Training to recognise the early signs of recurrence in schizophrenia

  • Review
  • Intervention

Authors


Abstract

Background

Schizophrenia has a lifetime prevalence of less than one per cent. Studies have indicated that early symptoms that are idiosyncratic to the person with schizophrenia (early warning signs) often precede acute psychotic relapse. Early warning signs interventions propose that learning to detect and manage early warning signs of impending relapse might prevent or delay acute psychotic relapse.

Objectives

To compare the effectiveness of early warning signs interventions plus treatment as usual involving and not involving a psychological therapy on time to relapse, hospitalisation, functioning, negative and positive symptomatology.

Search methods

Search databases included the Cochrane Schizophrenia Group Trials Register (July 2007 and May 2012) which is based on regular searches of BIOSIS, CENTRAL, CINAHL, EMBASE, MEDLINE and PsycINFO. References of all identified studies were reviewed for inclusion. We inspected the UK National Research Registe and contacted relevant pharmaceutical companies and authors of trials for additional information.

Selection criteria

We included all randomised clinical trials (RCTs) comparing early warning signs interventions plus treatment as usual to treatment as usual for people with schizophrenia or other non-affective psychosis

Data collection and analysis

We assessed included studies for quality and extracted data. If more than 50% of participants were lost to follow-up, the study was excluded. For binary outcomes, we calculated standard estimates of risk ratio (RR) and the corresponding 95% confidence intervals (CI), for continuous outcomes, we calculated mean differences (MD) with standard errors estimated, and for time to event outcomes we calculated Cox proportional hazards ratios (HRs) and associated 95 % CI. We assessed risk of bias for included studies and assessed overall study quality using the GRADE approach.

Main results

Thirty-two RCTs and two cluster-RCTs that randomised 3554 people satisfied criteria for inclusion. Only one study examined the effects of early warning signs interventions without additional psychological interventions, and many of the outcomes for this review were not reported or poorly-reported. Significantly fewer people relapsed with early warning signs interventions than with usual care (23% versus 43%; RR 0.53, 95% CI 0.36 to 0.79; 15 RCTs, 1502 participants; very low quality evidence). Time to relapse did not significantly differ between intervention groups (6 RCTs, 550 participants; very low quality evidence). Risk of re-hospitalisation was significantly lower with early warning signs interventions compared to usual care (19% versus 39%; RR 0.48, 95% CI 0.35 to 0.66; 15 RCTS, 1457 participants; very low quality evidence). Time to re-hospitalisation did not significantly differ between intervention groups (6 RCTs; 1149 participants; very low quality evidence). Participants' satisfaction with care and economic costs were inconclusive because of a lack of evidence.

Authors' conclusions

This review indicates that early warning signs interventions may have a positive effect on the proportions of people re-hospitalised and on rates of relapse, but not on time to recurrence. However, the overall quality of the evidence was very low, indicating that we do not know if early warning signs interventions will have similar effects outside trials and that it is very likely that further research will alter these estimates. Moreover, the early warning signs interventions were used along side other psychological interventions, and we do not know if they would be effective on their own. They may be cost-effective due to reduced hospitalisation and relapse rates, but before mental health services consider routinely providing psychological interventions involving the early recognition and prompt management of early warning signs to adults with schizophrenia, further research is required to provide evidence of high or moderate quality regarding the efficacy of early warning signs interventions added to usual care without additional psychological interventions, or to clarify the kinds of additional psychological interventions that might aid its efficacy. Future RCTs should be adequately-powered, and designed to minimise the risk of bias and be transparently reported. They should also systematically evaluate resource costs and resource use, alongside efficacy outcomes and other outcomes that are important to people with serious mental illness and their carers.

Résumé scientifique

Formation pour reconnaître les premiers signes de récidive de la schizophrénie

Contexte

La schizophrénie a une prévalence inférieure à un pour cent. Des études ont indiqué que les symptômes précoces qui sont caractéristiques de la personne atteinte de schizophrénie (signes avant-coureurs) précédaient souvent une rechute psychotique aiguë. Les interventions concernant les signes avant-coureurs suggèrent que le fait d'apprendre à détecter et à gérer les signes avant-coureurs d'une rechute imminente pourrait prévenir ou retarder la rechute psychotique aiguë.

Objectifs

Comparer l'efficacité d'interventions concernant les signes avant-coureurs associées au traitement habituel impliquant et n'impliquant pas une psychothérapie en termes de délai avant rechute, d'hospitalisation, d'état fonctionnel, de symptomatologie négative et positive.

Stratégie de recherche documentaire

Recherches effectuées dans les bases de données incluses dans le registre d'essais cliniques du groupe Cochrane sur la schizophrénie (juillet 2007 et mai 2012) qui se base sur des recherches régulières issues de BIOSIS, CENTRAL, CINAHL, EMBASE, MEDLINE et PsycINFO. Les bibliographies de toutes les études identifiées ont été examinées en vue de l'inclusion des études. Nous avons consulté le UK National Research Register et avons contacté les sociétés pharmaceutiques pertinentes et des auteurs d'essais afin d'obtenir des informations supplémentaires.

Critères de sélection

Nous avons inclus tous les essais cliniques randomisés (ECR) comparant des interventions concernant les signes avant-coureurs associées à un traitement habituel à un traitement habituel chez les personnes atteintes de schizophrénie ou d'une autre psychose non affective.

Recueil et analyse des données

Nous avons évalué la qualité des études incluses et extrait des données. Si plus de 50 % des participants étaient perdus de vue, l'étude était exclue. Pour les résultats binaires, nous avons calculé des estimations standard du risque relatif (RR) et les intervalles de confiance (IC) à 95 % correspondants, pour les résultats continus, nous avons calculé des différences moyennes (DM) avec des estimations des erreurs standard, et pour les résultats de délai jusqu'à l'événement, nous avons calculé des hazards ratios (HR) proportionnels de Cox et les IC à 95 % associés. Nous avons évalué le risque de biais pour les études incluses et avons évalué la qualité globale des études en utilisant l'approche GRADE.

Résultats principaux

Trente-deux ECR et deux ECR en grappe, dans lesquels étaient randomisées 3 554 personnes, ont répondu aux critères d'inclusion. Une seule étude a examiné les effets d'interventions concernant les signes avant-coureurs sans interventions psychologiques supplémentaires et un grand nombre des critères de jugement pour cette revue n'ont pas été rapportés ou ont été mal rapportés. Un nombre significativement inférieur de personnes ont fait une rechute avec les interventions relatives aux signes avant-coureurs par rapport aux soins habituels (23 % versus 43 % ; RR 0,53, IC à 95 % 0,36 à 0,79 ; 15 ECR, 1 502 participants ; preuves de qualité très médiocre). Le délai avant rechute n'a pas été significativement différent entre les groupes d'intervention (6 ECR, 550 participants ; preuves de qualité très médiocre). Le risque de réhospitalisation a été significativement inférieur avec des interventions concernant les signes avant-coureurs comparé aux soins habituels (19 % versus 39 % ; RR 0,48, IC à 95 % 0,35 à 0,66 ; 15 ECR, 1 457 participants ; preuves de qualité très médiocre). Le délai avant la réhospitalisation n'a pas été significativement différent entre les groupes d'intervention (6 ECR, 1 149 participants ; preuves de qualité très médiocre). La satisfaction des participants concernant les soins et les coûts économiques n'a pas été concluante en raison du manque de preuves.

Conclusions des auteurs

Cette revue indique que les interventions concernant les signes avant-coureurs peuvent avoir un effet positif sur la proportion de personnes réhospitalisées et sur les taux de rechute, mais non sur le délai avant récidive. Cependant, la qualité globale des preuves était très médiocre, ce qui indique que nous ignorons si les interventions concernant les signes avant-coureurs auront des effets semblables en dehors des essais et qu'il est très probable que des recherches supplémentaires modifient ces estimations. De plus, les interventions concernant les signes avant-coureurs ont été utilisées en association avec d'autres interventions psychologiques et nous ignorons si elles seraient efficaces seules. Elles peuvent avoir un meilleur rapport coût-efficacité en raison de taux d'hospitalisation et de rechute réduits, mais avant que les services de santé mentale n'envisagent de proposer systématiquement des interventions psychologiques impliquant la reconnaissance précoce et la prise en charge rapide des signes avant-coureurs chez l'adulte atteint de schizophrénie, des recherches supplémentaires doivent être menées pour fournir des preuves de grande qualité ou de qualité modérée concernant l'efficacité des interventions relatives aux signes avant-coureurs associées aux soins habituels sans interventions psychologiques supplémentaires ou pour déterminer les types d'interventions psychologiques supplémentaires qui pourraient améliorer leur efficacité. Les ECR à venir doivent avoir une puissance adéquate, être conçus pour minimiser le risque de biais et être rapportés de manière transparente. Ils doivent également évaluer de façon systématique les coûts en ressources et l'utilisation des ressources, ainsi que les critères d'efficacité et les autres critères de jugement qui sont importants pour les personnes atteintes d'une grave maladie mentale et leurs soignants.

Resumo

Treinamento para reconhecer os sinais precoces de recorrência em pacientes com esquizofrenia

Introdução

A esquizofrenia tem prevalência menor que 1%. Estudos indicam que, frequentemente, antes de os pacientes com esquizofrenia terem um novo surto agudo de psicose, eles apresentam sintomas específicos e individuais (sinais de alerta precoce). Os adeptos das intervenções para sinais de alerta precoce acreditam que, se o paciente aprender a detectar e controlar esses sinais, isso poderia prevenir ou retardar um novo surto psicótico agudo.

Objetivos

Comparar a efetividade de intervenções para sinais de alerta precoce associadas ao tratamento usual, envolvendo ou não terapia psicológica, sobre o intervalo até recaída, a necessidade de internação, a funcionalidade do paciente e sobre sintomas positivos e negativos.

Métodos de busca

Fizemos buscas na base de dados Cochrane Schizophrenia Group Trials Register (Julho de 2007 e Maio 2012) que recupera estudos publicados nas bases BIOSIS, CENTRAL, CINAHL, EMBASE, MEDLINE e PsycINFO. Revisamos as listas de referências de todos estudos incluídos para identificar possíveis novos estudos. Também fizemos buscas na UK National Research Register e entramos em contato com todas as empresas farmacêuticas relevantes e com os autores dos estudos para obter informações adicionais.

Critério de seleção

Incluímos os ensaios clínicos randomizados (ECRs) que compararam intervenções para sinais de alerta precoce mais tratamento usual versus tratamento usual, em pacientes com esquizofrenia ou outras psicoses não afetivas.

Coleta dos dados e análises

Extraímos dados dos estudos incluídos e avaliamos sua qualidade. Excluímos os estudos com perdas de mais de 50% dos participantes. Para os desfechos dicotômicos, calculamos o risco relativo (RR) e os respectivos intervalos de confiança (IC) de 95%. Para os desfechos contínuos, calculamos a diferença média (DM) e o erro padrão estimado. Para o desfecho tempo até o próximo evento, calculamos a taxa de incidência (hazard ratio-HR) proporcional de Cox e seu IC de 95%. Avaliamos o risco de viés dos estudos incluídos. Usamos a metodologia GRADE para avaliar a qualidade geral das evidências.

Principais resultados

Incluímos na revisão 32 ECR e 2 ECR por conglomerados (cluster), envolvendo um total de 3.554 pacientes. Apenas um estudo avaliou os efeitos das intervenções para sinais de alerta precoce sem intervenções psicológicas adicionais. Muitos dos desfechos da revisão não foram reportados ou foram mal reportados pelos ECR incluídos. A taxa de recaída foi significativamente menor nos pacientes do grupo das intervenções para sinais de alerta precoce do que nos pacientes do grupo de cuidados usuais (23% versus 43%; RR 0,53, IC 95% 0,36 a 0,79, 15 ECR, 1.502 participantes,qualidade da evidência muito baixa). Não houve diferença significativa entre os grupos para o desfecho tempo até a recaída (6 ECR, 550 participantes, evidência de qualidade muito baixa). O risco de re-internação foi significativamente menor no grupo de intervenção para sinais de alerta precoce em comparação com o grupo de cuidados usuais (19% versus 39%, RR 0,48, IC 95% 0,35 a 0,66, 15 ECR, 1.457 participantes, evidência de qualidade muito baixa). Não houve diferença estatisticamente significativa entre os grupos para o desfecho tempo até reinternação (6 ECR,1.149 participantes, evidência de qualidade muito baixa). Devido à falta de evidências, a satisfação dos pacientes com o cuidado recebido e os resultados relacionados a custo foram inconclusivos.

Conclusão dos autores

Esta revisão indica que as intervenções para sinais precoces podem reduzir a proporção de pacientes com esquizofrenia que precisam ser reinternados e as taxas de recaída, mas não afetam o tempo até a próxima recorrência. Porém, no geral, a qualidade das evidências foi muito baixa. Isso significa que não podemos ter certeza se esse tipo de intervenção traria resultados semelhantes quando usada fora dos ensaios clínicos e também significa que é muito provável que futuros estudos possam vir a modificar nossas estimativas do efeito dessa intervenção. Além disso, as intervenções para sinais de alerta precoce foram sempre usadas junto com outras intervenções psicológicas e, portanto, não sabemos se essas intervenções seriam efetivas se fossem usadas sozinhas. As intervenções analisadas podem ser custo-efetivas na medida em que elas reduzem as internações e as taxas de recaída. No entanto, antes de os serviços de saúde mental incorporarem essa conduta (de identificação e gestão imediata dos primeiros sinais de alerta para adultos com esquizofrenia) na sua rotina, outros estudos devem ser conduzidos para obtermos evidências de melhor qualidade e para sabermos qual tipo de intervenção psicológica deve ser associado para aumentar sua eficácia. Futuros ECR devem ser bem desenhados e conduzidos para minimizar o risco de viés e devem reportar seus achados de maneira transparente. Eles também devem avaliar sistematicamente os custos e a utilização de recursos, assim como os resultados de eficácia e os outros desfechos importantes para os pacientes com doença mental grave e seus cuidadores.

Notas de tradução

Tradução do Centro Cochrane do Brasil (Camila Rufino). Contato: tradutores@centrocochranedobrasil.org.br

Plain language summary

Training to recognise the early signs of recurrence in schizophrenia

Many people with schizophrenia experience periods of illness followed by relatively stable periods (although symptoms of illness such as hearing voices and seeing things often remain in the background). This means that many people with schizophrenia may become unwell again and need to go back into hospital. Training in early warning signs techniques encourages people to learn, detect and recognise the early warning signs of future illness. Studies indicate that noticing even small changes in signs and symptoms of schizophrenia can often predict future illness and relapse two to 10 weeks later. Early warning training may help to prevent or delay relapse, so reducing the chances of going into hospital. Recognition of early warning signs requires detailed history taking, sometimes with additional techniques such as diary keeping, completion of questionnaires and a plan of action based on anticipated early warning signs. Training can be undertaken by the individual or be group-based, involving health professionals, family members or carers. Successful training seems to require around 12 sessions and involves therapists of high competency.

