998). Two studies employed a brief CBT protocol (Brabban et al., 2009; Naeem et al., 2008). One study used a CBT Social Skills Training (Emmerson et al., 2009), one study focused on Workbased CBT (Kukla et al., 2014) and a further study utilised a CBT Coping Strategy Enhancement as the primary intervention (Tarrier Beckett et al., 1993). To ensure methodological rigour, only studies examining predictive factors which were selected prior to the CBT intervention were included.

2.1. Literature research A literature search was electronically conducted using CINAHL, Cochrane, EBSCO, EMBASE, ISI Web of Science, MEDLINE (Ovid), PsycARTICLES, PsycINFO, PubMed, and Scopus. No initial restrictions were placed on the year, or language of the studies. 2.2. Keywords used The main search terms included “psychosis”, “psychotic disorder”, “cognitive behavioural therapy”, “cognitive behavioural therapy”, “cognitive therapy”, “randomised controlled trial”, “randomized controlled trial”, “predictor”, and “treatment outcome” in various combinations as needed.

2.4. Data extraction The following variables were extracted from each of the 10 finalised studies: author, year, sample size, intervention, predictors tested, statistical significance, outcome measures, quality score, and secondary analysis. The summary extraction table is listed in Table 3. 2.5. Assessment of methodological quality The CONSORT checklist of RCT quality (Moher et al., 2012) was applied to each of the 10 studies (see Appendix 4). Seven of the included Table 1 Inclusion criteria. • Randomised controlled trials (RCTs), • Comparing at least 2 levels of one clinical, demographic or personal predictive factor, • Examining the factor's effect on treatment outcome with a cognitive behavioural intervention for psychosis, • No limits were imposed on year of publication, • Only articles published in full, with English text were included, • Any RCT in psychosis with an intervention naming cognitive behavioural therapy as an active component was included, • Only studies examining predictive factors which were selected prior to the CBT intervention were included.

Please cite this article as: O'Keeffe, J., et al., A systematic review examining factors predicting favourable outcome in cognitive behavioural interventions for psychosis, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.11.021

J. O'Keeffe et al. / Schizophrenia Research xxx (2016) xxx–xxx

3

scales were reported in two studies: Comprehensive Psychological Rating Scale (CPRS), Schedule of Assessment of Insight, Psychotic Symptoms Rating Scale (PSYRATS), and the Present State Examination (PSE).

studies were secondary analyses of larger, parent-studies. The level of methodological detail reported in these secondary studies was often minimal, thus a decision was made to also apply the CONSORT checklist to the parent studies (see Appendix 5) providing a more thorough analysis of methodological quality. Although some minor differences emerged in CONSORT scores between parent and secondary studies (see Table 2), these differences were not statistically different (t = 0.85, SD = 3.80, df = 7, p = 0.43), allowing the researcher to compare quality scores on each of the finalised 10 studies independently of their parent studies. The included studies were assessed against 37 criteria evaluating the quality of reporting under each of the following categories: Title and abstract, Introduction, Methods, Randomisation, Results, and Discussion. Within each category, items assessing characteristics of the study, such as trial design, statistical analyses, and clarity of reported results were included. All 10 studies assessed were randomised controlled trials (RCTs). The methodological quality of the studies ranged from 38% (Allott et al., 2011) to 65% (Tarrier Yusupoff et al., 1998), with an average quality value of 48%. Inter-rater reliability scores were obtained for 3 studies which were independently scored by another author. The kappa score of 0.90 indicated very good inter-rater reliability (Altman, 1999).

4. Predictor variable results 4.1. Demographic variables 4.1.1. Age Two studies suggest that older age of patient promotes better outcome from CBT. Haddock et al. (2006) in a study of 304 patients, found that older patients, over 21, did better in CBT, and that younger patients, under 21, responded better to counselling. In a related finding Kukla et al. (2014) with a sample of 50 participants, found that younger age of patient predicted poorer work engagement, and that older participants were more likely to stay longer in employment. Lincoln et al. (2014b) included age of patient in their analysis of predictor variables with a sample size of 73, and found no significant relationship between age and positive or negative symptoms. 4.1.2. Gender In a study with a sample size of 354, female patients were found to show greater overall symptom improvement and greater improvements in insight (Brabban et al., 2009). In contrast, Lincoln et al. (2014a) with a sample size of 73, found female gender to have a nonsignificant influence on both positive and negative symptoms of psychosis.

