Pilot programme of modular symptom-specific group cognitive behaviour therapy in a ‘Real World’ early intervention in psychosis service

Authors

  • David Raune,

    Corresponding author
    • Harrow and Hillingdon Early Intervention in Psychosis Service, Central and Northwest London Foundation NHS Trust, Ruislip Manor, UK
    Search for more papers by this author
  • Suzanne Law

    1. Harrow and Hillingdon Early Intervention in Psychosis Service, Central and Northwest London Foundation NHS Trust, Ruislip Manor, UK
    Search for more papers by this author

Corresponding author: Dr David Raune, Harrow and Hillingdon Early Intervention in Psychosis Service, Central and Northwest London Foundation NHS Trust, The Pembroke Centre, 90 Pembroke Road, Ruislip Manor HA4 8NQ, UK. Email: david.raune@nhs.net

Abstract

Aim

A modular symptom-specific (MSS) programme of early intervention group cognitive behaviour therapy for psychosis might confer additional therapeutic benefits and clinical and financial efficiency, but the approach is empirically untested. Therefore, we devised a novel MSS programme to test – in a culturally diverse ‘Real World’ early intervention service – its relevancy, feasibility, acceptability and safety.

Method

The MSS programme comprised six different groups across 33 sessions: Psycho-Education (3), Mood-Management (5), Delusions (8), Auditory Hallucinations (8), Past Auditory Hallucinations (3) and Negative Symptoms (6). All patients were eligible for psycho-education, but other groups required a specific symptom.

Results

Patients (n = 166, 37% White) showed wide heterogeneity in the activity, type and number of symptoms, and group-relevant symptoms were common. Eighteen groups have run so far, each type of group at least once, 58/166 (35%) of patients attended across 281/412 (68%) group sessions, 46/58 (79%) of patients re-attended, and the 58 patients attended groups a mean of 4.8 times. Patients were significantly (P < 0.05) more likely to attend the programme if they had a schizophrenia diagnosis, a longer psychotic illness length or one active group-relevant (psychotic/mood) symptom. Diagnostically uncertain psychosis patients were almost significantly (P = 0.05) less likely to attend the programme. duration of untreated psychosis and ethnicity were unrelated to attendance/re-attendance. No group-related risk incidents occurred.

Conclusion

A MSS programme approach is highly relevant to early intervention symptom heterogeneity and is feasible, culturally acceptable and safe; it also appears particularly efficient. Future research should now test its clinical efficacy.

Introduction

Early intervention group cognitive behaviour therapy for psychosis (EI-GCBTp) is an emerging treatment option involving therapeutic factors potentially superior to individual CBT at the EI stage.[1] Three EI-GCBTp programmes have shown that EI-GCBTp can lower positive and/or negative symptom of psychosis, but these studies used a treatment approach that involved patients attending one quite lengthy (16, 18 or 24 h) generic group, which included patients with heterogeneous symptom types.[2-4]

A quite different approach would be to use a modular symptom-specific (MSS) programme of EI-GCBTp in which all patients would be eligible for a CBT for psychosis psycho-education group, but patients would be streamlined for other groups according to symptomatic presentations such as symptom type, activity/remission, or treatment resistance, etc. This MSS approach would provide greater group therapeutic homogeneity[5, 6] and would necessitate patients attending fewer sessions that are not relevant to them, so might be more clinically and financially efficient, as well as perhaps more acceptable and ethical. Promising results have been obtained from several separate small symptom-specific GCBTp studies in chronic samples[6-11] and at EI;[12, 13] however, a whole programme MSS approach to GCBTp has never been empirically tested in psychosis, including at EI.

Barriers to implementing a MSS approach are that the approach is dependent on sufficient prevalences of group-relevant symptoms for which CBT has an evidence base, and sufficient heterogeneity in symptom presentation to make a MSS approach a better choice than a single-group-for-all approach. Also, it is unclear if sufficient patients, including those from different cultures,[14] would be willing to attend groups that are explicitly problem symptom-focused, for example because of shame, or not re-attend, for example because of the concentrated intensity of a symptom-specific approach. Novel symptom-specific groups such as GCBTp for single delusions or negative symptoms at the EI stage also merit some risk monitoring. Therefore, we devised a novel MSS EI-GCBTp programme to test its relevancy, feasibility, acceptability and safety in a culturally diverse young recent onset sample.