This review includes a total of 34 studies. It found that there are positive benefits of training in early warning signs. It reduces rates of relapse and re-hospitalisation (but not on time to recurrence). It should be noted that training in early warning signs was mainly used alongside other psychological therapies, so it is not entirely clear what proportion of the positive effect is due to training in early warning signs alone. Moreover, the overall quality of the evidence from these studies was judged to be very low. This means that we do not know if interventions using early warning signs, with or without additional psychological treatments, will have the same beneficial effects outside clinical trials.

Further research is required to decide whether training in early warning signs is effective on its own. Effects on quality of life, satisfaction with care, money spent, and burden of care for carers are unclear, so ideally should be known before training programmes are put into wider use. At this time, there is not enough evidence to support training in early warning signs alone.

This plain language summary was written by consumer, Ben Gray of RETHINK.

Résumé simplifié

Formation pour reconnaître les premiers signes de récidive de la schizophrénie

Un grand nombre de personnes atteintes de schizophrénie connaissent des périodes de maladie suivies de périodes relativement stables (bien que les symptômes de la maladie, tels que le fait d'entendre des voix et de voir des choses, restent souvent présents en arrière-plan). Cela signifie qu'un grand nombre de personnes atteintes de schizophrénie peuvent ne pas aller bien de nouveau et nécessiter un retour à l'hôpital. Les techniques de formation aux signes avant-coureurs encouragent les personnes à apprendre, à détecter et à reconnaître les signes avant-coureurs de la maladie à venir. Des études indiquent que le fait de remarquer des changements, même faibles, concernant les signes et symptômes de la schizophrénie permet souvent de prédire la maladie à venir et la rechute deux à 10 semaines plus tard. La formation aux signes avant-coureurs peut aider à prévenir ou à retarder la rechute, réduisant ainsi les risques de retour à l'hôpital. La reconnaissance des signes avant-coureurs nécessite un interrogatoire approfondi, parfois à l'aide de techniques supplémentaires, telles que le fait de tenir un journal, de remplir des questionnaires et telles qu'un plan d'action fondé sur des signes avant-coureurs prévus. La formation peut être suivie par une personne ou être collective ; elle s'adresse notamment aux professionnels de santé, aux membres de la famille ou aux soignants. La réussite d'une formation semble nécessiter la participation à environ 12 séances et l'intervention de thérapeutes d'une grande compétence.

Cette revue inclut au total 34 études. Elle a découvert qu'une formation aux signes avant-coureurs présentait des bénéfices clairs. Elle réduit le taux de rechute et de réhospitalisation (mais pas le délai avant la rechute). Il convient de remarquer que la formation aux signes avant-coureurs a principalement été utilisée en association avec d'autres psychothérapies, on ignore donc exactement la part de l'effet positif due uniquement à la formation aux signes avant-coureurs. De plus, la qualité globale des preuves issues de ces études a été évaluée comme très médiocre. Cela signifie que nous ignorons si des interventions utilisant les signes avant-coureurs, avec ou sans psychothérapies supplémentaires, auront les mêmes effets bénéfiques en dehors des essais cliniques.

Des recherches supplémentaires sont nécessaires pour déterminer si la formation aux signes avant-coureurs seule est efficace. Les effets sur la qualité de vie, la satisfaction des soins, les dépenses et la charge des soins pour les soignants sont incertains et doivent idéalement être connus avant que le recours à des programmes de formation soit étendu. A ce stade, les preuves sont insuffisantes pour soutenir le recours à la formation aux signes avant-coureurs seule.

Ce résumé en langage simplifié a été rédigé par un consommateur, Ben Gray, de RETHINK.

Notes de traduction

Traduit par: French Cochrane Centre 1st March, 2013
Traduction financée par: Pour la France : Ministère de la Santé. Pour le Canada : Instituts de recherche en santé du Canada, ministère de la Santé du Québec, Fonds de recherche de Québec-Santé et Institut national d'excellence en santé et en services sociaux.

Resumo para leigos

Treinamento para reconhecer os primeiros sinais de um novo surto de esquizofrenia

Muitas pessoas com esquizofrenia apresentam períodos de doença ativa seguidos por períodos de estabilidade relativa (embora os sintomas da doença, tais como ouvir vozes e ver coisas, frequentemente continuem ocorrendo). Isso significa que muitas pessoas com esquizofrenia podem voltar a piorar e precisar de uma nova internação. O treinamento em técnicas para reconhecer os primeiros sinais de recaída encoraja as pessoas a aprender, detectar e reconhecer os sinais de alerta de que vai acontecer uma nova recaída da doença. Estudos indicam que perceber pequenas mudanças nos sinais e sintomas da esquizofrenia pode muitas vezes predizer que vai ocorrer um novo surto e uma recaída da doença dentro das próximas duas a dez semanas. O treinamento para reconhecer sinais de alerta precoce pode ajudar a prevenir ou retardar a recaída, reduzindo as chances de internação. Identificar esses sinais de alerta requer que o médico faça uma história detalhada do paciente, e às vezes é necessário incluir também outras técnicas, tais como o paciente manter um diário dos seus sintomas, preencher questionários e ter um plano de ação para saber o que fazer diante dos sinais de alerta precoce. O treinamento pode ser realizado individualmente ou em grupo, e envolve profissionais de saúde, familiares ou cuidadores. Para que as orientações sejam efetivas, são necessárias em torno de 12 sessões envolvendo terapeutas de alta competência.

Esta revisão incluiu 34 estudos. A conclusão foi que o treinamento para reconhecer os sinais de alerta precoce pode ser benéfico. Esse treinamento reduz a taxa de recaídas e de reinternações (mas não modifica o tempo até a próxima recaída). É importante salientar que, em todos os estudos, o treinamento para reconhecer os primeiros sinais de alerta foi sempre usado em conjunto com outras técnicas de terapia psicológica. Por isso não está claro qual seria o benefício exato do treinamento isoladamente. Além disso, a qualidade geral das evidências existentes é muito baixa. Isso significa que nós não sabemos se as intervenções que utilizaram os sinais de alerta precoce, com ou sem tratamentos psicológicos adicionais, terão os mesmos efeitos benéficos fora dos estudos, isto é, na prática clínica.

Pesquisas adicionais são necessárias para saber se o treinamento de reconhecimento dos sinais de alerta precoce seria efetivo por si só. Os efeitos desse treinamento sobre a qualidade de vida, a satisfação com o atendimento, os gastos e a carga para os cuidadores não são claros. Por isso, antes de introduzir esse treinamento como uma rotina nos serviços de saúde, o ideal seria avaliar tudo isso.No momento, não há evidências suficientes que apoiem o uso desse treinamento (do reconhecimento dos primeiros sinais de alerta) como medida isolada.

Este resumo para leigos foi escrito por Ben Gray do RETHINK.

Notas de tradução

Tradução do Centro Cochrane do Brasil (Camila Rufino). Contato: tradutores@centrocochranedobrasil.org.br

Summary of findings(Explanation)

Summary of findings for the main comparison. Early warning signs training and treatment as usual compared to treatment as usual for schizophrenia
  1. 1Eight studies had an unclear risk of bias for sequence generation and 13 for allocation concealment. One study did not use assessors that were blinded and in seven it was unclear. Eight studies had an unclear risk of bias for attrition and two studies had received funding from industry.

    2There was high heterogeneity in the pooled results.

    3The 95% confidence intervals are wide and include both significant benefit and harm of the intervention.

    4The data were not provided for most studies and had to be imputed. Three studies had an unclear risk of bias for random sequence generation and all studies had an unclear risk for allocation concealment. In one study the assessors were not blinded and in another the risk of bias for blinding was unclear. In three studies there was an unclear risk of bias for attrition.

    5The data were not provided for most studies and had to be imputed. Two studies had an unclear risk of bias for random sequence generation and two were unclear for allocation concealment. One study did not have assessors that were blinded and in another it was unclear. Three studies had an unclear risk of attrition bias. One study was funded by industry.

    6Nine studies had an unclear risk of bias for sequence generation and 12 for allocation concealment. One study did not use assessors that were blinded and in seven it was unclear. One study had a high risk of attrition bias and eight studies had an unclear risk of bias for attrition. One study had received funding from industry.

    7The funnel plot indicates that there may be the equivalent number of ’negative’ trials that have not been included in this analysis.

    8Two studies had an unclear risk of bias for sequence generation and three for allocation concealment. One study did not use blind assessors and one study had an unclear risk of bias for attrition. Two studies were part funded by industry.

    9Only three studies reported this outcome.

    10The two RCTs found no significant difference in satisfaction with care between early warning signs training and treatment as usual, although the data may be skewed for one trial. The cluster RCT did find a significant difference in favour of early warning signs training.

    11Ten studies had an unclear risk of bias for sequence generation and fifteen for allocation concealment. One study did not use assessors that were blinded and in seven it was unclear. One study had a high risk of attrition bias and eight studies had an unclear risk of bias for attrition. Two studies had received funding from industry.

    12One study had skewed data and the other did not report SDs.

    13The two studies found very different results.

    14This outcome included relatively few participants.

    15 Kuipers 1997 showed no significant difference between the mean cost of care for the early warning signs training group and standard care group, but the data was skewed. McDonell 2006 found that the mean costs of care were less in the early warning signs group ($1641) than the standard psychiatric group ($5199), no SDs were reported.

    16 Some indication of publication bias from funnel plot.

Early warning signs training and treatment as usual compared to treatment as usual for schizophrenia
Patient or population: patients with schizophrenia
Settings: secondary care
Intervention: Early warning signs training and treatment as usual
Comparison: treatment as usual
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of Participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Treatment as usualEarly warning signs training and treatment as usual
Relapse: Rate of relapse
Follow-up: 3 to 36 months
432 per 1000229 per 1000
(155 to 341)
RR 0.53
(0.36 to 0.79)
1502
(15 studies)
⊕⊝⊝⊝
very low 1, 2,3
 
Relapse: Time to relapse
Follow-up: 12 to 18 months
See commentSee commentHR 0.53
(0.27 to 1.06)
550
(6 studies)

⊕⊝⊝⊝

very low 2, 3, 4

The data were added to the meta-analysis as generic inverse variance.
Service use: Rate of re-hospitalisation
Follow-up: 2 to 36 months
385 per 1000185 per 1000
(135 to 254)
RR 0.48
(0.35 to 0.66)
1457
(15 studies)
⊕⊝⊝⊝
very low 2, 6, 7
 
Service use: Time to re-hospitalisation
Follow-up: 12 to 24 months
See commentSee commentHR 0.58
(0.33 to 1.04)
1149
(6 studies)
⊕⊝⊝⊝
very low 2, 3, 5
The data were added to the meta-analysis as generic inverse variance.
Satisfaction with care
Follow-up: 24 weeks to 12 months
See commentSee commentNot estimableSee comment
(3 studies)
⊕⊝⊝⊝
very low 8, 9
2 RCTs (92 participants) and 1 cluster RCT (56 clusters) report on satisfaction with care.10
Adverse events: Lost to follow-up
Follow-up: 10 weeks to 24 months
105 per 1000101 per 1000
(76 to 134)
RR 0.96
(0.72 to 1.28)
1495
(17 studies)
⊕⊝⊝⊝
very low 3, 7, 11 ,16
 
Economic burden (cost of care)See commentSee commentNot estimable129
(2 studies)
⊕⊝⊝⊝
very low 12, 13, 14
Two studies report on costs of care, but data could not be pooled.15
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio; HR: Hazard ratio;
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Background

The lifetime prevalence of schizophrenia is just less than one per cent with onset usually occurring during adolescence or early adulthood. Diagnosis is predictive of an increased risk of suicide, impaired occupational and social functioning, and a heightened risk of physical illness. Most people with schizophrenia experience a cyclical pattern of illness with periods of acute psychotic episodes followed by stable periods of full or partial remission, although these are often accompanied by the presence of residual symptoms (APA 1997).

The use of secondary prevention strategies, in which patients are encouraged to recognise signs of impending exacerbations of illness, are common in the treatment of several chronic physical illnesses such as diabetes and heart disease and may be equally beneficial to those suffering from severe mental illness (Bustillo 1995).

Both retrospective (Bechdolf 1998; Herz 1980) and prospective (Birchwood 1989; Marder 1994) observational studies indicate that relatively small changes in signs and symptoms of schizophrenia can reliably predict acute episodes of psychosis two to 10 weeks later. The symptoms (observed and reported by the patient) and signs (mostly behaviours observed and reported by carers or mental health professionals) are idiosyncratic to the patient, but consistently predict recurrence of psychosis in those who have a relapsing and remitting course of schizophrenia (Birchwood 1989; Bustillo 1995; Herz 2000; Marder 1994 ).

Early warning interventions are based on the hypothesis that learning to detect and manage early warning signs (EWS) of impending relapse might prevent the onset of acute episodes (Fitzgerald 2001; Lam 1997). The term prodrome has been used to delineate early warning signs, but this term will be avoided here because prodrome is now commonly used to delineate a period immediately before the first onset of psychotic symptoms in schizophrenia (Bustillo 1995).

Originally the recognition of early warning signs was combined with intermittent medication strategies. It was hoped that this would help prevent the long-term side-effects of medication such as tardive dyskinesia, by reducing overall exposure to neuroleptic medication (Herz 1980; Jolley 1990). However, early randomised controlled trials suggested that over a two-year period, those randomised to early warning signs interventions and intermittent antipsychotic medication strategy experienced more acute episodes of schizophrenia than those randomised to a continuous use of maintenance antipsychotic medication only (Gaebel 1993; Jolley 1990).

More recently, however, added clinical benefit has been demonstrated in patients on maintenance antipsychotics when early warning signs interventions are combined with short-term additional treatment such as diazepam (Carpenter 1999), additional oral antipsychotic (Marder 1994), cognitive therapy (Gumley 2003) and multifaceted psycho-educational and family interventions (Herz 2000). If such interventions were effective in reducing recurrences they would offer significant benefits to sufferers and their caregivers. Some practice guidelines for schizophrenia (APA 1997) recommend working with patients and their families to address EWS of recurrence.