3. Results Study selection was undertaken by the first author (see p. 9, Fig. 1). 181 records were exported from 11 electronic databases. Forward and backward citation searches yielded 87 further records. Duplicates were removed and 222 records were searched by title. From these, 80 records contained no predictor variables and were excluded, 33 studies did not feature cognitive behavioural therapy, 56 did not use a population experiencing psychosis, and 42 records were excluded on the basis of methodological rigour – only studies employing a randomised controlled design were included. Two further studies were excluded following examination of the full text. One study (Dunn et al., 2006) did not employ RCT, and one study (Kemp et al., 1996) did not utilise a cognitive behavioural intervention. In the final analysis, 10 randomised controlled studies were examined. The numbers of subjects per study varied from 23 (Tarrier Beckett et al., 1993) to 354 (Brabban et al., 2009) with an average number of participants of 141. 23 predictor variables were reported in 10 studies. 15 variables appeared in separate studies and were not replicated. One variable ‘severity of symptoms’ appeared in five studies. ‘Age of patient’ and ‘insight’ both appeared in four studies, ‘delusion conviction’ and ‘education’ were considered in three studies, while ‘duration of disorder’, cognitive flexibility’, and ‘gender’ each featured in two studies. 45 outcome variable scales were reported in 10 studies, − only six scales were shared by two or more studies. The Brief Psychiatric Rating Scale (BPRS) was used by four studies. The Positive and Negative Syndrome Scale (PANSS) featured in three studies and each of the following

4.1.3. Educational achievement Higher educational achievement was found to be significant for negative symptoms in a study of 62 patients (Allott et al., 2011) and higher engagement in CBT protocol in a study with a sample size of 50 (Kukla et al., 2014) but non-significant for both positive and negative symptoms in one study of 73 patients (Lincoln et al., 2014b). 4.1.4. Occupational status People working or studying had better functional outcomes in one study (Allott et al., 2011). 4.1.5. Baseline functioning Global functioning at baseline was found by Allott et al. (2011) to be significantly related to improvement in positive symptoms. 4.2. Cognitive factors 4.2.1. Cognitive flexibility Contrasting findings exist in relation to the influence of patient's cognitive flexibility on outcomes. Garety et al. (1997) with a sample size of 60 found that cognitive flexibility predicted improvement in positive symptoms but Lincoln et al. (2014b) (N = 73) found no such relationship.

Table 2 CONSORT quality scores for parent and secondary studies. Parent study/year

Parent CONSORT score

N

Secondary study/year

2ndry CONSORT score

Turkington et al. (2002)

19

422

Sensky et al. (2000) Lysaker et al. (2009) Lewis et al. (2002) Jackson et al. 2008 Kuipers et al. (1997) Granholm et al. (2005) Mean

18 13 23 19 20 22 18.9a (51%)

90 50 304 62 60 76

Naeem et al. (2008) Brabban et al. (2009) Naeem et al. (2008) Kukla et al. (2014) Haddock et al. (2006) Allott et al. (2011) Garety et al. (1997) Emmerson et al. (2009)

17 22 17 19 19 14 17 17 17.8a (48%)

a

NS (t = 0.85, SD = 3.80, df = 7, p = 0.43).

Please cite this article as: O'Keeffe, J., et al., A systematic review examining factors predicting favourable outcome in cognitive behavioural interventions for psychosis, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.11.021

4

J. O'Keeffe et al. / Schizophrenia Research xxx (2016) xxx–xxx

Table 3 Data extraction.

Author

N

Design Intervention

Predictors tested

Significance Positive symptoms

Allott et al. (2011)

62

RCT

CBTp vs. befriending

Outcome measures Neg. symptoms

Premorbid adjustment

Emmerson et al. (2009)

Garety et al. (1997)

62

60

RCT

RCT

Brief CBT + TAU vs. TAU

Cognitive Behavioural Social Skills Training

R2 = 0.38; p = 0.003

CBTp + TAU vs. TAU

R2 = 0.26; p = 0.01

R2 = 0.35; p = 0.02

Overall symptoms

Insight

Female gender

p = 0.004

p = 0.04

Lower conviction of delusions

p = 0.02

Baseline Insight (BIS total) BIS insight factor “need for treatment” Cognitive flexibility

Number of recent admissions

Insight

304 RCT

CBT + TAU; counselling + TAU; TAU

50

RCT

Work-based CBT + TAU vs. TAU

Higher educational attainment Higher digit symbol scores Higher arithmetic scores Less severe negative symptoms

Hamilton Rating Scales for Depression

Counselling p = 0.04

p = 0.01

t (48) = 2.14, p = 0.04 F = 2.90, p = 0.04 F = 3.81, p = 0.03 F = 5.38, p = 0.02

14

Yes

22

Yes

17

No

17

Yes

19

Yes

Brief Psychiatric Rating Scale (BPRS) National Reading Test (NART) Amador Insight Scale

p b 0.05

Positive CBT engagement

Lincoln et al. (2014b)

Comprehensive Psychopathological Rating Scale Schedule of assessment of Insight Psychotic Symptoms Rating Scale Negative Symptoms Rating Scale Independent Living Skills Survey (ILSS)

p b 0.05

Younger client Older client

Brief Psychiatric Rating Scale (BPRS) R2 = 0.35; Scale for the Assessment p = 0.004 of Negative Symptoms (SANS) Delusions Inventory (PDI) Social and Occupational Functioning Assessment (SOFAS) R2 = 0.48; p = 0.002

b = 0.06, t (52) = 2.30, p b 0.05 b = 0.11, t (52) = 2.38, p b 0.05 p b 0.05

CBT Haddock et al. (2006)

Premorbid Adjustment Scale Wechsler Abbreviated Scale of Intelligence (WASI) Medication Adherence Rating Scale

R2 = 0.38; p = 0.003

Befriending group

Brabban et al. 354 RCT (2009)

Functional outcome

CBT group R2 = 0.19; Higher baseline p = 0.02 functioning (SOFAS) Higher educational achievement Lower avolition levels Working/studying

Quality max Secondary 37 analysis?