Method

The EI in Psychosis (EIP) service

The EIP service has a catchment area containing 449 820 people in North/West London, UK.[15] Cases of first-episode psychosis can be accepted if the duration of untreated psychosis (DUP) is under 12 months (or 6 months DUP for the patients who were accepted within the first year of the EIP service's inception) and the patient is aged 14–35 years (119 138 local people).

MSS groups overview

The intervention was devised by the EI team Clinical Psychologist who is a UK BABCP Accredited Cognitive Therapist and Supervisor who holds a PhD in first-episode psychosis (Kings College, London) and a post-graduate diploma with Distinction in CBT (Oxford University). Based on our previous symptom prevalence survey,[16] a programme was created consisting of six separate 1-h CBT groups across 33 group sessions, which attempted to provide some accommodation for symptom type, activity and treatment resistance: Psych-Education (three sessions), Mood-Management (five sessions), Past Auditory Hallucinations (three sessions), Current Persistent Distressing Auditory Hallucinations (eight sessions), Current Persistent Distressing Single Delusion (eight sessions), and Current Persistent Negative Symptoms (six sessions). The therapists were the clinical psychologist (DR) and assistant psychologist (SL) with clinical and/or audit assistance from temporary trainee clinical psychologists who passed through the EI service.

Inclusion/exclusion criteria

All patients were given The International Classification of Disease- version 10 diagnosis of psychosis by the Consultant or Speciality Psychiatrist. All patients were eligible to attend the group programme except those who do not speak English, have more than a mild learning disability, are likely to be disruptive to the group, or where it is judged that the group may increase rather than decrease symptomatic risk. The CBT Psycho-Education group was open to all patients; the Past Auditory Hallucinations group was open to patients who have had auditory hallucinations, but who are in symptomatic remission; Mood-Management groups were open to patients who currently had mood symptoms. The three active psychotic symptom groups were reserved for patients who had persistent (at least 3 months) distressing/life-interfering active delusions, auditory hallucinations or negative symptoms. In order to confirm symptom presence and persistence, and to estimate symptom severity, patients in these three groups were assessed on a range of instruments, including the Psychotic Symptom Rating Scale (PSYRAT)[17] for the Delusion group; selected items from the PSYRAT that were relevant to the group aims were used for the Auditory Hallucination group; and patients in the Negative Symptom group were assessed on the Scale for Assessment of Negative Symptoms (SANS).[18]

Within session CBT content

Sessions used standard group CBT structure,[19] with an informal atmosphere, age-relevant educational examples and humour to engage this young early course patient sample.[12] There was a strong Socratic and patient self-formulation emphasis. Within each session, specific techniques were drawn from the broad contemporary worldwide literature on CBT and on psychosis[20-22] and included: education, bibliotherapy, normalization, destigmatization; thought and visual image identification, monitoring and evaluation methods including appraisals of the symptom, associated meta-cognition, meta-emotional, and self-beliefs; reasoning biases (e.g. bias against disconfirmatory evidence and jumping to conclusions), and cognitive heuristics (availability and simulation); extensive (over 50) theoretically organized coping strategies were included in each group. Evidence-based biological and social factors that influence cognition in psychosis were also addressed in all groups.

Procedure

No referrals were sought. Instead, a frequently updated computerized team psychiatric grid lists all group-relevant symptoms of every patient in the EI service, and this was used to identify which CBT groups the patients were potentially eligible for. A psychology worker contacts the psychiatrist or the primary clinician and asks them to contact the patient, who is then contacted by the psychology worker. The presence of the symptom is confirmed before the group on the telephone or in a face-to-face meeting and re-confirmed on the day of the first group session. In the case of the three active psychotic symptom groups, patients also receive a pre-group therapeutic assessment (including PSYRAT or SANS items) in the weeks before the group is due to start, and weekly before each session for the Delusion and Negative symptom groups. Risk is checked by weekly liaison with the psychiatrists, primary clinicians and patients. Almost all groups in this study ran between August 2010 and September 2011.