Monitoring for early warning signs in schizophrenia has often been included as a component of psychological interventions, for example, in combination with other cognitive strategies or as one aspect of family intervention. There is a need to identify which components of therapy have a positive influence on outcomes (Haddock 1998). Two narrative reviews have specifically evaluated studies in schizophrenia seeking to reduce relapse through interventions based on early warning signs (Birchwood 2001; Herz 1995) and conclusions to date have been promising. These reviews did not systematically search for studies, however, and did not attempt meta-analysis so conclusions are open to bias. Furthermore, they did not consider the full range of possible outcomes in schizophrenia such as the individual's well being and functioning and resulting difficulties incurred by friends and relatives.

It is also possible that early warning signs interventions in schizophrenia might be associated with adverse outcomes, for example increased depression due to increased self-focus required by symptom monitoring. For people with a diagnosis of schizophrenia, those identified as less aware of their symptoms were less likely to report being depressed (Dixon 1998). Symptom awareness has also been linked to recurrent suicidal thoughts (Amador 1996). Another effect might be increased medication due to the possible effect of increased symptom reporting without reducing relapse (Gaebel 1993).

Finally, the effectiveness of early warning signs interventions in schizophrenia might vary according to features of the intervention such as the frequency of monitoring, use of standardised checklists of early warning signs or early warning signs interventions tailored to the individual from previous relapses in that individual, participation of caregivers and health professionals in the monitoring, group- versus individually-delivered early warning signs intervention, severity of illness and the presence or absence of an acute episode of schizophrenia at the time of the intervention, or the existence of co-morbid conditions. Demographic factors such as age and gender should also be taken into account. There is evidence that patients with first episode schizophrenia can benefit from EWS interventions coupled with intermittent medication, but multi-episode cases of schizophrenia must have maintenance medication (Gaebel 2002).

Description of the condition

Most people with schizophrenia experience a cyclical pattern of illness with periods of acute psychotic episodes followed by stable periods of full or partial remission, although these are often accompanied by the presence of residual symptoms (APA 1997).

Description of the intervention

Recognition of early warning signs requires detailed history taking, with or without additional techniques such as diary keeping, completion of questionnaires and card sorting techniques, and a plan of action based on the early warning signs. The interventions may be individually-based or group-based and can involve family members or carers. They can be carried out in any treatment setting.

How the intervention might work

Early warning interventions are based on the hypothesis that learning to detect and manage EWS of impending relapse might allow treatment or coping strategies to reduce symptoms or reduce the risk of harm thereby preventing or delaying relapse and/or hospitalisations (Fitzgerald 2001, Lam 1997).

Why it is important to do this review

Early warning signs interventions are carried out in routine clinical care for people with schizophrenia but their availability and delivery vary considerably from one clinical service to another even in the same locality.

Objectives

The primary objective was to compare the effectiveness of early warning signs intervention plus treatment as usual versus treatment as usual for those with schizophrenia. Treatment as usual could involve medication plus psychological therapy which did not incorporate early warning signs principles or only medication as part of its care plan.

Another objective was to compare the effectiveness of intermittent medication used on recognition of early warning signs in people not taking maintenance neuroleptic medication with treatment as usual involving maintenance neuroleptic medication.

Other areas of interest were early warning signs as the primary focus of the intervention or as part of another intervention; early warning signs intervention delivered to patient only or patient and carer or health professional; early warning signs intervention delivered to carer only versus carer and patient; early warning signs intervention delivered individually versus group intervention.

Methods

Criteria for considering studies for this review

Types of studies

All relevant randomised controlled trials were included. Where a trial was described as 'double-blind' but it implied that the study was randomised and the demographic details of each group were similar, it was included. We excluded quasi-randomised studies, such as those allocated by using alternate days of the week.

Types of participants

Adults over the age of 16 with a diagnosis of schizophrenia or a non-affective, non-organic serious or chronic mental illness based on standardised psychiatric criteria (including Research Diagnostic Criteria, DSM IV, ICD-10), at any stage of the disorder and in any treatment setting, or carers/health professionals of people with this diagnosis.

We excluded studies where

  1. the primary diagnosis was another disorder (for example, eating disorders);

  2. participants were selected on the basis of a non-psychiatric medical condition (for example, cancer, human immune deficiency syndrome);

  3. samples selected for co-morbidity, i.e. addictions and mental illness;

  4. 'at risk' populations who had not yet received a definite diagnosis of schizophrenia;

  5. early onset samples, with no established diagnosis of schizophrenia.

Types of interventions

1. Early warning signs intervention plus treatment as usual

Interventions designed to systematically train people with a diagnosis of schizophrenia to recognise early warning signs of recurrence of episodes. The essential elements of early warning signs monitoring were defined as: i) education to increase awareness of early warning signs and identification of these early warning signs; ii) person with schizophrenia learning to self-monitor for early warning signs (e.g. paranoia), or carers and/or health professionals learning to monitor on behalf of the person with schizophrenia; iii) early action to prevent development of recurrence (e.g. seeking early help or self-coping methods).

Each person in the study must have received at least one hour of training specifically in recognising and managing early warning signs. Therefore, early warning signs interventions require at least one hour of face-to-face clinical contact. Any intervention that included early warning signs interventions either as a major focus or as one component of an intervention package were included. The interventions could have been individually-based or group-based, encompassed self-help approaches, involved family members or carers, and carried out in community, primary care or secondary care settings.

2. Treatment as usual not involving a psychological therapy
3.Treatment as usual plus another psychological therapy

Studies where both groups simultaneously received the same psychological intervention involving early warning signs interventions were excluded (for example, psychological therapy versus psychological therapy in addition to medication).

Types of outcome measures

Primary outcomes

1. Relapse
1.1 Rate of relapse
1.2 Time to relapse

2. Service utilisation
2.1 Rate of re-hospitalisation
2.2 Time to re-hospitalisation

Secondary outcomes

1. Mental State
1.1 No clinically important change in general mental state (as defined by individual studies)
1.2 Mean endpoint general mental state score
1.3 Mean change in general mental state scores
1.4 No clinically important change in specific symptoms
1.5 Mean endpoint specific symptom score
1.6 Mean change in specific symptom scores

2. Social functioning
2.1 Average endpoint general social functioning score
2.2 Average change in general social functioning scores
2.3 No clinically important change in specific social functioning
2.4 Average endpoint specific social functioning score
2.5 Average change in specific social functioning scores

3. Quality of life
3.1 No clinically important change in quality of life
3.2 Average endpoint quality of life score
3.3 Average change in quality of life scores
3.4 No clinically important change in specific aspects of quality of life
3.5 Average endpoint specific aspects of quality of life
3.6 Average change in specific aspects of quality of life

4. Satisfaction with care
4.1 Recipient of care
4.2 Informal care givers
4.3 Professional carers

6. Adverse events
6.1 Incidence of adverse effects, general or specific
6.2 Leaving the study early
6.3 Measured acceptance of treatment
6.4 Use of antiparkinsonian treatment
6.5 Sudden and unexpected death

7. Economic burden (cost of care)

8. Knowledge

9. Burden of care

10. Medication compliance

Search methods for identification of studies

Electronic searches

1. Cochrane Schizophrenia Group Trials Register (July 2007)

The register was searched using the phrase:

[((sign* OR early sign* OR signs of earl* OR intervention* OR detect*) OR  ((relaps* OR recurren*) AND (prevent* OR predict* OR reduc* OR control*))) AND (((monitor* OR help* OR evaluat* OR) and (self* OR family*)) OR symptom manag* OR manage symptom* OR aware* OR prodrom* OR re-occur* OR reoccur*) in REFERENCE title, abstract and index fields] OR [(training* OR relapse* OR consumer* OR self* OR early*) in STUDY intervention field]

This register is compiled by systematic searches of major databases, handsearches and conference proceedings (see group module).

2. Cochrane Schizophrenia Group Trials Register (May 2012)

We ran another search in 2012. The Trials Search Co-ordinator searched the Cochrane Schizophrenia Group’s Trials Register (3 May 2012).

[((sign* OR early sign* OR signs of earl* OR intervention* OR detect*) OR  ((relaps* OR recurren*) AND (prevent* OR predict* OR reduc* OR control*))) AND (((monitor* OR help* OR evaluat* OR) and (self* OR family*)) OR symptom manag* OR manage symptom* OR aware* OR prodrom* OR re-occur* OR reoccur*) in REFERENCE title, abstract and index fields] OR [(training* OR relapse* OR consumer* OR self* OR early*) in STUDY intervention field]

The Cochrane Schizophrenia Group’s Trials Register is compiled by systematic searches of major databases, handsearches of journals and conference proceedings (see Group Module). Incoming trials are assigned to relevant existing or new review titles.

Searching other resources

1. Reference lists

We searched all references of articles selected for inclusion for further relevant trials.

2. Personal contact

We contacted the first author of each included study for information regarding unpublished trials.

3. NHS National Research Register (NRR):

Using abstracts identified from searching the Cochrane Schizophrenia Group Register, the NRR was inspected for information regarding ongoing research trials, and we contacted the corresponding authors regarding unpublished data.

Data collection and analysis

Selection of studies

Review author IV inspected all abstracts of studies identified as above and identified potentially relevant reports. In addition, to ensure reliability, review authors MAF and RM inspected a random sample of these abstracts, comprising 30% of the total. Where disagreement occurred, it was resolved by discussion, or where there was still doubt, the full article was acquired for further inspection. The full articles of relevant reports were acquired for reassessment and carefully inspected by IV and MAF for a final decision on inclusion (see Criteria for considering studies for this review). IV and MAF were not blinded to the names of the authors, institutions or journal of publication. Where difficulties or disputes arose, we asked review author RM for help and where it was impossible to decide, these studies were added to those awaiting assessment and the authors of the papers contacted for clarification.

For the May 2012 search two members of the Enhance Review inspected full reports of trials selected by the original team of authors.

Data extraction and management

1. Extraction

Review authors IV and MF extracted data from all included studies found in the July 2007 search. To ensure reliability, RM independently extracted data from a random sample of these studies, comprising 30% of the total. Again, any disagreement was discussed, decisions documented and, if necessary, authors of studies contacted for clarification. With remaining problems RM helped clarify issues and those final decisions documented. Data presented only in graphs and figures were extracted whenever possible, but only included if two review authors independently had the same result. Where possible, we extracted data relevant to each component centre of multi-centre studies separately.
For the May 2012 search two members of the Enhance Reviews team extracted data, using above methods, for the included studies.

2. Management
2.1 Forms

A form for data collection was created, piloted in three trials independently by two review authors, and revised after author discussion. We extracted data onto these forms.

2.2 Scale-derived data

We included continuous data from rating scales only if: a) the psychometric properties of the measuring instrument have been described in a peer-reviewed journal (Marshall 2000); and b) the measuring instrument was not written or modified by one of the trialists for that particular trial.

Ideally, the measuring instrument should either be i. a self-report or ii. completed by an independent rater or relative (not the therapist). We realise that this is not often reported clearly but in the Description of studies section, we noted if this is the case or not.

2.3 Endpoint versus change data

There are advantages of both endpoint and change data. Change data can remove a component of between-person variability from the analysis. On the other hand, calculation of change needs two assessments (baseline and endpoint) which can be difficult in unstable and difficult to measure conditions such as schizophrenia. We decided primarily to use endpoint data and only use change data if the former were not available. Endpoint and change data were combined in the analysis as we used mean differences rather than standardised mean differences throughout (Higgins 2011, chapter 9.4.5.2 ). All data in the analyses are endpoint data unless marked in a footnote as change data.

2.4 Skewed data

Continuous data on clinical and social outcomes are often not normally distributed. To avoid the pitfall of applying parametric tests to non-parametric data, we aimed to apply the following standards to all data before inclusion:

a) standard deviations (SDs) and means are reported in the paper or obtainable from the authors;
b) when a scale starts from the finite number zero, the SD, when multiplied by two, is less than the mean (as otherwise the mean is unlikely to be an appropriate measure of the centre of the distribution, (Altman 1996);
c) if a scale starts from a positive value (such as the Positive and Negative Syndrome Scale (PANSS) (Kay 1986) which can have values from 30 to 210), the calculation described above was modified to take the scale starting point into account. In these cases skew is present if 2 SD > (S-S min), where S is the mean score and S min is the minimum score. Endpoint scores on scales often have a finite start and end point and these rules can be applied. When continuous data are presented on a scale that includes a possibility of negative values (such as change data), it is difficult to tell whether data are skewed or not. Skewed data from studies of less than 200 participants were entered in additional tables (Table 1) rather than into an analysis. Skewed data pose less of a problem when looking at means if the sample size is large and was entered into the syntheses.

Table 1. Skewed and incomplete data
  1. FAS: Family Attitude Scale
    FIBS: Family Burden Interview Schedule
    HoNOS: Health of the Nation Outcome Scale
    PAS: Psychiatric Assessment Scale
    REHAB: Rehabilitation Evaluation Hall and Baker
    SD: standard deviation
    SDSS: Social disability screening scale

OutcomeStudyResults
Mental state: average score in positive symptoms (PAS positive, high = bad)Drury 1996The early warning signs group (N = 20) had a mean of 2.06 (SD 2.57) and the control group (N = 20) had a mean of 4.59 (SD 2.72).
Mental state: average score in positive symptoms (PAS negative, high = bad)Drury 1996The early warning signs group (N = 20) had a mean of 1.33 (SD 1.8) and the control group (N = 20) had a mean of 1.71 (SD 2.2).
Social functioning: average score (REHAB scale, high = bad)Anzai 2002The early warning signs group (N = 14) had a mean of 18.2 (SD 13) and the control group (N = 15) had a mean of 32.5 (SD 22.7).
Social functioning: average score (HoNOS scale, high = bad)Bradley 2006The early warning signs group (N = 25) had a mean of 9.26 (SD 4.63) and the control group (N = 24) had a mean of 7.66 (SD 4.85).
Social functioning: average score (SDSS, high = bad)Chen 2003The early warning signs group (N = 31) had a mean of 4.04 (SD 3.89) and the control group (N = 31) had a mean of 7.63 (SD 4.27).
Social functioning: average score (SDSS, high = bad)Zhang 2004The early warning signs group (N = 54) had a mean of 4.8 (SD 3) and the control group (N = 50) had a mean of 9.8 (SD 3.9).
Social functioning: average months of employmentXiong 1994The early warning signs group (N = 30) had a mean of 7.83 (SD 3) and the control group (N = 28) had a mean of 4.64 (SD 7.39).
Burden of care: average score (FBIS, high = bad)Bradley 2006The early warning signs group (N= 24) had a mean of 18.95 (SD 15.39) and the control group (N = 22) had a mean of 9.38 (SD 8.1).
Burden of care: average score (FAS, high = good)Li 2003The early warning signs group (N = 46) had a mean of 21 (SD 14.54) and the control group (N = 55) had a mean of 25.15 (SD 20).
Mean number of missed dosesXiang 2001The early warning signs group had a mean of 0.2 missed doses (SD 0.6) and the control group had a mean of 4.7 missed doses (SD 2.2).
Economic burden (cost of care)Kuipers 1997The early warning signs training group had a mean care package cost of £1220 (SD 736) and the standard treatment group had a mean care package cost of £1403 (SD 887) after 18 months follow-up. There was no significant difference between the costs.
Economic burden (cost of care)McDonell 2006The mean cost of inpatient service utilisation was $5199 for the standard psychiatric care group and $1641 for the early warning signs training group at 3 years post-randomisation.
2.5 Common measure

To facilitate comparison between trials, we converted variables that can be reported in different metrics, such as days in hospital (mean days per year, per week or per month) to a common metric (e.g. mean days per month). 