Poor work engagement

Insight increase p = 0.01

Present State Examination (PSE) Beck Depression/Anxiety Inventory (BDI/BAI) Maudsley Assessment of Delusion Schedule (MADS) Positive and Negative Syndrome Scale (PANSS) Psychotic Rating Scales (PSYRATS) Social Functioning Scale California Therapeutic Alliance Scale (CALPAS) Psychotherapy Status Report (PSR) Positive and Negative Syndrome Scale (PANSS) Rosenberg Self Esteem Scale (RSES) Wechsler Adult Intelligence Scale III (WAIS III) Wisconsin Card Sorting Test (WCST) Employment outcomes

Please cite this article as: O'Keeffe, J., et al., A systematic review examining factors predicting favourable outcome in cognitive behavioural interventions for psychosis, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.11.021

J. O'Keeffe et al. / Schizophrenia Research xxx (2016) xxx–xxx

5

Table 3 (continued)

Author

N

Design Intervention

Predictors tested

Significance Positive symptoms

Lower arithmetic scores Younger age

Naeem et al. (2008)

Tarrier Beckett et al. (1993)

73

RCT

353 RCT

23

RCT

CBTp vs. wait

Brief CBTp vs. TAU

Coping Strategy Enhancement vs. problem solving vs. TAU

Female gender

Positive symptoms NS

F = 5.37, p = 0.01 Neg. symptoms NS

Age

NS

NS

Years of education NS

NS

Duration of disorder Comorbid axis 1 disorder Comorbid axis 2 disorder Symptom severity (PANSS) Delusion conviction (PDI) Lack of insight (PANSS-G-12) Depression Social skills Cognitive flexibility (TMT-B) Memory Jumping to conclusions Internal attributions of neg. events Theory of mind CBT treatment group High symptom severity (CPRS) Insight

NS

p = 0.02

NS

NS

NS

NS

Wechsler Memory Scale (Logical Memory 1) Trail Making Test (TMT)

NS

NS

Beads Test

NS

NS

NS

NS

NS NS NS

NS NS NS

NS NS

NS p = 0.04

NS

NS

NS p = 0.01

NS

RCT

CBT vs. supportive counselling vs. TAU

p = 0.01 p = 0.03

Low delusion rating

p = 0.02

Group membership (CSE or PS)

p = 0.003

Higher pre-treatment scores

72 Tarrier Yusupoff et al. (1998)

Functional outcome

Wald = 8.01, p = 0.00 Wald = 7.31, p = 0.00 F = 7.91, p = 0.008

More severe positive symptoms Lower self esteem

Lincoln et al. (2014a)

Outcome measures Neg. symptoms

Quality max Secondary 37 analysis?

p = 0.0001

Allocation to CBT

p = 0.005

Short duration of illness Low severity of symptoms

p = 0.008 p = 0.02

19

Yes

15

No

17

Yes

19

No

24

No

Structured Clinical Interview for DSM-IV (SCID) Positive and Negative Syndrome Scale (PANSS) Calgary Depression Rating Scale Delusions Inventory (PDI)

Comprehensive Rating Scale (CPRS) David's Insight Scale Health of the Nation Outcome Scale (HoNOS)

Present State Examination (PSE) Psychiatric Assessment Scale (PAS) Brief Psychiatric Rating Scale (BPRS) Special Functioning Scale (SFS) Brief Psychiatric Rating Scale (BPRS) Present State Examination (PSE)

Italics refer to 'p' values. Bold for quality scores on CONSORT.

Please cite this article as: O'Keeffe, J., et al., A systematic review examining factors predicting favourable outcome in cognitive behavioural interventions for psychosis, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.11.021

6

J. O'Keeffe et al. / Schizophrenia Research xxx (2016) xxx–xxx

Fig. 1. PRISMA flow-chart of the study inclusion process. PRISMA 2009 flow diagram.

4.3. Clinical factors

4.2.2. Insight Four studies examined the influence of patient's insight on the effectiveness of CBT interventions. In a study of 62 patients with psychosis, Emmerson et al. (2009) found that a patient's Baseline Insight, and also their response to a specific Insight factor acknowledging a need for treatment, predicted their improvement in positive symptomology. Similarly, higher patient insight prior to treatment was predictive of better outcome in positive symptoms in two further studies with both small sample size (N = 60) (Garety et al., 1997), and large sample size (N = 353) (Naeem et al., 2008). One study (N = 73) (Lincoln et al., 2014a) found no relationship between level of insight and positive or negative symptomology. The divergence in these results may be accounted for by the methodology employed. Lincoln et al. (2014b) urged caution in interpreting their single findings, due to the numerous single regression analyses that were not adjusted to reduce type 1 error. Any results from this study should therefore be interpreted with caution. Since these methodological weaknesses are likely to have affected the results of the insight factors, it is reasonable to conclude that higher insight is a predictive factor of CBT outcome in psychosis.