Analysis

To test the relevancy of the MSS approach, we calculated the prevalence and the number of different types of symptoms patients had to see if there was homogeneity or heterogeneity in patients' symptomatic presentation. To test the feasibility of the MSS approach, we calculated the number of separate patients and number of attendances by patients at groups. To test the acceptability of the MSS approach, we calculated the number of patients who re-attended the programme and the percentage of sessions that were attended/missed. To test predictors of group programme attendance/re-attendance, we tested demographics (age, gender, ethnicity); general clinical characteristics (diagnosis, DUP, psychosis illness length); and symptom characteristics (type of symptom and number of symptoms, as listed in Table 4). For the predictors of attendance/re-attendance, we used Pearson chi-square/Fisher's exact test and t-tests/Mann–Whitney U.

Results

Sample demographic and clinical characteristics (n = 166, Table 1)

Table 1. Sample demographic, clinical and symptomatic characteristics (n = 166)Thumbnail image of

The EIPS sample comprised 166 patients. Four patients (2.4%) were excluded (three non-English-speaking, one learning disabled). Table 1 shows the overall sample demographic and clinical characteristics. Demographically, the sample was young (mean age 23.1), nearly two-thirds (63.9%, n = 106/166) were male, and just over one-third (37.3%, n = 62/166) was Caucasian. Clinically, the sample is recent onset (under 2 years psychosis length, n = 159, seven patients medication start was unavailable), short DUP (under 3 months, n = 148, 21 patients' onset date was within the previous 12 months, but could not be precisely dated) and nearly half (44.6%, n = 74/166) had a schizophrenia spectrum diagnosis. Almost all patients were on psychiatric medication.

Sample symptom characteristics (n = 166, Table 1)

In the sample of 166 patients eligible for the group programme, symptoms potentially relevant to the symptom-specific groups were prevalent: 137 (82.5%) of patients were eligible for at least one of the five symptom-specific groups. It can be seen that there is high variability in the type, number and activity/remission of symptoms. Not shown in Table 1 is that the precise types of symptom co-morbidity were extremely diverse, with 22 mutually exclusive combinations involving small percentages of the 166 patients showing different mutually exclusive combinations of symptoms (e.g. mood and past auditory hallucinations, 7.8%, n = 13). Importantly, patients' symptom co-morbidity meant that they were, on average, only eligible – in addition to Psycho-Education group – for a mean of 1.4 of the maximum four symptom-specific groups that were available. Some limited data is also available on the severity of the patients who attended the persistently active psychotic symptom groups. For the Delusions group (n = 5), the mean PSYRAT score was 17.8, with five items scoring a mean of 3/4 (conviction, preoccupation frequency and duration, distress frequency and intensity) and the life interference question averaging 2.5/4). For the Current Auditory Hallucinations group (n = 6), only items related to group aims were used as this lowered assessment burden on the patients: the mean PSYRAT auditory hallucination frequency = 2.1 (once a day to once an hour), the mean duration = 2.5 (several minutes to an hour), amount of voice content negative = 3 (majority of content unpleasant), degree of negative content = 3.1 (personal abuse relating to self-concept), amount of distress = 2.3 (minority of times distressing), intensity of distress = 2.8 (moderate/very distressing), life disruption = 3.1 (severe disruption). For the negative symptoms, group the mean SANS score was 53 (range 55, min = 21, max = 76).