2.6 Conversion of continuous to binary

Where possible, efforts were made to convert outcome measures to dichotomous data. This can be done by identifying cut-off points on rating scales and dividing participants accordingly into 'clinically improved' or 'not clinically improved'. It is generally assumed that if there is a 50% reduction in a scale-derived score such as the Brief Psychiatric Rating Scale (BPRS, Overall 1962) or the PANSS (Kay 1986), this can be considered as a clinically significant response (Leucht 2005a; Leucht 2005). If data based on these thresholds were not available, we used the primary cut-off presented by the original authors.

2.7 Direction of graphs

Where possible, we entered data in such a way that the area to the left of the line of no effect indicates a favourable outcome for treatment with antipsychotic combinations.

2.8 'Summary of findings' table

We used the GRADE approach to interpret findings (Schünemann 2008) and used GRADE profiler (GRADE 2004) to import data from RevMan 5 (Review Manager 2011) to create 'Summary of findings' tables. These tables provide outcome-specific information concerning the overall quality of evidence from each included study in the comparison, the magnitude of effect of the interventions examined, and the sum of available data on all outcomes we rate as important to patient-care and decision making. If possible, we anticipated selecting the following main outcomes for inclusion in the 'Summary of findings' table.

  1. Relapse: time to relapse

  2. Relapse: rate of relapse

  3. Service use: time to re-hospitalisation

  4. Service use: rate of re-hospitalisation

  5. Satisfaction with care

  6. Adverse events: lost to follow-up

  7. Economic burden (cost of care)

Assessment of risk of bias in included studies

Review authors IV and MF independently assessed the risk of bias of each trial using The Cochrane Collaboration's 'Risk of bias' tool (Higgins 2011). This set of criteria is based on evidence of associations between overestimate of effect and high risk of bias of the article such as sequence generation, allocation concealment, blinding, incomplete outcome data and selective reporting. Due to the nature of the intervention, it was not possible to blind participants or personnel and we marked this as low risk of bias for all studies. If the raters disagreed, the final rating was made by consensus, with the involvement of another member of the review group. Where inadequate details of randomisation and other characteristics of trials were provided, we contacted authors of the studies in order to obtain further information. Non-concurrence in quality assessment was reported, but if disputes arose as to which category a trial is to be allocated, again, resolution was made by discussion with RM. The level of risk of bias was noted in both the text of the review and in the Summary of findings for the main comparison. Two members of the Enhance Review team performed the risk of bias for the included studies from the May 2012 search.

Measures of treatment effect

1. Binary data

For binary outcomes we calculated a standard estimation of the risk ratio (RR) and its 95% confidence interval (CI). It has been shown that RR is more intuitive (Boissel 1999) than odds ratios and that odds ratios tend to be interpreted as RR by clinicians (Deeks 2000). For binary data presented in the 'Summary of findings' table, where possible, we calculated illustrative comparative risks as the Number Needed to Treat/Harm (NNT/H) statistic with its confidence intervals is intuitively attractive to clinicians but is problematic both in its accurate calculation in meta-analyses and interpretation (Hutton 2009).

2. Continuous data

For continuous outcomes we estimated mean difference (MD) between groups. We preferred not to calculate effect size measures (standardised mean difference (SMD)). However, if scales of very considerable similarity were used, we presumed there was a small difference in measurement, and we calculated effect size and transformed the effect back to the units of one or more of the specific instruments.

3. Time to event data

The outcomes “time to relapse” and "time to re-hospitalisation" were analysed using the Cox proportional hazard ratio (HR) and associated 95 % CI which takes into account the number and timing of events, and the time until last follow-up for each patient who has not experienced the event (Borenstein 2009; Tierney 2007).

Unit of analysis issues

1. Cluster trials

Studies increasingly employ 'cluster randomisation' (such as randomisation by clinician or practice) but analysis and pooling of clustered data poses problems. Firstly, authors often fail to account for intra-class correlation in clustered studies, leading to a 'unit of analysis' error (Divine 1992) whereby P values are spuriously low, confidence intervals unduly narrow and statistical significance overestimated. This causes type I errors (Bland 1997; Gulliford 1999). If clustering had not been accounted for in primary studies, we planned to present data in a table, with a (*) symbol to indicate the presence of a probable unit of analysis error. In subsequent versions of this review we will seek to contact first authors of studies to obtain intra-class correlation coefficients (ICCs) for their clustered data and to adjust for this by using accepted methods (Gulliford 1999). Where clustering has been incorporated into the analysis of primary studies, we presented these data as if from a non-cluster randomised study, but adjusted for the clustering effect. We have sought statistical advice and have been advised that the binary data as presented in a report should be divided by a 'design effect'. This is calculated using the mean number of participants per cluster (m) and ICC [Design effect = 1+(m-1)*ICC] (Donner 2002). If the ICC was not reported it was assumed to be 0.1 (Ukoumunne 1999). If cluster studies had appropriately analysed taking into account ICCs and relevant data documented in the report, synthesis with other studies was possible using the generic inverse variance technique.

2. Cross-over trials

A major concern of cross-over trials is the carry-over effect. It occurs if an effect (e.g. pharmacological, physiological or psychological) of the treatment in the first phase is carried over to the second phase. As a consequence, on entry to the second phase the participants can differ systematically from their initial state despite a wash-out phase. For the same reason cross-over trials are not appropriate if the condition of interest is unstable (Elbourne 2002). As both effects are very likely in severe mental illness, we only used data of the first phase of cross-over studies.

3. Studies with multiple treatment groups

Where a study involved more than two treatment arms, if relevant, the additional treatment arms were presented in comparisons. If data were binary we planned to simply add and combine within the two-by-two table. If data were continuous we planned to combine data following the formula in section 7.7.3.8 (Combining groups) of the Cochrane Handbook for Systematic Reviews of Interventions. For studies where each pair-wise comparison was relevant, we planned to include separately, with the shared intervention groups divided out approximately evenly among the comparisons: only the total number of participants were divided up and the means and standard deviations left unchanged, according to section 16.5.4 (How to include multiple groups from one study) of the Cochrane Handbook for Systematic Reviews of Interventions. This differs from the original protocol (see Differences between protocol and review). Where the additional treatment arms were not relevant, these data were not reproduced.

Dealing with missing data

1. Overall loss of credibility

If loss of data at follow-up becomes too great, then the study itself runs the risk of losing credibility (Xia 2009). We chose that, for any particular outcome, had more than 50% of data be unaccounted for, we did not reproduce these data or use them within analyses. Where, however, more than 50% of those in one arm of a study were lost, but the total loss was less than 50%, we marked such data with (*) to indicate that such a result may well be prone to bias.

2. Binary

In the case where attrition for a binary outcome was between 0% and 50% and where these data were not clearly described, data were presented on a 'once-randomised-always-analyse' basis (an intention-to-treat analysis). Those leaving the study early were all assumed to have the same rates of negative outcome as those who completed, with the exception of the outcome of death and adverse effects. For these outcomes the rate of those who stayed in the study - in that particular arm of the trial - were used for those who did not. A sensitivity analysis was undertaken to test how prone the primary outcomes were to change when 'completer' data only were compared to the intention-to-treat analysis using the above assumptions.

3. Continuous
3.1 Attrition

In the case where attrition for a continuous outcome was between 0% and 50% and completer-only data were reported, we reproduced these.

3.2 Standard deviations

If standard deviations (SDs) were not reported, we first tried to obtain the missing values from the authors. If not available, where there were missing measures of variance for continuous data, but an exact standard error (SE) and confidence intervals (CIs) were available for group means, and either 'P' value or 't' values were available for differences in mean, we calculated them according to the rules described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011): When only the SE is reported, SDs are calculated by the formula SD = SE * square root (n). Chapters 7.7.3 and 16.1.3 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) present detailed formula for estimating SDs from P values, t or F values, CIs, ranges or other statistics. Where these formulae do not apply, we calculated the SDs according to a validated imputation method which is based on the SDs of the other included studies (Furukawa 2006). Although some of these imputation strategies can introduce error, the alternative would have been to exclude a given study’s outcome and thus to lose information. We nevertheless examined the validity of the imputations in a sensitivity analysis excluding imputed values.

3.3 Last observation carried forward

We anticipated that in some studies the method of last observation carried forward (LOCF) will be employed within the study report. As with all methods of imputation to deal with missing data, LOCF introduces uncertainty about the reliability of the results (Leucht 2007). Therefore, where LOCF data have been used in the trial, if less than 50% of the data have been assumed, we reproduced these data and indicated that they are the product of LOCF assumptions.

Assessment of heterogeneity

1. Clinical heterogeneity

We considered all included studies initially, without seeing comparison data, to judge clinical heterogeneity. We simply inspected all studies for clearly outlying people or situations which we had not predicted would arise. When such situations or participant groups arose, these were fully discussed.

2. Methodological heterogeneity

We considered all included studies initially, without seeing comparison data, to judge methodological heterogeneity. We simply inspected all studies for clearly outlying methods which we had not predicted would arise. When such methodological outliers arose these were fully discussed.

3. Statistical heterogeneity
3.1 Visual inspection

We visually inspected graphs to investigate the possibility of statistical heterogeneity.

3.2 Employing the I2 statistic

Heterogeneity between studies was investigated by considering the I2 method alongside the Chi2 'P' value. The I2 provides an estimate of the percentage of inconsistency thought to be due to chance (Higgins 2011). The importance of the observed value of I2 depends on i. the magnitude and direction of effects and ii. the strength of evidence for heterogeneity (e.g. 'P' value from Chi2 test, or a confidence interval for I2). An I2 estimate greater than or equal to around 50% accompanied by a statistically significant Chi2 statistic was interpreted as evidence of substantial levels of heterogeneity (Section 9.5.2 - Higgins 2011). When substantial levels of heterogeneity were found in the primary outcome, we explored reasons for heterogeneity (Subgroup analysis and investigation of heterogeneity).

Assessment of reporting biases

Reporting biases arise when the dissemination of research findings is influenced by the nature and direction of results (Egger 1997). These are described in Section 10 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We are aware that funnel plots may be useful in investigating reporting biases but are of limited power to detect small-study effects. We did not use funnel plots for outcomes where there were 10 or fewer studies, or where all studies were of similar sizes. In other cases, where funnel plots were possible, we sought statistical advice in their interpretation.

Data synthesis

We understand that there is no closed argument for preference for use of fixed-effect or random-effects models. The random-effects method incorporates an assumption that the different studies are estimating different, yet related, intervention effects. This often seems to be true to us and the random-effects model takes into account differences between studies even if there is no statistically significant heterogeneity. There is, however, a disadvantage to the random-effects model. It puts added weight onto small studies which often are the most biased ones. Depending on the direction of effect, these studies can either inflate or deflate the effect size. A fixed-effect model was used, unless we demonstrated statistically significant heterogeneity (P < 0.10) for a specific outcome, in which case the random-effects models was preferred. The reader is, however, able to choose to inspect the data using the random-effects model.

Subgroup analysis and investigation of heterogeneity

1. Subgroup analyses

We conducted subgroup analyses for the primary outcomes comparing:

  • early warning signs training as the main focus of the intervention versus early warning signs training not the main focus of the intervention

  • patients only versus patients and carers/health workers

  • carers/health workers only versus patients and carers/health workers

  • group training versus individual training.

2. Investigation of heterogeneity

If inconsistency was high (I2 > 75%), this was reported. First, we investigated whether data had been entered correctly. Second, if data were correct, the graph was visually inspected and outlying studies were successively removed to see if heterogeneity was restored.

When unanticipated clinical or methodological heterogeneity were obvious, we simply stated hypotheses regarding these for future reviews or versions of this review. We did not anticipate undertaking analyses relating to these.

Sensitivity analysis

1. Implication of randomisation

We aimed to include trials in a sensitivity analysis if they were described in a way that implied randomisation. For the primary outcomes, we included these studies and if there was no substantive difference when the implied randomised studies were added to those with better description of randomisation, then all data were employed from these studies.

2. Assumptions for lost data

Where assumptions had to be made regarding missing SDs data (see Dealing with missing data), we compared the findings on primary outcomes when we used our assumption compared with complete data only. We undertook a sensitivity analysis to test how prone results were to change when 'completer' data only were compared to the imputed data using the above assumption. If there was a substantial difference, we reported results and discussed them but continued to employ our assumption.

3. Imputed data

Where time to event data had to be imputed, we compared the imputed data with data that did not have to be imputed and reported any differences in results.

Results

Description of studies

Please see Characteristics of included studies, Characteristics of excluded studies, and Characteristics of ongoing studies.