4.3.3. Delusional conviction One study found that lower delusional conviction was shown to be significantly predictive of reduction in overall psychotic symptoms (Brabban et al., 2009). A contrasting finding was reported by Lincoln et al. (2014a) who observed no relationship between delusional conviction and response in positive or negative symptoms.

4.2.3. Neuropsychological factors Kukla et al. (2014) found that higher digit symbol scores and higher arithmetic scores both predicted positive CBT engagement. Lincoln et al. (2014a) found no predictive value in measures of memory or of theory of mind.

4.3.4. Number of recent admissions One study (Garety et al., 1997) found a significant relationship between greater number of recent hospital admissions and improvement in positive symptomology. This variable was not assessed in any of the other nine studies.

4.3.1. Avolition levels Lower levels of avolition were found to be associated with improvement in negative symptoms in one study (Allott et al., 2011) but were not included in any of the other studies. 4.3.2. Duration of illness Two studies found that a shorter duration of illness predicted positive outcome (Tarrier Yusupoff et al., 1998) and improvement of negative symptoms (Lincoln et al., 2014b).

Please cite this article as: O'Keeffe, J., et al., A systematic review examining factors predicting favourable outcome in cognitive behavioural interventions for psychosis, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.11.021

J. O'Keeffe et al. / Schizophrenia Research xxx (2016) xxx–xxx

4.3.5. Severity of symptoms Two moderate sized studies (N = 50 + 72) found that patients with less severe psychotic symptomology at baseline fared better in CBT interventions (Kukla et al., 2014; Tarrier Yusupoff et al., 1998). In direct contrast, two studies, one with a large sample (N = 353), (Naeem et al., 2008), and one with a small sample (N = 23) (Tarrier Beckett et al., 1993) found that more severe psychotic symptomology at baseline was predictive of better outcome. A further study with a sample size of 73 found no relationship between symptom severity and outcome (Lincoln et al., 2014b). It is also possible to add Garety et al.'s (1997) above study in support of the contention that more severe baseline psychotic symptomology, as expressed by more recent hospital admissions, is predictive of CBT success, particularly for positive symptoms. Symptom severity appears not to be a strong predictor of CBT outcome, but greater symptom severity at baseline appears marginally more predictive of CBT outcome than lower baseline symptomology. Overall, previous studies examining variables predicting CBT outcome in treatment of psychosis have looked at various demographic, clinical and cognitive factors of patients experiencing psychosis. Female patients tended to fare better, as did older patients, and those with higher education and higher insight into their illness. Shorter duration of illness was also found to be predictive and severity of psychotic symptomology, while not entirely predictive, was not a deterrent to successful outcome with CBT for psychosis. 5. Discussion This current systematic review set out to examine the demographic, clinical and cognitive factors which may predict successful outcome in CBT with patients experiencing psychosis. The findings of this review suggest that it may be possible to identify factors which predict better response to cognitive behavioural interventions with adults experiencing psychosis. Many factors have shown predictive value in some studies but not in others. Other factors have been examined in only one study and thus their generalisability is limited. In this section, we will examine some of the possible reasons for discrepancy in findings, and where contradictory findings occur, the methodological quality of the opposing studies will be used to arbitrate between them. We will use this information to consider whether any factor or set of factors are more reliably predictive of CBT success, and, also suggest possible direction for future research in this area. Insight has long been considered an important component in an individual's experience of mental illness (Burton et al., 2011). Insight has been divided into clinical insight, which concerns the more traditional factors of an awareness of having a mental disorder, and a recognition of the need for treatment (Amador et al., 1993) and cognitive insight which involves a person's capacity to evaluate unusual experiences and question their own style of thinking (Beck et al., 2004). In this review, three studies found clinical insight to be predictive of CBT success (Emmerson et al., 2009; Garety et al., 1997; Naeem et al., 2008). These studies used validated instruments specifically and solely designed to measure insight; the Birchwood Insight Scale (Birchwood et al., 1994), David's Insight Scale (David et al., 1992), and the Amador Insight Scale (Amador et al., 1993). The studies employed a large combined sample size and each study scored 46% on the CONSORT quality scale. The one study (Lincoln et al., 2014a) which found no link between lack of insight and CBT success used a relatively small sample size, used one of the 32 sub-scales of the Positive and Negative Syndrome Scale (Kay et al., 1987) to assess insight, and scored 41% on the CONSORT scale. Considering the consistent findings across three validated scales, specifically targeting insight, in combination with the methodological quality scores, it is reasonable to conclude that clinical insight is a predictive factor for successful outcome in CBT for psychosis. The findings on symptom severity pose some difficulties in interpretation. Two studies (Kukla et al., 2014; Tarrier Yusupoff et al., 1998) found that lower baseline symptom severity was predictive of CBT