Overall group attendance n = 58) Table 2)

Table 2. Group attendance: (n = 58/166 (34.9%))
Number of groups runAllPsycho-educationMood managementPast AHCurrent AHDelusionNegative
18942111
n/max (%)n/max (%)n/max (%)n/max (%)n/max (%)n/max (%)n/max (%)
  1. aTwo of the 11 groups were five sessions; the rest were three sessions.
  2. AH, auditory hallucination; Max, maximum potentially eligible based on symptom prevalence in the sample, or maximum mean number of sessions possible.
Number of separate patients attended58/166 (34.9)45/166 (27.1)19/70 (27.1)10/72 (13.9)6/24 (25)5/31 (16.1)9/40 (22.5)
Number of group sessions attended281/412 (68.2)93/145 (64.1)a61/95 (64.2)27/30 (90)40/48 (83.3)30/40 (75)30/54 (55.6)
Number of individual assessment sessions attended for group85/120 (71)00011/12 (93)35/45 (78)39/63 (62)
Total number of sessions attended366/532 (68.8)93/145 (64.1)61/95 (64.2)27/30 (90)51/70 (72.9)65/85 (76.5)69/117 (59)
Number of patients Re-attended session46/58 (79.3)29/45 (64.4)16/19 (84.2)10/10 (100)6/6 (100)5/5 (100)6/9 (66.7)
 M (SD)/maxM (SD)/maxM (SD)/maxM (SD)/maxM (SD)/maxM (SD)/maxM (SD)/max
Mean number of group sessions attended4.8 (4.3)/7.12.1 (1)/33.2 (1.6)/52.7 (0.5)/36.7 (1.5)/86 (2)/83.3 (2.1)/6

Table 2 shows that in this initial phase of the pilot study, all six groups have been run at least once to make a total of 18 separate group episodes, involving just over a third (34.9%, 58/166) of all eligible patients, across 281 group sessions, plus 85 individual therapeutic assessment sessions (delusion = 35, negative symptom = 39, auditory hallucination = 11), totalling 366 group-related CBT sessions. Patients attended a mean of 4.8 group sessions (standard deviation (SD) = 4.3, median = 3, mode = 3, range = 1–18). Including the 85 individual therapeutic assessment sessions, patients attended a mean of 6.3 group-related sessions overall. For the active psychotic symptoms groups, the mean number of group-related sessions attended – including the individual assessment sessions – was 9.3 group-related sessions (delusions = 13; auditory hallucinations = 8.5; negative symptoms = 7.6). The number of separate patients who attended a symptom-specific group out of the 137 potentially eligible was 36 (26.3%).

Group combination attendances

Of the 58 service users who attended the programme, the mean number of separate groups attended was 1.6 (SD = 0.8), with four being the most anyone has attended and one group being the least. With respect to attendance at mutually exclusive combinations of groups, patients were very diverse in their attendance, as patients attended in 18 different attendance permutations (e.g. Psycho-Education and Mood-Management group, 12.1%, 7/58).

Predictors of attendance n = 58) versus non-attendance n = 108) of the group programme Tables 3 and 4)