In Additional Tables (Table 2 and Table 3) we categorised all studies by:

Table 2. Comparison of nature of intervention of studies that reported primary outcome as relapse/ re-hospitalisation
  1. CBT: cognitive behavioural therapy
    EWS: early warning signs

Study IDEWS as primary focusEWS delivered to patient onlyEWS delivered to patient and carer/ health workerEWS delivered to carer/ health worker onlyEWS delivered individuallyEWS delivered in groupsEWS promoting only help seeking from othersEWS promoting additional self-directed coping
Anzai 2002
Bauml 2007
Bradley 2006  
Buchkremer 1997  
Chan 2007not reported not reported 
Chen 2003not reportednot reported
Chien 2004
Drury 1996 

 

✗ 

not reported 
Garety 2008✓ (CBT)✓ (family intervention)✓ (CBT)✓ (family intervention)

no (CBT)

yes (family intervention)

✓ (CBT)

no (family intervention) 

Granholm 2005
Gumley 2003  
Harris 2009
Herz 2000
Hogarty 1997a
Horan 2009
Kopelowicz 1998b
Kopelowicz 2003  
Kuipers 1997
Li 2003optional
Liberman 1998
McDonell 2006  
Merinder 1999  
Norman 2002  
Shin 2002✗ optional 
Shon 2002unclear
Tait 2002  ✓ includes cognitive therapy
Turkington 2006  
Van Meijel 2006  
Vreeland 2006
Weng 2005  
Xiang 2001
Xiong 1994
Zhan 2003not reportednot reported
Zhang 2004
Table 3. Comparison of description of intervention of studies that reported primary outcome as relapse/ re-hospitalisation
Study IDFocus of interventionAt least one hour/ session focus on EWSEWS intervention delivered byNo. of sessionsControl group
Anzai 2002Community re-entry programyesWard Nurses18 one-hour sessionsOccupational rehabilitation
Bauml 2007Psychoeducation of patients and relatives (reduce re-hospitalisation and improve compliance)yesPsychiatrist or clinical psychologist

8 one-hour groups, 4 sessions weekly and 4 monthly

8 bi-weekly 90-minute sessions for relatives

Outpatient treatment, antipsychotic medication
Bradley 2006  Multiple family group treatmentyes

Psychoeducation:Psychiatrist, psychologists, social workers, occupational therapists

Family groups: Therapists

3 single family joining sessions.

2 half-day psychoeducation sessions.

Multiple family group fortnightly for 12 months

Appointments with case manager and doctor for medication and individual psychosocial rehabilitation. Family contact as required
Buchkremer 1997  Psychoeducational medication training and key person counsellingyesProject staff10 sessions, first 5 at weekly intervals and 5 at fortnightly intervals

1.Cognitive psychotherapy: 7 sessions weekly and 8 sessions fortnightly

2. Leisure time group activities led by students of psychology or educational sociology

Chan 2007Psychoeducation & relapse preventionnoOccupational therapistTen 50-minute sessions over two weeks for patientsTraditional ward occupational therapy including work, rest, leisure activities
Chen 2003Medication & symptom management training programmenot reportedNot reportedNot reportedAntipsychotic therapy
Chien 2004PsychoeducationyesTrained psychiatric nurses12 bi-weekly 2-hour sessions over 6 months for patients and families together

1. Mutual support group for families only

2. Standard outpatient care

Drury 1996 Cognitive TherapyNoNot reported8 hours a week – four individual and four group sessions, 2 sessions with family engagementRecreation therapy,  leisure and social activities outside the ward
Garety 2008CBT or family interventionNoEach session with 2 therapists - mix of doctoral clinical psychologists and nurses all trained in CBT9 months (intended minimum 12 months, max 20 months) – 1-hour sessions with a mean number of 14.3 sessionTAU- standard of care delivered according to national and local service protocols including antipsychotic medication.
Granholm 2005Cognitive behavioural skills trainingYesEach session led by 2 therapists - mix of doctoral or senior graduate level psychologists24 two-hour weekly group sessionsTAU – continuation of prior treatment
Gumley 2003  Targeted CBT for relapseyesClinical psychologist5 session engagement phase between 0-12 weeks; intensive targeted phase (2-3 sessions per week) after for 12 monthsOngoing medication, regular psychiatric review, follow-up from a key worker, access to multidisciplinary community mental health team
Harris 2009Medication management training programmeyesNot reportedCommunity mental health professionals (CMHPs; nurses, occupational therapists, social workers) received a 10-day programme: 3 days assessment training, followed by seven fortnightly study days. These were followed by 1-hour individual supervision sessions, at the CMHP’s workplace, every month for 6 months, giving a total educational input of 9 months.Waiting list
Herz 2000Relapse preventionyesMaster's level nurse clinician or certified social worker

One-hour weekly supportive group therapy OR 30-45-minute individual supportive therapy

90-minute multifamily psychoeducation groups bi-weekly for 6 months and monthly after

Individual supportive therapy and medication management biweekly for 15-30 minutes
Hogarty 1997aPersonal therapy (relapse prevention) and/or family therapy (psychoeducation/ management)yesPersonal and family therapy: master's level psychiatric nurse clinical specialists and doctoral- level clinical psychologists

30-45-minute weekly for about 3 years

1.9 additional monthly medication management sessions

Supportive therapy
Horan 2009Symptom managementyesNot reported12 one-hour groups that met twice weeklySocial cognitive intervention
Kopelowicz 1998bCommunity re-entry programyes2 trainers per session – from a multi-disciplinary staff of 10Eight 45-minute group sessions, twice a day for 4 days in a week.Occupational therapy
Kopelowicz 2003  Family assisted skills training programyesNurses, psychologists and social workers

90-minute sessions four times per week for 3 months.

Weekly group sessions with families included

Case management by social workers and monthly 20-minute psychiatric visits to see a psychiatrist
Kuipers 1997CBTYesTherapistWeekly sessions that become fortnightly- up to one hour and up to nine monthsStandard care- case management and medication
Li 2003Comprehensive patient/ family education guide (CP/FEG)yesWard NursesIn hospital- Patient 8 hours; family member 36 hours. Post-discharge- patient and family member 2 hours per month for 3 months.Not reported
Liberman 1998Social and Independent Living Skills Program consisting of basic conversation, recreation for leisure, medication management, and symptom managementyesOccupational therapist and paraprofessionals6 months of intensive, clinic-based treatment 3 hours a day, 4 days per week.Psychosocial occupational therapy
McDonell 2006  Multiple family group treatmentyesclinicians24 sessions in year 1 and 12 sessions in year 2Outpatient services: medication management, case management, phone consultation, crisis services, group psychotherapy, occupational therapy, individual psychotherapy
Merinder 1999  Patient and relative psychoeducationyesnot stated8 weekly sessions separately for patients and relativesPsychopharmacological treatment, psychosocial rehabilitation efforts, some supportive psychotherapy
Norman 2002  Stress management programyesNurse case managers12 weeks of group sessions, and 12 individual sessionsSocial activities program similar in frequency and length to treatment group
Shin 2002Psychoeducational group  program with individual supportive therapyyes

Group sessions – psychiatric social worker

Individual sessions – Master’s degree student

10 weekly group sessions (90 minutes each) & individual sessions (45 minutes each)Individual supportive therapy
Shon 2002Self management educational groupYes2 psychiatric nurses & 2 social workers12 group sessions (70 minutes each)Non-equivalent control group - undefined
Tait 2002  EWS recognition and targeted cognitive therapyyesTherapist

5 one-hour sessions on initial engagement and formulation

Early signs monitoring questionnaire dispatched fortnightly for an average of 10 months

Targeted cognitive therapy, intensive for a brief period, about 2-3 sessions per week

Routine appointments with psychiatrist and key worker (community psychiatric worker, social worker or occupational therapist)
Turkington 2006  Modify patient's condition and family approachyesTrained Community Psychiatric Nurseup to 6 hour-long sessions of CBT and/or 3 sessions with carerCare of community mental health teams
Van Meijel 2006  Relapse prevention plan using EWS monitoringyesTrained nursesTherapy for a total of mean (SD) 155 (94) daysCare as usual
Weng 2005  Psychiatric rehabilitationyesInterdisciplinary teams2 months, possibly weekly sessionsStandard inpatient services and activities
Xiang 2001Medication & symptom management training programmeyesPsychiatrist and a psychologist90 to 120-minute sessions, 3 sessions per week for 20 weeksAntipsychotic medication and standard psychological intervention
Xiong 1994Family based interventionyesClinicians

1. Introduction- two to three 45-minute sessions- family +/- patient

2. Monthly 45-minute counselling with patient (average 8.1 contacts per year) and family and 90 minute family group

3. Maintenance- families seen once in 2-3 months and monthly group meetings up to 18 months

2-3 months of medication post discharge +/- out patient follow-up
Zhan 2003Medication & symptom management training programmeyesPsychiatrist and a psychologist60 to 90 minutes per session, 1 session per week for 12 weeksPharmacotherapy and standard psychological intervention
Zhang 2004Life skills, medication & symptom management training programmeyesPsychiatrist and psychiatric nurseGroup training for 90 to 120 minutes, 2 times a week. Then community follow-up by nurses, number of sessions not reportedConventional rehabilitation service (outpatient follow-up every 3 months, or telephone counselling as needed)

1. Target of intervention

1.1 Intervention targeted at patient's own recognition of early warning symptoms;
1.2 Interventions targeted at recognition of early warning symptoms by carers or health professionals;
1.3 Interventions that are targeted at both patient and carers.

2. Individual or group treatment

3. Focus of intervention: early warning signs is primary focus of intervention or one component of a multifaceted intervention.

4. Intervention as an addition or replacement for treatment as usual

4.1 Intervention targeting early warning symptoms plus treatment as usual (TAU) versus treatment as usual not including psychological treatment;
4.2 Early warning signs intervention plus treatment as usual versus another psychological treatment not involving early warning signs intervention plus treatment as usual;
4.3 Early warning signs intervention with intermittent medication use in presence of early warning signs without continued medication versus treatment as usual using continued medication;
4.4 Early warning signs intervention involves only help seeking from others or early warning signs intervention involves other self-directed coping.

Results of the search

The search of electronic databases retrieved a total of 1784 citations. Agreement about which reports may have been randomised was 100% and 205 of the original reports were selected and ordered. Thirty-four trials met all the criteria for study selection for this review (see 'Characteristics of included studies').

Included studies

The current review includes 41 reports describing 34 studies (Anzai 2002; Bauml 2007; Bradley 2006; Buchkremer 1997; Chan 2007; Chen 2003; Chien 2004; Drury 1996; Garety 2008; Granholm 2005; Gumley 2003; Harris 2009; Herz 2000; Hogarty 1997a; Horan 2009; Kopelowicz 1998a; Kopelowicz 2003; Kuipers 1997; Li 2003; Liberman 1998; McDonell 2006; Merinder 1999; Norman 2002; Shin 2002; Shon 2002; Tait 2002; Turkington 2006; Van Meijel 2006; Vreeland 2006; Weng 2005; Xiang 2001; Xiong 1994; Zhan 2003; Zhang 2004). This review includes data on 3554 randomised people from within these 34 separate trials.

Details of contact with authors can be found in Appendix 1.

1. Methods

All studies were stated to be randomised. Thirty-two studies were randomised controlled trials (RCTs) and two were cluster-RCTS (Harris 2009; Li 2003). Due to the nature of the intervention, none of the studies were double-blind, 16 studies were single-blind, stating that the assessors were blinded to the participant's treatment allocation (Bauml 2007; Bradley 2006; Chen 2003; Chien 2004; Garety 2008; Granholm 2005; Herz 2000; Kopelowicz 2003; Liberman 1998; Merinder 1999; Norman 2002; Turkington 2006; Vreeland 2006; Xiang 2001; Zhan 2003; Zhang 2004), four studies were not blinded or had inadequate blinding (Buchkremer 1997; Gumley 2003; Harris 2009; Kuipers 1997) and blinding was unclear in the remaining 14 trials. For further details please see sections below on Allocation (selection bias) and Blinding (performance bias and detection bias).

2. Duration

Most trials were undertaken between nine months and two years (Bauml 2007; Bradley 2006; Buchkremer 1997; Chan 2007; Chen 2003; Chien 2004; Drury 1996; Gumley 2003; Harris 2009; Herz 2000; Kopelowicz 1998a; Kopelowicz 2003; Kuipers 1997; Li 2003; Liberman 1998; Merinder 1999; Norman 2002; Tait 2002; Turkington 2006; Van Meijel 2006; Xiang 2001; Xiong 1994; Zhan 2003; Zhang 2004), six studies were less than six months (Anzai 2002; Granholm 2005; Horan 2009; Shin 2002; Vreeland 2006; Weng 2005) and two studies had three years follow-up (Hogarty 1997a; McDonell 2006).

3. Participants

All participants were people with a chronic mental illness, mostly with schizophrenia and schizophrenia-like disorders. All studies except five included both sexes; Anzai 2002, Weng 2005 and Zhan 2003 did not report the sex of the participants, Chan 2007 and Liberman 1998 included only male participants. The mean age for studies that reported this ranged from 28 years to 54 years.

4. Setting

All studies were in a secondary care setting. Eleven studies took place in North America, 11 in Europe, 11 in East Asia and one in Australia.

5. Interventions

For full details of the interventions please see Characteristics of included studies, Table 2 and Table 3.

Thirty-two studies include early warning signs recognition as part of other psychological interventions. Six studies had recognition of early warning signs as a primary focus (Garety 2008; Herz 2000; Liberman 1998; Shon 2002; Tait 2002; Van Meijel 2006). In all studies early warning signs interventions provided education to the patient about the nature of schizophrenia as well as early warning sign recognition and management in addition to treatment as usual. Twenty-three studies had the control group with only treatment as usual, three studies had a control group including treatment as usual and another psychological therapy, in four studies the control groups included treatment as usual and occupational therapy (Anzai 2002; Chan 2007; Kopelowicz 1998a; Liberman 1998), three also included social and recreational activities (Drury 1996; Norman 2002; Vreeland 2006), and in one study, the control group received social skills training (Horan 2009). Six studies delivered the intervention to individuals, 16 delivered it in groups, five to both groups and individually, and in the remaining studies it was either unclear or not reported.

In 20 studies the early warning signs training was delivered to the patients only. In 11 studies the families or carers of the participants were included in the early warning signs programme, and in one study family involvement was optional (Shin 2002). In two of these studies, treatment as usual also included family support services (Bradley 2006; Chien 2004).  Gumley 2003 and Garety 2008 used cognitive behavioural therapy (CBT) as part of the early warning signs intervention and in Buchkremer 1997, one of the early warning signs interventions was cognitive psychotherapy.

In three studies the intervention was delivered to the carer/health professional only. In Li 2003, a cluster-RCT, although the study included interventions for both the patients and their families, only families received early warning signs training. In Garety 2008, there were two treatment pathways: one for patients who had a carer and one for patients without carers. In the no-carer pathway patients were randomised to CBT with early warning signs training or treatment as usual. In the carer pathway, patients were randomised to CBT with early warning signs training, treatment as usual or family intervention in which only the family members received early warning signs training. Harris 2009 was a cluster-RCT in which community health professionals were trained in an experimental medication management training programme that included early warning signs work and relapse prevention.

6. Outcomes scales

Scales that provided useable data are described below.