7

success, two studies found that higher symptom severity predicted better outcomes (Naeem et al., 2008; Tarrier Beckett et al., 1993) and one study (Lincoln et al., 2014b) found no relationship between symptom severity and outcome. A further study (Garety et al., 1997) found that response to CBT is best predicted at baseline by a higher rate of recent admissions, implying a higher symptom severity rating. Considering sample sizes and methodological quality: three studies with an average CONSORT quality score of 48% found higher symptom severity to be predictive, and two studies, with an average CONSORT score of 58% found lower baseline symptom severity to be predictive. Since all studies used similar outcome measures, it must be assumed that baseline symptom severity is not a consistently predictive variable in determining CBT success with psychosis, though there is some modest indication that higher symptom severity is likely to result in better outcomes. It is possible that patients with higher initial levels of psychotic symptomology showed more rapid improvement and that those patients with more moderate initial symptomatology had less scope to improve, because they were operating closer to their optimal level of functioning. Either way, it should be incorporated into best practice that severity of symptoms should not be used as a contra-indication of CBT for psychosis. The age (Haddock et al., 2006; Kukla et al., 2014) and gender (Brabban et al., 2009) of the patient have been shown to be predictive of CBT success in relatively large-scale studies. One study, however, (Lincoln et al., 2014a) failed to find any relationship between these factors and CBT success. In fact, Lincoln et al. (2014b) failed to find significant predictive value in many factors which other studies have found to be predictive. Listed amongst the factors Lincoln et al. (2014a) found insignificant were: gender, age, years of education, comorbid axis 1 or 2 disorder, symptom severity, delusional conviction, insight, depression, social skills, cognitive flexibility, memory, internal attributions, and theory of mind. Since this one study is a counterpoint to many other, often larger, and more methodologically sound studies, it is worthwhile examining its methodological strengths and weaknesses. The Lincoln study scored 41% on the CONSORT quality analysis. This is in contrast with 51% for each of the studies examining age and 59% for the Brabban et al. (2009) study examining gender. The sample size used by Lincoln of 72, though small, was comparable to samples from previous CBTp trials. The trial lacked a control condition for follow-up. Perhaps more seriously Lincoln et al. (2014b) urge caution in interpreting their single findings due to risk of type 1 error. These methodological weaknesses are likely to have affected the results of the age and gender factors, and allied to the CONSORT findings, it is reasonable to conclude that female gender and older age (N21) are predictive factors of CBT outcome in psychosis. If female gender is more responsive to CBTp this presents some treatment problems to clinicians, since a greater proportion of males, particularly young males present with psychosis (Spauwen et al., 2003). This may partly be explained by higher premorbid functioning and social functioning, and a higher age of onset in females (Ochoa et al., 2012). There may be a case to be made for gender-specific programmes to be developed. Shorter duration of illness was found to be predictive of outcome, both on positive symptoms (Tarrier Yusupoff et al., 1998) and negative symptoms (Lincoln et al., 2014a). This highlights the need for early intervention and it lends support to community outreach approaches targeted to identify early psychosis in the community. Educational achievement was shown to be predictive of outcome in two studies (Allott et al., 2011; Kukla et al., 2014), with an average CONSORT score of 45% but (Lincoln et al., 2014b), with a CONSORT score of 41% (see above) found no predictive value. Given the caveats outlined above in the Lincoln study, it is reasonable to conclude that higher educational achievement has some predictive power in CBT for psychosis. Other factors are listed in only one study and include: occupational status, lower avolition and global baseline functioning (Allott et al., 2011), and number of recent admissions (Garety et al., 1997). These findings must be treated with caution and definitive commentary awaits replication in further studies. Some factors featured two studies

Please cite this article as: O'Keeffe, J., et al., A systematic review examining factors predicting favourable outcome in cognitive behavioural interventions for psychosis, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.11.021