Table 3. Demographic and clinical predictors: attended (n = 58) V not attended (n = 108); re-attended (n = 46) V once (n = 12)
 Not attendedAttendedStatisticsAttended onceRe-attendedStatistics
% (n = 108)% (n = 58)% (n = 12)% (n = 46)
% (n)% (n) % (n)% (n) 
  1. Note: for gender, ethnicity and diagnosis each category is compared with all other participants who are not in that category.
  2. *P < 0.05; **P < 0.01.
  3. aNot attended n = 95; attended n = 53; attended once n = 11; re-attended n = 42.
  4. bNot attended n = 104; attended n = 55; attended once n = 12; re-attended n = 43.
  5. d.f., degrees of freedom; DUP, duration of untreated psychosis; FET, Fisher's exact test; M–W, Man–Whitney U-test; n/a, no analysis possible because of insufficient sample size; SD, standard deviation.
Gender  χ2 = 1.8, d.f. = 1, P = 0.18  FET P = 0.48
Male60.2 (65)70.7 (41) 83.3 (10)67.4 (31) 
Female39.8 (43)29.3 (17) 16.7 (2)32.6 (15) 
Ethnicity      
White36.1 (39)39.7 (23)χ2 = 0.20, d.f. = 1, P = 0.6541.7 (5)39.1 (18)χ2 = 0.03, d.f. = 1, P = 0.87
Asian32.4 (35)36.2 (21)χ2 = 0.24, d.f. = 1, P = 0.6225.0 (3)39.1 (18)FET P = 0.51
Black21.3 (23)20.7 (12)χ2 = 0.01, d.f. = 1, P = 0.9325.0 (3)19.6 (9)FET P = 0.7
Mixed6.5 (7)3.4 (2)FET P = 0.58.3 (1)2.2 (1)N/A
Any other background3.7 (4)0 (0)N/A0 (0)0 (0)N/A
Diagnosis      
Schizophrenia spectrum38.9 (42)55.2 (32)χ2 = 4.05, d.f. = 1, P = 0.04*58.3 (7)54.3 (25)χ2 = 0.06, d.f. = 1, P = 0.81
Affective psychosis12.0 (13)6.9 (4)FET P = 0.420 (0)8.7 (4)N/A
Acute and transient psychosis25.0 (27)19.0 (11)χ2 = 0.78, d.f. = 1, P = 0.3825.0 (3)17.4 (8)FET P = 0.68
Drug-induced psychosis4.6 (5)8.6 (5)χ2 = 1.06, d.f. = 1, P = 0.316.7 (2)6.5 (3)N/A
Unspec. non-organic psychosis7.4 (8)6.9 (4)FET P = 10 (0)8.7 (4)N/A
Delusional disorder1.9 (2)1.7 (1)N/A0 (0)2.2 (1)N/A
Other2.8 (3)1.7 (1)N/A0 (0)2.2 (1)N/A
Diagnostically uncertain7.4 (8)0 (0)FET P = 0.050 (0)0 (0)N/A
 M (SD)M (SD) M (SD)M (SD) 
Age22.9 (5.3)23.4 (4.2)t = −0.6, d.f. = 164, P = 0.5124.4 (5.0)23.1 (4.0)t = −0.94, d.f. = 56, P = 0.35
DUPa3.0 (6.0)2.6 (4.9)M–W = 2451.5, P = 0.782.9 (7.4)2.5 (4.1)M–W = 174.5, P = 0.2
Psychosis illness length totalb19.6 (11.6)26.3 (14.0)M–W = 1956, P = 0.001**22.3 (9.6)27.4 (4.0)M–W = 214.5, P = 0.37
Table 4. Symptom predictors: attended ((n = 58) V not attended ((n = 108); re-attended ((n = 46) V attended only once (n = 12)
 Not attendedAttendedStatisticsAttended onceRe-attendedStatistics
% (n = 108)% (n = 58)% (n = 12)% (n = 46)
% (n)% (n) % (n)% (n) 
  1. *P < 0.05.
  2. d.f., degrees of freedom; FET, Fisher's exact test; N/A, no analysis possible because of insufficient sample size; PE, psycho-eduction; SSG, symptom-specific group.
Prevalence of at least one group-relevant symptom79.6 (86)87.9 (51)χ2 = 1.804 d.f. = 1, P = 0.17975.0 (9)91.3 (42)χ2 = 2.384 d.f. = 1, P = 0.123
Past auditory hallucination42.6 (46)44.8 (26)χ2 = 0.077 d.f. = 1, P = 0.78241.7 (5)45.7 (21)χ2 = 0.061 d.f. = 1, P = 0.805
Active delusion18.5 (20)19.0 (11)χ2 = 0.005 d.f. = 1, P = 0.9448.3 (1)21.7 (10)FET P = 0.429
Active auditory hallucination13.9 (15)15.5 (9)χ2 = 0.081 d.f. = 1, P = 0.7768.3 (1)17.4 (8)FET P = 0.668
Active negative symptom21.3 (23)29.3 (17)χ2 = 1.325 d.f. = 1, P = 0.25041.7 (5)26.1 (12)χ2 = 1.115 d.f. = 1, P = 0.291
Prevalence of at least one active psychotic symptom35.2 (38)46.6 (27)χ2 = 2.046 d.f. = 1, P = 0.15358.3 (7)43.5 (20)χ2 = 0.844 d.f. = 1, P = 0.358
Active mood symptom37.0 (40)51.7 (30)χ2 = 3.338 d.f. = 1, P = 0.0741.7 (5)54.3 (25)χ2 = 0.613 d.f. = 1, P = 0.434
Active mood and psychotic symptom20.4 (22)32.8 (19)χ2 = 3.114 d.f. = 1, P = 0.0833.3 (4)32.6 (15)FET P = 1.000
Prevalence of at least one active symptom (mood or psychosis)52.8 (57)69.0 (40)χ2 = 4.071 d.f. = 1, P = 0.04*66.7 (8)69.6 (32)χ2 = 0.037 d.f. = 1, P = 0.847
No relevant group symptom (PE only)20.4 (22)12.1 (7)χ2 = 1.804 d.f. = 1, P = 0.17925.0 (3)8.7 (4)N/A
Eligible for 1 SSG41.7 (45)39.7 (23)χ2 = 0.063 d.f. = 1, P = 0.80233.3 (4)41.3 (19)FET P = 0.746
Eligible for 2 SSG25.9 (28)27.6 (16)χ2 = 0.053 d.f. = 1, P = 0.81716.7 (2)30.4 (14)FET P = 0.479
Eligible for 3 SSG8.3 (9)17.2 (10)χ2 = 2.954 d.f. = 1, P = 0.0925.0 (3)15.2 (7)FET P = 0.417
Eligible for 4 SSG3.7 (4)3.4 (2)FET P = 1.00004.3 (2)N/A