6.1 Mental state

i) Positive and Negative Syndromes Scale - PANSS (Kay 1986) in Chen 2003, Garety 2008, Gumley 2003, Kopelowicz 2003 and Vreeland 2006.
This scale is used for measuring symptom reduction of patients with schizophrenia and was conceived as an operationalised, drug-sensitive instrument that provides balanced representation of positive and negative symptoms and gauges their relationship to one another and to global psychopathology. It consists of four scales measuring positive and negative syndromes, their differential and general severity of illness. The name refers to the two types of symptoms in schizophrenia as defined by the American Psychiatric Association: positive symptoms, which refer to an excess or distortion of normal functions, and negative symptoms, which represent a diminution or loss of normal functions. High scores suggest greater psychopathology.
PANSS positive syndrome range from seven to 48
PANSS negative syndrome range from seven to 48
PANSS general psychopathology range from 16-96

ii) Brief Psychopathology Rating Scale – BPRS (Overall 1962) in Bauml 2007, Bradley 2006, Buchkremer 1997, Chien 2004, Kuipers 1997, Li 2003, Liberman 1998; Merinder 1999, Shin 2002, Weng 2005, Xiang 2001 and Zhan 2003.
A brief rating scale used to assess the severity of a range of psychiatric symptoms, including psychotic symptoms. The scale has 16 items, and each item can be defined on a seven-point scale varying from 'not present' (0) to 'extremely severe' (6). High scores suggest greater psychopathology.

iii) Scale for the Assessment of Negative Symptoms - SANS (Andreasen 1983) in Bradley 2006, McDonell 2006 and Zhang 2004.
This scale allows a global rating of the following negative symptoms: alogia (impoverished thinking), affective blunting, avolition-apathy, anhedonia-asociality, and attention impairment. Assessments are made on a six-point scale from zero (not at all) to five (severe). Higher scores indicate more symptoms.

iv) Scale for the Assessment of Positive Symptoms - SAPS (Andreasen 1983) in Zhang 2004.
This scale is designed to assess positive symptoms that occur in schizophrenia. On a zero to five-point system, the scale measures four domains including hallucinations, delusions, bizarre behaviour and positive formal thought disorder. Higher scores indicate more symptoms.

v) Psychiatric Assessment Scale – PAS (Krawiecka 1977) in Drury 1996.
This is a simple, five-point scale that measures the symptomatology of chronic psychotic patients. It consists of eight categories of symptoms: depression, anxiety, hallucinations, delusions, flattened incongruous affect, psychomotor retardation, incoherence and irrelevance of speech, and poverty of speech. High scores indicate greater psychopathology.

vi) Krawiecka-Goldberg-Vaughan scale - KGV (Krawiecka 1977) in Harris 2009.
This scale measures symptoms of psychotic illnesses such as schizophrenia and bipolar affective disorder. The measure comprises 14 symptoms and each symptom is rated on a scale from zero to three. Higher scores indicate more symptoms.

6.2 Social Functioning

i) Rehabilitation Evaluation Hall and Baker - REHAB (Baker 1988) in Anzai 2002.
This is a 23-item behaviour rating scale for use with people with chronic psychiatric disability. It measures personal, social and life skills that are important in reintegrating into the community. Scores range from five to 115. Higher scores indicate worse functioning.

ii) Health of the Nation Outcome Scale – HoNOS (Wing 1998) in Bradley 2006.
This scale combines 12 outcomes scales, each measured on a five-point scale, which covers physical, personal and social problems associated with mental illness. High score indicate worse functioning.

iii) Global Assessment of Functioning - GAF (APA 1987) in Bauml 2007 and Merinder 1999.
This scale measures the level of psychological, social, and occupational functioning of psychiatric patients. Possible scores range from one to 90. High scores indicate better functioning.

iv) Global Assessment of Functioning GAF-DIS (Goldman 1992) in Vreeland 2006.
This version of the GAF focuses on functioning, and does not combine this with symptoms. Scores also range from one to 90. High scores indicate better functioning.

v) GAS Global Assessment Scale - GAS (Endicott 1976) in Buchkremer 1997, Li 2003 and Liberman 1998.
This is an observer-rated scale for evaluating the overall functioning of an individual during a specified time period on a continuum from psychological or psychiatric sickness to health. Score ranges from zero to 100, where a higher score indicates better functioning.

vi) Independent Living Skills Survey – ILSS (Wallace 2000) in Kopelowicz 2003.
This survey measures the functional living skills of individuals with severe and persistent mental illness. It includes 103 items covering basic community living skills and is completed by a key relative on an 11-point Likert scale. A high score indicates better functioning.

vii) Specific Level of Functioning assessment scale - SLOF (Schneider 1983) in Chien 2004.
This scale includes 43 behavioural items measured on a five-point Likert scale. There are six domains that are measured: physical functioning, personal care skills, interpersonal relationships, social acceptability, activities of community living, and work skills. A high score indicates better functioning.

viii) Time budget interview (Jolley 2006) in Garety 2008.
The time budget interview is a measure of social functioning that records activity. It is an interview with participants, in which a week-long diary is retrospectively completed in four time periods for each day. It covers domestic activities, social contacts, work and leisure. The activities are then scored, higher scores represent more complex activities, and better functioning.

ix) Social and Occupational Functioning Assessment Scale - SOFAS (APA 1994) in Garety 2008.
This scale measures social and occupational functioning. It is rated from zero to100, high scores indicate better functioning.

x) SDSS (WHO 1998) in Chen 2003 and Zhang 2004.
This scale measures 10 aspects of social activities and roles. It is a Chinese simplified version of the WHO's Disability Assessment Schedule. Higher scores indicate worse functioning.

xi) Social Functioning Scale - SFS (Birchwood 1990) in Harris 2009.
This is a clinician rated scale that measures the level of social functioning seven areas. Higher scores indicate higher functioning.

xii) Nurses' Observation Scale for Inpatient Evaluation - NOSIE (Honigfeld 1965) in Chen 2003, Li 2003 and Zhan 2003.
This is a 30-item scale designed to assess the behaviour of patients on an inpatient unit. It is simple to administer, and may be used to assess patients that may be too ill to participate in more interactive rating scales. High scores indicate better functioning.

6.3 Satisfaction with care

i) International Association of Psychosocial Rehabilitation Services toolkit – IAPSRS (Arns 2001) in Vreeland 2006.
This is self-reported measure of 20-items on four sub scales: empowerment, mastery, life satisfaction and programme satisfaction. A high score indicates more satisfaction.

ii) Versona Service Satisfaction Scale – VSSS (Ruggeri 1993) in Merinder 1999.
This scale is a method for measuring service satisfaction of patients with mental health services. It consists of 82 items, measured on a five-point Likert scale. High scores indicate more satisfaction.

iii) California Psychotherapy Alliance Scale - CALPAS (Gaston 1993) in Harris 2009.
This is a 24-item scale that measures participants' perception of involvement in treatment. Higher scores indicate higher perception of involvement.

6.4 Knowledge

i) Knowledge about Schizophrenia Questionnaire – KASQ (Ascher-Svanum 1999) in Vreeland 2006.
This scale assess schizophrenic patients’ knowledge about their illness and its management, including the diagnosis of schizophrenia and its prevalence, aetiology, course and prognosis, knowledge about medication and its side-effects, psychological treatments, stress factors and legal issues. It is a 25-item multiple-choice test. High scores indicate greater knowledge.

ii) Insight scale – IS (Birchwood 1994) in Merinder 1999.
This is an eight-item questionnaire, which measures on a three-point scale insight in psychosis across three factors: awareness of illness, need for treatment, and attribution of symptoms. Higher scores indicate better insight.

ii) Scale of Unawareness of Mental Disorder - SUMD (Amador 1993) in Chan 2007.
This is a 20-item scale measuring the past and present awareness of having a mental disorder, the effect of medications, the consequences of mental illness, and the awareness and attribution of the symptom items. It is measured on a five-point scale. Higher scores indicate worse insight.

iii) Insight and Treatment Attitude Questionnaire - ITAQ (McEvoy 1989) in Chen 2003 and Zhang 2004.
The ITAQ is a semi-structured interview of 11 items that measures awareness of illness (first five items) and attitude to medication/hospitalisation and follow-up evaluation (six items). Scores range from zero (no insight) to 22 (full insight).

iv) Drug Attitude Inventory - DAI (Hogan 1983) in Harris 2009.
This is a self-rating scale that measures attitude to treatment. It consists of 30 statements about attitudes to psychotropic medication. Total scores can range from +30 to -30, with positive scores predicting adherence to medication.

v) Knowledge About Schizophrenia Interview - KASI (Tarrier 1990) in Li 2003.
This interview measures families and carers' knowledge about schizophrenia, including six areas: diagnosis, symptomology, aetiology, medication, prognosis and management. Higher scores indicate more knowledge.

6.5 Quality of Life

i) Quality of Life Scale - QLS (Heinrich 1984) in Bradley 2006.
This six-point quality of life scale has been designed as an outcome instrument for schizophrenic deficit syndrome as well as to measure impaired functioning in studies of chronic schizophrenia, to assess the deficit syndrome's impact on the patient's life. There are seven severity steps (zero to six, six being adequately functioning and zero being deficient). The time frame is one month. Four item categories have been identified by factor analysis 1) interpersonal relationships (seven items), 2) instrumental role (four items), 3) intrapsychic function (seven items) and 4) commonplace objects and activities. Higher scores indicate greater quality of life.

ii) Lancashire Quality of Life Profile - LQLP (Priebe 1998) in Bauml 2007.
This scale includes both objective and subjective measures of quality of life. The 27-item subjective measures cover nine domains: work and education, leisure and participation, religion, finance, living situation, legal and safety, family relations, social relations and health. it is scored on a seven-point scale. High scores indicate greater quality of life.

iii) EuroQoL (Brazier 1993) in Garety 2008.
The EuroQol is a self-reported questionnaire, measuring health-related quality of life. Higher scores indicate greater quality of life.

iv) Lehman Quality of Life Scale - LQoLS (Lehman 1988) in Liberman 1998.
This is a structured, 45-minute qualify of life Interview for the chronically mentally ill. Higher scores indicate greater quality of life.

6.6 Burden of care

i) Family Burden Interview Schedule – FBIS (Pai 1981) in Bradley 2006 and Chien 2004.
This scale consists of six measures of burden, with each item rated on a three-point scale. It includes financial burden, effects on family routine, effects on family leisure, effects on family interaction, effects on physical health of family members and effects on mental health of other family members. High scores indicate greater caregiver burden.

ii) Family Attitude Scale - FAS (Kavanagh 1997) in Li 2003.
This is a 30-item self-report questionnaire with brief measures of anger, burden or ‘stress’ of relatives towards a patient. Respondents identify how often each statement is true, on a five-point scale from “never” to “everyday”, and responses are summed to give a zero to 120 score.

Excluded studies

We excluded 86 studies that did not meet the criteria for the review, mainly because they either did not include education to increase awareness of early signs of recurrence or the early warning signs intervention was used for both control and experimental groups (see Characteristics of excluded studies).

Studies awaiting assessment

No studies are awaiting assessment.

Ongoing studies

We found seven ongoing studies (see Characteristics of ongoing studies).

Risk of bias in included studies

We prepared a 'Risk of bias' assessment for each trial. Our judgments regarding the overall risk of bias in individual studies is illustrated in Figure 1 and Figure 2. Overall, we felt the risk of bias in the included studies to be moderate.

Figure 1.

'Risk of bias' graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 2.

'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study.

Allocation

All studies were reported as randomly assigned. However, 19 studies did not clearly describe how generation of randomisation sequence was undertaken and had an unclear risk of bias. Only six of the studies had a low risk of bias for allocation concealment, the risk of bias was unclear in the remaining studies.

Blinding

None of the studies were double-blind due to the nature of the intervention, therefore, all studies were rated as low risk of bias. Fourteen studies used assessors that were blinded and were rated as low risk of bias; in 17 studies the risk of bias was unclear and two studies used assessors that were not blind to allocation of treatment condition and these were rated as high risk of bias.

Incomplete outcome data

Fourteen of the studies had a low risk of bias for incomplete outcomes data, 17 trials had an unclear risk of bias and risk of bias was high in two studies.

Selective reporting

Twenty studies were rated as low risk of bias for selective outcome reporting, 11 studies had a high risk of bias for selective outcome reporting, and it was unclear in two studies.

Other potential sources of bias

Seven studies were of low risk of bias for other potential sources of bias, five studies were subject to other biases as they were either partly or fully funded by the pharmaceutical industry, and the remainder had an unclear risk of bias.

Effects of interventions

See: Summary of findings for the main comparison Early warning signs training and treatment as usual compared to treatment as usual for schizophrenia

Comparison 1: Early warning signs training + treatment as usual (TAU) versus TAU

Primary outcomes

For the primary outcomes, we have presented the data in subgroups according to who received the training, but the outcomes were measured in the patients. See subgroup analyses for details.

1. Relapse
1.1 Relapse: 1. Relapse

Fifteen trials with a follow-up duration of two to 36 months provided data on the rate of relapse of participants (Analysis 1.1). Early warning signs intervention was found to be superior to treatment as usual in preventing relapses (15 RCTs, n = 1502, risk ratio (RR) 0.53 95% confidence interval (CI) 0.36 to 0.79). This data however showed a very high heterogeneity (I2 n = 81%), mostly due to the heterogeneity observed in studies where the intervention was delivered only to the patients (I2 n = 89%). See subgroup analyses below for more details.

1.2 Relapse: 2. Time to first relapse

In six trials with a follow-up duration of 12 to 18 months, time taken to relapse after treatment was no longer when early warning signs intervention was delivered, compared to treatment as usual (6 RCTs, n = 550, hazard ratio (HR) 0.53, 95% CI 0.27 to 1.06; Analysis 1.2). The pooled results showed high heterogeneity (I2 = 72%).

2. Service use
2.1 Service use: 1 Re-hospitalisation

Fifteen trials with a follow-up duration of two months to three years provided data regarding the outcome rate of re-hospitalisation. Early warning signs intervention was found to be superior to treatment as usual in preventing re-hospitalisation of participants (15 RCTs, n = 1457, RR 0.48, 95% CI 0.35 to 0.66; Analysis 1.3). The pooled data showed moderately high heterogeneity (I2 = 62%).

In addition, Buchkremer 1997 and Bauml 2007 report a rate of re-hospitalisation for available participants at a follow-up of five years and seven years respectively (Analysis 1.4). No significant difference was found in favour of early warning signs intervention over treatment as usual (2 RCTs, n = 156, RR 0.82 95% CI 0.49 to 1.36).

2.2 Service use: 2 Time to re-hospitalisation

In six trials with a follow-up duration of one to two years, time to re-hospitalisation was no longer when early warning signs intervention was delivered, compared to treatment as usual (6 RCTs, n = 1149, HR 0.58, 95% CI 0.33 to 1.04; Analysis 1.5). This data however showed a very high heterogeneity (I2 = 83%), mostly due to the heterogeneity observed in studies where the intervention was delivered only to the patients (I2 = 89%). See subgroup analyses below for more details.