8

J. O'Keeffe et al. / Schizophrenia Research xxx (2016) xxx–xxx

with directly opposed results: cognitive flexibility, delusional conviction and neuropsychological factors. Again, further, targeted research is required to arrive at a conclusive verdict as to the predictive value of these factors. 5.1. Clinical and policy implications This systematic review will have some practice implications for clinicians, and policy implications for service managers. When streamlining patients chosen to receive CBT interventions, consideration should be given in allocation of such scarce resources. In triaging cases, the predictive factors outlined above should be used as a guide. In prioritising those patients for whom CBTp is more likely to be effective, it should also be a priority for service planners to provide input for individuals less likely to respond well. Both could be achieved by including a component in all screenings for CBTp assessing patient insight. If the screening reveals low insight at baseline, an insight-orientated block of therapy should be offered prior to commencement of CBTp proper. Assuming an increase in insight from the pre-CBTp block, each patient would then be best placed to avail of the full benefits of CBTp. Other policy implications may include a review of the national guidelines for treatment of psychosis. Rather than offering CBTp universally to all patients experiencing psychosis, a more individualised, client-centred approach is indicated, taking the above-mentioned predictive factors into account. For example, using age as a predictive factor, Haddock et al. (2006) demonstrated that CBTp was effective for older patients (N21), but for younger patients with psychosis (b 21), counselling had better outcomes. The importance of early detection of psychosis has again been highlighted in this current review. CBTp is more effective for patients with a shorter duration of illness. Future mental health policy should see resources targeted into identification of prodromal signs of psychosis in schools and colleges, and swifter access for patients with early psychosis to medical and psychological therapies. 5.2. Limitations This study had some limitations which may affect the strength of its findings. The searches were undertaken solely by the first author, who is a trained CBTp practitioner, thus selection bias cannot be ruled out. The studies included used many different outcome measures, and a wide variety of predictive factors, making valid comparisons between studies more difficult and more open to interpretation. The scale and variety of outcome variables used by authors in this study contrast with other, more recent studies of CBTp where authors narrow the outcome measures to a smaller number, such as reduction of distress and negative emotion (Lincoln et al., 2016). Future systematic reviews targeting CBTp effectiveness may yield more focused results and facilitate a better match between patient need and available resources. The focus of this current review was on available studies using CBTp for a broad spectrum of psychosis. More recent approaches have targeted single symptoms of psychosis with a high degree of reported success. Freeman's (2016) ‘Feeling Safe’ programme for paranoia selects persecutory delusions as the target of intervention and uses exposure and modification of maintenance beliefs to bring about change. Similarly, Birchwood et al. (2014) reported treatment success targeting command hallucinations using CBT. The outcome of studies focusing on a single symptom of the spectrum of psychosis, and factors predicting their outcome is a promising area which will merit systematic review when the available papers accumulate. By limiting the scope of the review to randomised controlled studies only, some important aspects of the relationship between predictive factors and psychosis could not be explored. The mechanism, for example, by which higher clinical insight promotes greater CBT outcome, was beyond the scope of this current review. Furthermore, non-specific factors such as therapeutic alliance (Dunn et al., 2006) remain outside of our analysis.

5.3. Conclusion In summary, this systematic review yielded a mix of results with a few consistent findings. It concluded that female gender, older age, and higher clinical insight at baseline each predicted better outcome in CBT interventions with psychotic patients, as did a shorter duration of the illness, and higher educational attainment. Higher symptom severity at baseline seemed to yield better outcome, but the results here are, at best, mixed. Factors examined in single studies, while promising, lack the generalisability to be utilised in routine practice, and could not be validated in this systematic review. Future studies should apply randomised controlled studies to those factors we were unable to confirm as predictive in this study, and should include studies targeting single symptoms of psychosis. Factors such as occupational status, baseline functioning, cognitive flexibility, avolition and neuropsychological functioning each could add to the emerging profile of responders to CBT for psychosis. References Allott, K., Alvarez-Jimenez, M., Killackey, E.J., Bendall, S., McGorry, P.D., Jackson, H.J., 2011. Patient predictors of symptom and functional outcome following cognitive behaviour therapy or befriending in first-episode psychosis. Schizophr. Res. 132 (2–3), 125–130. Altman, D.G., 1999. Practical Statistics for Medical Research. London Chapman and Hall. Amador, X.F., Strauss, D.H., Yale, S.A., Flaum, M.M., Endicott, J., Gorman, J.M., 1993. Assessment of insight in psychosis. Am. J. Psychiatr. 150 (6), 873–879. Andreasen, N.C., Olsen, S., 1982. Negative v positive schizophrenia: definition and validation. Arch. Gen. Psychiatry 39 (7), 789–794. Beck, A.T., Baruch, E., Balter, J.M., Steer, R.A., Warman, D.M., 2004. A new instrument for measuring insight: the Beck Cognitive Insight Scale. Schizophr. Res. 68 (2–3), 319–329. Birchwood, M., Michail, M., Meaden Tarrier, N., Lewis, S., Wykes, T., Davies, L., Dunn, G., Peters, E., 2014. Cognitive behaviour therapy to prevent harmful compliance with command hallucinations (COMMAND): a randomised controlled trial. Lancet Psychiatry 1, 23–33. Birchwood, M., Smith, J., Drury, V., Healy, J., Macmillan, F., Slade, M., 1994. A self-report insight scale for psychosis: reliability, validity and sensitivity to change. Acta Psychiatr. Scand. 89 (1), 62–67. Brabban, A., Tai, S., Turkington, D., 2009. Predictors of outcome in brief cognitive behavior therapy for schizophrenia. Schizophr. Bull. 35 (5), 859–864. Burton, C.Z., Vella, L., Twamley, E.W., 2011. Clinical and cognitive insight in a compensatory cognitive training intervention. Am. J. Psychiatr. Rehabil. 14 (4), 307–326. Butler, A.C., Chapman, J.E., Forman, E.M., Beck, A.T., 2006. The empirical status of cognitive-behavioral therapy: a review of meta-analyses. Clin. Psychol. Rev. 26 (1), 17–31. Button, K.S., Turner, N., Campbell, J., Kessler, D., Kuyken, W., Lewis, G., ... Wiles, N., 2015. Moderators of response to cognitive behavioural therapy as an adjunct to pharmacotherapy for treatment-resistant depression in primary care. J. Affect. Disord. 174, 272–280. Ciao, A.C., Accurso, E.C., Fitzsimmons-Craft, E.E., Le Grange, D., 2015. Predictors and moderators of psychological changes during the treatment of adolescent bulimia nervosa. Behav. Res. Ther. 69, 48–53. David, A., Buchanan, A., Reed, A., Almeida, O., 1992. The assessment of insight in psychosis. Br. J. Psychiatry 161, 599–602 (NOV.). Drury, V., Birchwood, M., Cochrane, R., Macmillan, F., 1996. Cognitive therapy and recovery from acute psychosis: a controlled trial. II. Impact on recovery time. Br. J. Psychiatry 169 (5), 602–607. Dunn, H., Morrison, A.P., Bentall, R.P., 2006. The relationship between patient suitability, therapeutic alliance, homework compliance and outcome in cognitive therapy for psychosis. Clin. Psychol. Psychother. 13 (3), 145–152. Emmerson, L.C., Granholm, E., Link, P.C., McQuaid, J.R., Jeste, D.V., 2009. Insight and treatment outcome with cognitive-behavioral social skills training for older people with schizophrenia. J. Rehabil. Res. Dev. 46 (8), 1053–1058. Freeman, D., 2016. Persecutory delusions: a cognitive perspective on understanding and treatment. Lancet Psychiatry 3, 685–692. Granholm, E., McQuaid, J.R., McClure, F.S., Auslander, L.A., Perivoliotis, D., Pendrelli, P., Patterson, T., Jests, D.V., 2005. A randomized, controlled trial of cognitive behavioral social skillls training for middle-aged and older outpatients with chronic schizophrenia. Am. J. Psychiatry 162, 520–529. Garety, P., Fowler, D., Kuipers, E., Freeman, D., Dunn, G., Bebbington, P., ... Jones, S., 1997. London-East Anglia randomised controlled trial of cognitive-behavioural therapy for psychosis. II: predictors of outcome. Brit. J. Psych. 171, 420–426. Haddock, G., Lewis, S., Bentall, R., Dunn, G., Drake, R., Tarrier, N., 2006. Influence of age on outcome of psychological treatments in first-episode psychosis. Br. J. Psychiatry 188, 250–254. Health Services Executive (HSE), 2013. National Clinical Programme for Mental Health. Department of Health, Ireland. Jackson, H.T., McGarry, P.D., Killackey, E., Bendall, S., Allot, K., Dungeon, P., Gleeson, J., Johnson, T., Harrington, S., 2008. Acute -phase and 1-year follow up results of a randomized controlled trial of CBT versus befriending for first-episode psychosis: the ACE project. Psychol. Med. 38 (5), 725–735.