A diagnosis of schizophrenia spectrum disorder (vs. a combined group of other diagnoses) significantly (P = 0.04) predicted attendance (Table 3), whereas patients who had psychosis, but who were diagnostically uncertain, were almost significantly less likely to attend (P = 0.05) (Table 3). Having a longer illness length (by 6.7 months) significantly (P < 0.01) predicted an increased likelihood of attending the programme (Table 3). Patients were also significantly (P = 0.04) more likely to attend the programme if they had at least one active symptom (mood or psychotic), with trends for an active mood symptom (P = 0.07) or active mood plus active psychosis symptom (P = 0.08); and if there were three co-morbid symptoms relevant to the symptom-specific groups (P = 0.09) (Table 4).

Re-attendance Table 2)

Forty-six patients out of the 58 patients (79%) re-attended the programme, with re-attendance rates varying between 64% (Psycho-Education group) and 100% (Current and Past Auditory Hallucinations groups, and Delusions group). For the patients who did attend a group at least once, the overall percentage of group sessions that were attended (rather than missed) was 68.8%, with attendances for group sessions varying from 59% (Negative Symptoms) to 90% (Past Auditory Hallucinations). Nearly half (47%, 24/51) of those patients eligible to attend more than one group actually did so.

Predictors of re-attendance: attended once versus re-attended the programme, n = 46 versus n = 12 Tables 3 and 4)

No demographics, nor general clinical characteristics nor symptom characteristics were significantly associated with re-attendance.

Risk

Some patients who were accepted into the programme had serious symptom-related self-harm issues, but the psychiatrists, primary clinicians and patients identified no risk incidents that were related to any of the groups.

Discussion

We aimed to test the relevancy, feasibility, acceptability and safety of a pilot programme of MSS EI-GCBTp whose homogeneity and symptom-focused intensity might be additionally therapeutic, yet whose symptomatic streamlining, including novel groups for single delusions or negative symptoms at the EI stage, might be clinically and financially efficient.

Relevancy of MSS

The prevalence and heterogeneity of symptom type, activity, number and co-morbidity strongly supported the relevancy of a MSS approach.

Feasibility of MSS

In a routine service with limited resources, we managed – almost entirely over a 12.5-month period – to run 18 groups, all groups at least once, and recruit 58/166 (35%) of patients across 281 attended sessions, which supports the feasibility of a MSS approach. We have also successfully piloted symptom-specific GCBTp for single delusions and GCBTp for negative symptoms at the EI stage.