Secondary outcomes

For the outcomes mental state, social functioning, burden of care, medication compliance and economic outcomes, some studies reported skewed data, which were not entered into the analysis. These are reported in Table 1.

1. Mental state
1.1 Mental state: general 1a. Average score (BPRS, high = bad)

Ten trials with a follow-up duration of 10 weeks to two years provided data regarding the outcome mental state on the BPRS. Early warning signs intervention was found to be superior to treatment as usual in improving the mental state of participants (9 RCTs, n = 685, 1 cluster RCT, n = 8 clusters, mean difference (MD) -4.55, 95% CI -8.21 to -0.90; Analysis 1.6). The pooled results show very high heterogeneity (I2 = 93%).

When the outlying studies Xiang 2001 and Zhan 2003 are removed from the analysis, there is still a significant difference in favour of early warning signs training, although the magnitude of the effect is less (8 RCTS, n = 339, 1 cluster RCT, n = 8 clusters, MD -2.64 95% CI -4.62 to -0.67) much of the heterogeneity is reduced (I2 = 38%).

Weng 2005 also reported data for mean change in mental state on the BPRS but we could not pool this data because they measured the mean change from discharge to follow-up, rather than pre-intervention to follow-up.

1.2 Mental state: general 1b. Avergae score (PANSS, high = bad)

Three trials with a follow-up duration of nine months to two years provided data regarding the outcome mental state on the PANSS. Early warning signs intervention was not found to be superior to treatment as usual in improving the mental state of participants (3 RCTs, n = 385, MD -5.89, 95% CI -15.86 to 4.09; Analysis 1.7). The pooled results show very high heterogeneity (I2 = 91%).

When the outlier Chen 2003 is removed, there is no heterogeneity (I2 = 0%), and the early warning signs training is not found to be superior to treatment as usual in improving mental state on the PANSS (2 RCTS, n = 322, MD -1.45 95% CI -4.70 to 1.79).

1.3 Mental state: general 1c. Average score (KGV, high = bad)

One cluster randomised trial with a follow-up duration of nine months measured mental state on the KGV scale in patients whose community mental health professional had received early warning signs training. A significant difference was found in mental state in favour of early warning signs training (1 cluster RCT, n = 50 clusters, MD -2.60 95% CI -4.07 to -1.13; Analysis 1.8).

1.4 Mental state: specific 2. Average score in positive symptoms (various scales)

(Analysis 1.9)

1.4.1 BPRS (high = bad)

One trial with a follow-up duration of 18 months provided data on mean endpoint score on positive symptoms on the BPRS scale. Early warning signs intervention was not superior to treatment as usual in improving positive symptoms (1 RCT, n = 64, MD -0.90, 95% CI -3.52 to 1.72)

1.4.2 PANSS positive (high = bad)

Four trials with a follow-up duration of 24 weeks to two years provided data on mean endpoint score on positive symptoms on the PANSS positive scale. Early warning signs intervention was not found to be superior to treatment as usual in improving the mental state of participants (4 RCTs, n = 508, MD -0.41, 95% CI -2.70 to 1.87). The pooled results show high heterogeneity (I2 = 86%).

When Chen 2003, which is an outlier, is removed from the analysis, early warning signs intervention remains not superior to treatment as usual (3 RCTS, n = 445, MD 0.42 95% CI -1.09 to 1.92), but the level of heterogeneity is moderate (I2 = 50%).

1.4.3 SAPS (high = bad)

One trial with a follow-up duration of one year provided data on positive symptoms on the PAS positive scale. Early warning signs intervention was found to be superior to treatment as usual in improving positive symptoms (1 RCT, n = 104, MD -2.04, 95% CI -3.22 to -0.86).

1.5 Mental state: specific 3. Average score in negative symptoms (various scales)

(Analysis 1.10)

1.5.1 PANSS negative (high = bad)

Four trials with a follow-up duration of 24 weeks to two years provided data on negative symptoms on the PANSS negative scale. Early warning signs intervention was not found to be superior to treatment as usual in improving the mental state of participants (4 RCTs, n = 508, MD -0.74, 95% CI -3.13 to 1.64). The pooled results show high heterogeneity (I2 = 86%).

When Chen 2003, which is an outlier, is removed from the analysis, early warning signs intervention remains not superior to treatment as usual (3 RCTS, n = 445, MD 0.10 95% CI -1.48 to 1.69), the level of heterogeneity is moderate (I2 = 54%).

1.5.2 SANS (high = bad)

Three trials with a follow-up duration of one year to three years provided data on negative symptoms on the SANS scale. Early warning signs intervention was not found to be superior to treatment as usual in improving the mental state of participants (3 RCTs, n = 194, MD -3.95, 95% CI -15.25 to 7.35). The pooled results show extremely high heterogeneity (I2 = 99%).

When the study Zhang 2004, which is an outlier, is removed from the analysis, early warning signs intervention remains not superior to treatment as usual (2 RCTs, n = 90, MD -0.95, 95% CI -5.41 to 7.30), the level of heterogeneity is moderate (I2 = 58%).

2. Social functioning
2.1 Social functioning: 1. Not employed

Two trials with a follow-up duration of two months to one year provided data on the number of participants not in employment at the end of the study (Analysis 1.11). Early warning signs intervention was found to be superior to treatment as usual (2 RCTs, n = 185, RR 0.68, 95% CI 0.57 to 0.82). The pooled results showed no heterogeneity (I2 = 0%).Xiong 1994 also reported on the mean number of months of employment of participants at 18 months follow-up. They found a significant difference in favour of early warning signs intervention (1 RCT, n = 58, MD -4.86, 95% CI -8.78 to -0.94).

2.2 Social functioning: 2. Average scores (various scales)

(Analysis 1.12; Analysis 1.13)

Social functioning was measured on 11 different scales, with one to three studies providing data for each scale. On nine of these scales early warning signs training was not found to be superior to treatment as usual in improving social functioning, and on two of the scales used a significant difference was found in favour of the intervention.

2.2.1 GAF (high = good)

Two trials with a follow-up duration of one to two years provided data on social functioning on the GAF scale. Early warning signs intervention was not found to be superior to treatment as usual in improving social functioning (2 RCTs , n = 88, MD 2.51 95% CI -3.84 to 8.85). Pooled data showed no heterogeneity (I2 = 0%).

2.2.2 GAF-DIS (high = good)

One trial with a follow-up duration of 24 weeks provided data on social functioning on the GAF-DIS scale. Early warning signs intervention was not found to be superior to treatment as usual in improving social functioning (1 RCT, n = 61, MD -6.10, 95% CI -12.34 to 0.14).

2.2.3 ILSS (high = good)

One trial with a follow-up duration of nine months provided data social functioning on the ILSS scale. Early warning signs intervention was not found to be superior to treatment as usual in improving social functioning (1 RCT, n = 84, MD 0.35, 95% CI -0.70 to 1.40).

2.2.4 Time budget interview (high = good)

One trial with a follow-up duration of two years provided data on social functioning using the time budget interview. Early warning signs intervention was not found to be superior to treatment as usual in improving social functioning (1 RCT, n = 224, MD 1.61 95% CI -4.04 to 7.25).

2.2.5 SOFAS (high = good)

One trial with a follow-up duration of two years provided data on social functioning on the SOFAS scale. Early warning signs intervention was not found to be superior to treatment as usual in improving social functioning (1 RCT, n = 236, MD 1.21 95% CI -2.71 to 5.13).

2.2.6 SLOF (high = good)

One trial with a follow-up duration of 18 months provided data on social functioning on the SLOF scale. In contrast to the results on the other scales, early warning signs intervention was found to be superior to treatment as usual in improving social functioning (1 RCT, n = 64, MD 27.50, 95% CI 17.03 to 37.97).

2.2.7 SAS (high = good)

One trial with a follow-up duration of two years provided data on social functioning on the SOFAS scale. Early warning signs intervention was not found to be superior to treatment as usual in improving social functioning (1 RCT, n = 80, MD -0.43 95% CI -2.20 to 1.34).

2.2.8 SFS (high = good)

One trial with a follow-up duration of nine months provided data on social functioning on the SFS scale. Early warning signs intervention was not found to be superior to treatment as usual in improving social functioning (1cluster RCT, n = 56 clusters, MD 4.40, 95% CI -3.16 to 11.96).

2.2.9 GAS (high = good)

Three trials with a follow-up duration of nine months to two years provided data on social functioning on the GAS scale. Early warning signs intervention was not found to be superior to treatment as usual in improving social functioning (2 RCTs, n = 178, and 1 cluster RCT, n = 8 clusters, MD 1.92 95% CI -0.54 to 4.38). The pooled results showed no heterogeneity (I2 = 0%).

2.2.10 NOSIE (high = good)

Three trials with a follow-up duration of nine months to one year provided data on social functioning on the NOSIE scale. Early warning signs intervention was found to be superior to treatment as usual in improving social functioning (2 RCTs, n = 208, 1 cluster RCT, n = 8 clusters, MD 25.12, 95% CI 5.75 to 44.48). The pooled results showed very high heterogeneity (I2 = 96%).

3. Quality of life
3.1 Quality of life: Average scores (various scales)

(Analysis 1.14)

3.1.2 QLS (high = good)

One trial with a follow-up duration of one year provided data on mean endpoint score in quality of life on the QLS scale. Early warning signs intervention was not found to be superior to treatment as usual in improving quality of life (1 RCT, n = 46, MD -5.05, 95% CI -16.12 to 6.02).

3.1.2 LQLP (high = good)

One trial with a follow-up duration of two years provided data on mean endpoint score in quality of life on the LQLP scale. Early warning signs intervention was not found to be superior to treatment as usual in improving quality of life (1 RCT, n = 48, MD -0.40, 95% CI -1.30 to 0.50).

3.1.3 EuroQoL (high = good)

One trial with a follow-up duration of two years provided data on mean endpoint score in quality of life on the EuroQoL scale. Early warning signs intervention was not found to be superior to treatment as usual in improving quality of life (1 RCT, n = 206, MD -4.80, 95% CI -10.79 to 1.19).

3.1.4 LQoLS (high = good)

One trial with a follow-up duration of two years provided data on mean endpoint score in quality of life on the EuroQoL scale. Early warning signs intervention was not found to be superior to treatment as usual in improving quality of life (1 RCT, n = 80, MD 0.09 95% CI -0.24 to 0.42).

4. Satisfaction with care
4.1 Satisfaction with care: Average scores (various scales)

(Analysis 1.15; Analysis 1.16)

4.1.1 CALPAS (high = good)

One trial with a follow-up duration of nine months provided data on mean change in satisfaction with care on the CALPAS scale. Early warning signs intervention was found to be superior to treatment as usual in improving satisfaction with care (1 cluster RCT, n = 56 clusters, MD 0.30, 95% CI 0.12 to 0.48).

4.1.2 IAPSRS (high = good)

One trial with a follow-up duration of 24 weeks provided data on mean endpoint score in satisfaction with care on the IAPSRS scale. Early warning signs intervention was not found to be superior to treatment as usual in satisfaction with care (1 RCT, n = 60, MD 0.10, 95% CI -0.11 to 0.31).

4.1.3 VSSS (high = good)

One trial with a follow-up duration of one year provided data on mean change in satisfaction with care on the VSSS scale. Early warning signs intervention was not found to be superior to treatment as usual in improving satisfaction with care, although the data may be skewed (1 RCT, n = 32, MD 2.15, 95% CI -9.66 to 13.96).

5. Adverse events
5.1 Adverse events: 1. Lost to follow-up

Seventeen trials with follow-up duration of 10 weeks to two years provided data on the outcome lost to follow-up (Analysis 1.17). No significant difference was found between early warning signs intervention and treatment as usual in the number of participants lost to follow-up (17 RCTs, n = 1495, RR 0.96 95% CI 0.72 to 1.28). The pooled data showed no heterogeneity (I2 = 0%).

Shon 2002 also reported data on losses to follow-up but the length of follow-up is not stated in the report and so we have not pooled these data. Two participants in the early warning signs intervention group were lost to follow-up and none in the treatment as usual group.

5.2 Adverse events: 2. Death

Four trials with a follow-up duration of one to two years provided data on the number of deaths (Analysis 1.18). No significant difference was found between early warning signs intervention and treatment as usual in the number of deaths during the trials (4 RCTs, n = 462, RR 0.91, 95% CI 0.29 to 2.89). The pooled data showed low heterogeneity (I2 = 4%).

6. Knowledge
6.1 Knowledge: 1. Average scores (various scales)

(Analysis 1.19; Analysis 1.20; Analysis 1.21)

6.1.1 KASQ (patients, high = good)

One trial with a follow-up duration of 24 weeks provided data for this outcome. Early warning signs intervention was not found to be superior to treatment as usual in increasing participants' knowledge about their illness (1 RCT, n = 61, MD 1.60, 95% CI -0.84 to 4.04).

6.1.2 IS (patients, high = good)

One trial with a follow-up duration of 24 weeks provided data for this outcome. Early warning signs intervention was not found to be superior to treatment as usual in increasing participants' knowledge about their illness (1 RCT, n = 40, MD 0.96, 95% CI -0.48 to 2.40).

6.1.3 ITAQ (patients, high = good)

Two trials with a follow-up duration of one year provided data for this outcome. Early warning signs intervention was found to be superior to treatment as usual in increasing participants' knowledge about their illness (2 RCTs, n = 166, MD 7.21, 95% CI 0.94 to 13.48). The pooled data showed very high heterogeneity (I2 = 93%).

6.1.4 SUMD (patients, high = bad)

One trial with a follow-up duration of one year provided data for this outcome. Early warning signs intervention was not found to be superior to treatment as usual in increasing participants' knowledge about their illness (1 RCT, n = 81, MD -1.18, 95% CI -2.46 to 0.10).

6.1.5 DAI (patients, high = good)

One trial with a follow-up duration of one year provided data for this outcome. Early warning signs intervention was not found to be superior to treatment as usual in increasing participants' knowledge about their illness (1 cluster RCT, n = 56 clusters, MD 1.40, 95% CI -0.97 to 3.77).

6.1.6 KASI (relatives, high = good)

One trial with a follow-up duration of one year provided data for this outcome. Early warning signs intervention was found to be superior to treatment as usual in increasing families' knowledge about the illness of their family member (1 cluster RCT, n = 8 clusters, MD 2.83, 95% CI 0.93 to 4.73).