Please cite this article as: O'Keeffe, J., et al., A systematic review examining factors predicting favourable outcome in cognitive behavioural interventions for psychosis, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.11.021

J. O'Keeffe et al. / Schizophrenia Research xxx (2016) xxx–xxx Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The Positive and Negative Syndrome Scale (PANSS) for schizophrenia. Schizophr. Bull. 13 (2), 261–276. Kemp, R., Hayward, P., Applewhaite, G., Everitt, B., David, A., 1996. Compliance therapy in psychotic patients: randomised controlled trial. BMJ 312 (7027), 345–349. Kirkbride, J.B., Errazuriz, A., Croudace, T.J., Morgan, C., Jackson, D., Boydell, J., ... Jones, P.B., 2012. Incidence of schizophrenia and other psychoses in England, 1950–2009: a systematic review and meta-analyses. PLoS One 7 (3). Kuipers, E., Garety, P., Fowler, D., Dunn, G., Bebbington, P., Freeman, D., Hadley, C., 1997. London-East Anglia randomised controlled trial of cognitive-behavioural therapy for psychosis: 1: effects of the treatment. British Journal of Psychiatry 171, 319–327. Kukla, M., Davis, L.W., Lysaker, P.H., 2014. Cognitive behavioral therapy and work outcomes: correlates of treatment engagement and full and partial success in schizophrenia. Behav. Cogn. Psychother. 42 (5), 577–592. Lewis, S., Tarrier, N., Haddock, G., Bentall, R., Kinderman, P., Kingdon, D., Siddle, R., Drake, R., Everitt, J., Leadley, K., Benn, A., Grazebrook, K., Haley, C., Akhtar, S., Davies, L., Palmer, S., Faragher, B., Dunn, G., 2002. Randomised controlled trial of cognitive-behavioural therapy in early schizophrenia: acute-phase outcomes. British Journal of Psychiatry 181, s91–s97 SUPPL43. Lincoln, T.M., Jung, E., Wiesjahn, M., Schlier, B., 2016. What is the minimal dose of cognitive behavior therapy for psychosis? An approximation using repeated assessments over 45 sessions. Eur. Psychol. 38, 31–39. Lincoln, R.W., Westermann, S., Ziegler, M., Kesting, M.L., Heibach, E., Mehl, S., 2014b. Who stays, who benefits? Predicting dropout and change in cognitive behaviour therapy for psychosis. Psychiatry Res. 216 (2), 198–205. Lincoln, T.M., Rief, W., Westermann, S., Ziegler, M., Kesting, M.-L., Heibach, E., Mehl, S., 2014a. Who stays, who benefits? Predicting dropout and change in cognitive behaviour therapy for psychosis. Psychiatry Res. 216 (2), 198–205. Lysaker, P.H., Davis, L.W., Bryson, G.J., Bell, M.D., 2009. Effects of cognitive behaviuoral therapy on work outcomes in vocational rehabilitation for participants with schizophrenia spectrum disorders. Schizophr. Res. 107, 186–191. Moher, D., Hopewell, S., Schulz, K.F., Montori, V., Gøtzsche, P.C., Devereaux, P.J., ... Altman, D.G., 2012. CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials. Int. J. Surg. 10 (1), 28–55. Morrison, A.P., French, P., Stewart, S.L., Birchwood, M., Fowler, D., Gumley, A.I., ... Dunn, G., 2012a. Early detection and intervention evaluation for people at risk of psychosis: multisite randomised controlled trial. BMJ 344. Morrison, A.P., Hutton, P., Wardle, M., Spencer, H., Barratt, S., Brabban, A., ... Turkington, D., 2012b. Cognitive therapy for people with a schizophrenia spectrum diagnosis not taking antipsychotic medication: an exploratory trial. Psychol. Med. 42 (05), 1049–1056. Morrison, A.P., Renten, J.C., Williams, S., Dunn, H., Knight, A., Krautz, M., ... Dunn, G., 2004. Delivering cognitive therapy to people with psychosis in a community mental health setting: an effectiveness study. Acta Psychiatr. Scand. 110 (1), 36–44.