Acceptability of MSS

Nearly 80% of patients (46/58) who attended the programme re-attended, 68% of sessions were attended (rather than missed, similar to two generic programme studies),[3, 4] and nearly half (47%, 24/51) of those patients eligible to attend more than one group actually did so. Compared with previous EI-GCBTp programme studies,[2-4] our study included a wider range of symptom types, with similar or higher psychotic symptom severity than chronic symptom-specific psychosis studies[6, 8, 9, 23] and an EI-GCBTp programme.[3] Our study was also by far the most ethnically diverse. Our results therefore support the robust acceptability of the MSS approach across symptom presentations and cultures.

Efficiency of MSS

Our results suggest that a MSS approach is potentially a more clinically and financially efficient approach than running one long generic group that all patients attend because there is high heterogeneity in symptomatic presentations.

Patient engagement and retention in MSS

A new finding is that a longer psychosis length (but not the DUP part) predicts attendance, suggesting that persisting with patients in the service who do not initially attend might be fruitful over the succeeding (on average 6 more) months. Our results suggest that greater effort may need to be put into engaging patients who do not have a schizophrenia diagnosis (consistent with Malla et al.,[2] but not with Fanning et al.,[24] especially diagnostically uncertain psychotic patients. Asymptomatic patients may benefit from increased engagement effort (consistent with Kingdon & Kirchen),[25] but number of symptoms made no difference to attendance apart from a weak trend for three co-morbid group relevant symptoms to predict increased likelihood of attendance. With reference to retaining patients in EI-GCBTp, the high re-attendance rate (80%) may have underpowered our study to detect differences, although our finding that the Negative Symptom group had the least retention is consistent with Fanning et al.[24] who found that negative symptoms predicted dropout of EI-GCBTp.

Safety of MSS

The inclusion of some patients who had high symptom-risk issues and the novel application of GCBTp for single delusions, and GCBTp for negative symptoms, at the EI stage, did not lead to any group-related risk incidents; so the safety of a MSS approach was supported.

Dosage of MSS

On average, patients in our study attended almost 5 h, far fewer hours than successful generic group studies[2-4] or most of the single symptom-specific studies,[6, 8, 9] but similar to one chronic delusion study[7] and one EI auditory hallucination study.[12] A MSS approach is more concentrated on a single symptom type and so might require fewer sessions to work, especially at the EI stage.[1, 3]

Blend of MSS groups

Our programme blend of groups is but one selection that could be run in EI. We recommend that decisions about the blend of MSS EI-GCBTp groups that are run be dependent on resources and local survey data of symptom characteristics.

Limitations

A limitation to our study is that our survey of symptom prevalence relied on the psychiatrists, primary clinicians and patients for confirmation of symptom presence rather than validated psychometric instruments (except the active psychosis groups, which included PSYRAT or SANS items), because in a ‘Real World’, service there are insufficient resources to assess all patients in the service using validated instruments. We also did not record how many patients were still first episode, but our results are generalizable to samples with a short DUP and illness length. We did not record the reasons why only one-third of all patients have attended so far, but we aimed to engage all patients by the end of the 3 years of EI tenure. Our results concerning attendance predictors were based on an exploratory analysis so need to be replicated. It should be borne in mind that MSS EI-GCBTp is only one important component of a wider multimodal EI service, including non-CBT groups that focus on social performance rather than psychiatric symptoms.[26, 27]

Conclusion

Our pilot study illustrates a novel MSS EI-GCBTp programme approach that is highly relevant and accommodating of EI symptom heterogeneity and is feasible, culturally acceptable and safe; it also appears particularly efficient. A MSS approach may confer additional therapeutic benefits so future research should now test its clinical efficacy.

Acknowledgements

The part-time therapists who helped to deliver the MSS EI-GCBTp programme as part of their Doctorates in Clinical Psychology in London UK were: Alison Griffiths (Delusions therapy, University College London), Anna Georgiades (Negative Symptoms, therapy and audit, Royal Holloway), Daphne Paradisopoulos (Auditory Hallucinations, audit, Royal Holloway), and Jill Domoney (Delusions, audit, University College London). This study would not have been possible without the help of the Central and Northwest London NHS Foundation Trust Harrow and Hillingdon EIPS Team Manager, Leanne Frizzel.

Ancillary