9

Naeem, F., Kingdon, D., Turkington, D., 2008. Predictors of response to cognitive behaviour therapy in the treatment of schizophrenia: a comparison of brief and standard interventions. Cogn. Ther. Res. 32 (5), 651–656. National Institute for Clinical Excellence (NICE), 2009. Schizophrenia: core Interventions in the Treatment and Management of Schizophrenia in Primary and Secondary Care (Update). National Collaborating Centre for Mental Health, (UK). Ochoa, S., Usall, J., Cobo, J., Labad, X., Kulkarni, J., 2012. Gender differences in schizophrenia and first-episode psychosis: a comprehensive literature review. Schizophrenia Research and Treatment, pp. 2–9 (Article ID 916198). Perivoliotis, D., Grant, P.M., Peters, E.R., Ison, R., Kuipers, E., Beck, A.T., 2010. Cognitive insight predicts favorable outcome in cognitive behavioral therapy for psychosis. Psychosis 2 (1), 23–33. Premkumar, P., Peters, E.R., Fannon, D., Anilkumar, A.P., Kuipers, E., Kumari, V., 2011. Coping styles predict responsiveness to cognitive behaviour therapy in psychosis. Psychiatry Res. 187 (3), 354–362. Sensky, T., Turkington, D., Scott, J.L., Scott, J., Siddle, R., O'Carroll, M., Barnes, T.R.E., 2000. A randomized controlled trial of cognitive-behavioural therapy for persistent symptoms in schizophrenia resistent to medication. Arch. Gen. Psychiatry 57 (2), 165–172. Spauwen, J., Krabbendam, L., Lieb, R., Wittchen, H.U., van Os, J., 2003. Sex differences in psychosis: normal or pathological? Schizophr. Res. 62 (1), 45–49. Tarrier Beckett, R., Harwood, S., Baker, A., Yusupoff, L., Ugarteburu, I., 1993. A trial of two cognitive-behavioural methods of treating drug-resistant residual psychotic symptoms in schizophrenic patients: I. Outcome 162. Tarrier Yusupoff, L., Kinney, C., McCarthy, E., Gledhill, A., Haddock, G., Morris, J., 1998. Randomised controlled trial of intensive cognitive behaviour therapy for patients with chronic schizophrenia. Br. Med. J. 317 (7154), 303–307. Thomas, N., Rossell, S., Farhall, J., Shawyer, F., Castle, D., 2011. Cognitive behavioural therapy for auditory hallucinations: effectiveness and predictors of outcome in a specialist clinic. Behav. Cogn. Psychother. 39 (2), 129–138. Torp, N.C., Dahl, K., Skarphedinsson, G., Compton, S., Thomsen, P.H., Weidle, B., Ivarsson, T., 2015. Predictors associated with improved cognitive-behavioral therapy outcome in pediatric obsessive-compulsive disorder. J. Am. Acad. Child Adolesc. Psychiatry 54 (3), 200–207. Turkington, D., Kingdon, D., Turner, T., 2002. Effectiveness of a brief cognitive-behavioural therapy intervention in the treatment of schizophrenia. British Journal Of Psychiatry 180, 523–527. Turkington, D., Wright, N., Tai, S., 2013. Advances in cognitive behavior therapy for psychosis. Int. J. Cogn. Ther. 6 (2), 150–170. Wykes, T., Steel, C., Everitt, B., Tarrier, N., 2008. Cognitive behavior therapy for schizophrenia: effect sizes, clinical models, and methodological rigor. Schizophr. Bull. 34 (3), 523–537.

Please cite this article as: O'Keeffe, J., et al., A systematic review examining factors predicting favourable outcome in cognitive behavioural interventions for psychosis, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.11.021