7. Burden of care
7.1 Burden of care: 1. Average scores (FBIS, high = bad)

Two trials with a follow-up duration of one year and 18 months provided data for the outcome burden of care (Analysis 1.22). Early warning signs intervention was not found to be superior to treatment as usual in reducing the burden of care (2 RCTs, n = 110, MD 3.29, 95% CI -8.43 to 15.00). The pooled data showed high heterogeneity (I2 = 87%).

8. Medication compliance
8.1 Medication compliance: Not compliant to medication

Three trials with a follow-up duration of nine months to one year provided data for the outcome medication compliance, although Kopelowicz 2003 only reports data post-intervention and not at nine months (Analysis 1.23). Early warning signs training was found to be superior to treatment as usual in improving medication compliance (4 RCTs, n = 374, MD 0.57 95% CI 0.42 to 0.77). The pooled data showed no heterogeneity (I2 = 0%). Xiang 2001 also reported data on mean number of missed doses, but the data were skewed (see Table 1).

9. Economic burden (cost of care)

Two studies presented data on cost of care. Kuipers 1997 showed no significant difference between the mean cost of care for the early warning signs training group and standard care group. McDonell 2006 found that the mean costs of care were less in the early warning signs group than the standard psychiatric group, but the data could not be analysed as no standard deviations were reported. See Table 1 for details.

10. Subgroup analysis
10.1 Relaspe: Time to first relapse

Time taken to relapse after treatment was longer when early warning signs intervention compared to treatment as usual was delivered to patients only (3 RCTs, n = 372, HR 0.26, 95% CI 0.13 to 0.53). In contrast, there was no difference in the time taken to relapse when early warning signs intervention was delivered to both the patients and their carer/health professional (3 RCTs, n = 178, HR 1.00 95% CI 0.59 to 1.68). The pooled results showed low heterogeneity (I2 = 35% and 15%, respectively).

10.2 Relapse: Rate of relapse

Early warning signs intervention was found to be superior to treatment as usual in preventing relapses when the intervention was delivered only to the patients (8 RCTs, n = 933, RR 0.43, 95% CI 0.20 to 0.92), but showed high heterogeneity (I2 = 89%). In contrast, there was no significant difference where the training was delivered to both the patients and their carer/health professional (6 RCTs, n = 491, RR 0.68 95% CI 0.46 to 1.02), or to only the family or the health professionals caring for the patients (2 RCTs, n = 78, RR 0.58 95% CI 0.24 to 1.37). These results showed moderate or no heterogeneity (I2 = 0% and 63%, respectively).

10.3 Service use: Time to re-hospitalisation

Time to re-hospitalisation after treatment was longer when early warning signs intervention was delivered to both the patients and their carer/health professional compared to treatment as usual (3 RCTs, n = 740, HR 0.62, 95% CI 0.46 to 0.83), but showed high heterogeneity (I2 = 89%). In contract, there was no difference in the time taken to relapse when early warning signs intervention was delivered to patients only (3 RCTs, n = 409, HR 0.37 95% CI 0.07 to 1.95). These results showed no heterogeneity (I2 = 0%).

10.4 Service use: Rate of re-hospitalisation

Early warning signs intervention was found to be superior to treatment as usual in preventing re-hospitalisation when the intervention was delivered only to the patients (7 RCTs, n = 791, RR 0.30, 95% CI 0.14 to 0.66), but showed high heterogeneity (I2 = 80%); or when early warning signs training was delivered to both the patients and their carer/health professional (7 RCTs, n = 625, RR 0.64 95% CI 0.51 to 0.79), these results showed no heterogeneity (I2 = 0%). In contrast, there was no significant difference where the training was delivered to only the family or the health professionals caring for the patients (1 RCT, n = 41, RR 0.53 95% CI 0.15 to 1.82).

11. Sensitivity analysis

All studies were described as randomised and we did not have to make assumptions regarding missing SDs data so sensitivity analyses were not undertaken for these reasons.

11.1 Imputed data
11.1.1 Relapse: Time to relapse

When the data that was imputed were removed from the analysis there remains no significant difference in the length of time taken to relapse between early warning signs training and treatment as usual (2 RCTs, n = 226, HR 0.60 95% CI 0.14 to 2.59).

11.1.2 Service use: Time to re-hospitalisation

When the data that was imputed were removed from the analysis there remains no significant difference in the length of time taken to re-hospitalisation between early warning signs training and treatment as usual (3 RCTs, n = 258, HR 0.87 95% CI 0.41 to 1.83).

Discussion

Summary of main results

The summary below reflects the outcomes chosen for the Summary of findings for the main comparison, and considered the main findings of this review that can support evidence-based decision making. Nine studies reported an improvement in general mental state suggesting that early warning signs intervention is not associated with an increase in depressive symptoms (Dixon 1998) as we had postulated before undertaking the review. For all outcomes included in the 'Summary of findings' tables, the quality of evidence was found to be very low. Overall, 32 randomised controlled trials and two cluster randomised trials were identified that met our criteria for an early warning signs intervention in schizophrenia. The studies were conducted over varying time periods with differing periods of follow-up and the actual interventions themselves were of varying lengths, from two weeks to three years. This made comparing the data difficult.

Rate of relapse and rate of re-hospitalisation: We found significant differences in favour of early warning signs interventions in the number of participants relapsing and the number of participants being re-hospitalised. The results are highly heterogeneous, which is not explained by differences in the populations, although the heterogeneity is higher for studies where the intervention was delivered only to patients. Weng 2005 had a much shorter duration of follow-up than the other included trials of two months, but removing this study does not explain the heterogeneity. The heterogeneity may be related to the differences in the nature of the interventions and the length of time over which they were delivered.

Time to relapse and time to re-hospitalisation: Much of the data for these outcomes needed to be imputed as it was not reported in the studies in a way that could be added to the meta-analysis. No significant differences were found in the length of time taken to for participants to relapse or to be re-hospitalised. There was high heterogeneity in the results. For time to relapse, this is probably due to differences in the populations, as subgroup analyses show that when the early warning sign interventions are delivered to patients only, the intervention does significantly increase the length of time before relapsing.  

Only three studies reported on satisfaction with care. Two randomised controlled trials (RCTs) found no significant difference in satisfaction with care between early warning signs training and treatment as usual (although one may have skewed data), and one cluster RCT did find a significant difference in favour of early warning signs training. We found no evidence of a difference in losses to follow-up between the treatment groups. Two studies report on costs of care, but not in a form that could be added to the meta-analysis. One showed no significant difference between the mean cost of care for the early warning signs training group and standard care group, although the data were skewed. In contrast, another found that the mean costs of care were less in the early warning signs group ($1641) than the standard psychiatric group ($5199).

Nine studies reported an improvement in general mental state suggesting that early warning signs intervention is not associated with an increase in depressive symptoms (Dixon 1998) as we had postulated before undertaking the review.

Overall completeness and applicability of evidence

1. Completeness

Most studies reported at least one primary outcome. Many of the trials did not present usable data, such as means or SDs, or the data was from un-validated scales. The results for time to relapse and time to re-hospitalisation have to be treated with some caution as we had to impute data from 58% of the studies because of incomplete reporting of these outcomes. The studies often used different assessment scales and there is no agreement currently on how comparisons can be made between these scales for people suffering from schizophrenia. Therefore, we were unable to pool some of the data, which reduces the possibility of detecting a treatment effect. Very few studies reported data for satisfaction with care, quality of life, medication compliance and economic burden.

2. Applicability

Included studies of early warning signs interventions have international applicability because they have been conducted in many European countries (including United Kingdom), Asia (China, Japan, Korea), Australia and North America. All studies have been carried out in both genders and in the age range 16 to 65 years so the results cannot be generalised to children, adolescents or the elderly. It is unclear to what extent the results can be generalised to people with schizophrenia with comorbid personality disorder, axis 1 disorder, substance use disorders or medical comorbidity because most studies did not report if people with these conditions were included or not.

One limitation of the applicability of the results of the review is that only one study examined the effectiveness of early warning signs interventions alone without any other psychological intervention being applied at the same time. This study (Van Meijel 2006) found a positive benefit of early warning signs intervention on rate of relapse of the similar effect size as the overall review but did not report outcomes such as hospitalisation, social functioning and mental state. Therefore, the only conclusions that can be drawn from the review are on the overall effectiveness of psychological interventions incorporating early warning signs interventions and it is currently unclear how many of the benefits of these interventions can be attributed to training to recognise and manage early warning signs as opposed to other psychological intervention techniques. We also had no data to draw any conclusions concerning the effectiveness of early warning signs interventions coupled with intermittent medication alongside lower dose maintenance antipsychotic medication.

Six studies (Chen 2003; Shon 2002; Weng 2005; Xiang 2001; Zhan 2003; Zhang 2004) examined the medication management and symptom management skills training modules, based on Liberman’s Social Skills training for schizophrenic patients (Liberman 1998), which includes early warning signs training. These showed positive results for time to relapse, rate of relapse, time to hospitalisation, and rate of re-hospitalisation, some of which were outliers in the data. These studies were conducted mostly in China. Liberman 1998, which was conducted in the USA, did not report these outcomes.

Quality of the evidence

The quality of the current evidence is very low based on GRADE. Most studies did not address incomplete data adequately. Only 14 of the 34 included studies described an adequate sequence generation, and only six described the methods used to conceal allocation. Although it is not possible with this type of intervention to blind participants, only 15 studies used assessors that were blind to treatment allocation.

A matter of concern in relation to the robustness of our findings is that much of the provided primary outcome data for time to relapse and time to hospitalisation could not be directly used in the meta-analysis and needed to be imputed. One could argue that this may be due to selective reporting of outcome measures and thereby introducing bias. Furthermore, the results are limited to the extent of follow-up from the trials (Tierney 2007), which was no more than two years for these outcomes, and therefore there is only limited information about the long-term effects of early warning signs interventions. Rates of relapse and re-hospitalisation are also reported at varying time lengths depending on duration of the study, hence compromising comparability.

Potential biases in the review process

A thorough search strategy was used in this review. There may still be gaps in the search strategy such as unpublished data (grey literature), which are difficult to get hold of. We included studies where we were confident that an early warning signs intervention had been attempted. Some trials did not give precise descriptions of the nature of the psychological intervention tested and we did not get a response when we wrote to some authors for clarification.

For most outcomes we are not able to tell if there was publication bias, as no more than 11 trials reported data for these, and a funnel plot could not be performed. For rate of relapse, there does not appear to be publication bias, although for rate of re-hospitalisation and losses to follow-up there appears to be publication bias as the funnel plot indicates that there may be the equivalent number of 'negative' trials that have not been included in this analysis.

Agreements and disagreements with other studies or reviews

We do not know of any other systematic reviews on this topic.

Authors' conclusions

Implications for practice

1. For people with schizophrenia

Early warning signs intervention may reduce the likelihood of relapse and readmission, as well as reduce the length of hospital stay when they are part of a more complete package of care. It may well have other outcomes, that are, at this point, under-researched. However, we do not know if early warning signs will be effective in reducing the likelihood of relapse and readmission, or reduce the length of hospital stay, when they are part of a complete package of care, or if they are delivered on their own

2. For clinicians

We do not know if the reduction in relapse and length of hospital stay with training to recognise early warning sings of relapse will be useful for clinicians to use as a part of their treatment programme. There were some indications from trials that early warning signs delivered to patients/carers/health providers had differential effects on different outcomes, but no consistent patterns of benefits emerged that could reliably inform clinical decisions

3. For managers and policy makers

This review found no evidence of moderate or high quality to suggest that mental health services might wish to consider routinely providing psychological interventions involving the recognition and management of early warning signs to adults with schizophrenia. They may have suggested a cost-effective intervention, had the evidence of benefit in reduced hospitalisation and relapse rates in the pooled estimates been judged of moderate or high quality. There is also insufficient evidence to support the use of early warning signs interventions alone or with intermittent medication use alongside low-dose antipsychotic maintenance medication.

Implications for research

1. For two of the primary outcomes we had to impute time to event data; if we had the main data reported by authors, this data would be more accurate. Following CONSORT for good reporting of clinical trials more closely would have helped to considerably increase the amount of data available in this review.

2. Further research is required to establish whether early warning signs interventions on their own are effective or whether they are effective with additional intermittent medication use alongside low-dose antipsychotic maintenance medication. Such an approach might be low cost, should widely applicable interventions for people with schizophrenia or schizoaffective disorder prove effective.

3. Trials should make use of validated scales and report fully the means and SDs of results, which would allow more data to be pooled in analyses and improve the interpretation of early warning signs interventions.

4. The method of sequence generation, allocation concealment and blinding should be more clearly described to enable researchers to be more confident that risks of bias are low in trials.

5. Future early warning signs interventions trials should report data on relapse and re-hospitalisation, as well as quality of life and economic burdens.

Acknowledgements

The review authors would like to thank the Cochrane Schizophrenia Group for supporting this review. The Cochrane Schizophrenia Group maintains and provides a template for the methods section of their reviews. We have used and adapted this.

We would like to acknowledge and thank Ashely Jones and Paula Williamson who provided statistical expertise for the original version of this review.

Enhance Reviews provided support for this review in the data extraction, analysis and write-up of results.

Data and analyses

Download statistical data

Comparison 1. EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU)
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Relapse: 1. Relapse151502Risk Ratio (M-H, Random, 95% CI)0.53 [0.36, 0.79]
1.1 training delivered to patient only8933Risk Ratio (M-H, Random, 95% CI)0.43 [0.20, 0.92]
1.2 training delivered to carer/health professional only278Risk Ratio (M-H, Random, 95% CI)0.58 [0.24, 1.37]
1.3 training delivered to both patient and carer/health professional6491Risk Ratio (M-H, Random, 95% CI)0.68 [0.46, 1.02]
2 Relapse: 2. Time to first relapse6550hazards ratio (Random, 95% CI)0.53 [0.27, 1.06]
2.1 training delivered to patient only3372hazards ratio (Random, 95% CI)0.26 [0.13, 0.53]
2.2 training delivered to both patient and carer/health professional3178hazards ratio (Random, 95% CI)1.00 [0.59, 1.68]
3 Service use: 1a. Re-hospitalisation151457Risk Ratio (M-H, Random, 95% CI)0.48 [0.35, 0.66]
3.1 training delivered to patient only7791Risk Ratio (M-H, Random, 95% CI)0.30 [0.14, 0.66]
3.2 training delivered to carer/health professional only141Risk Ratio (M-H, Random, 95% CI)0.53 [0.15, 1.82]
3.3 training delivered to both patient and carer/health professional7625Risk Ratio (M-H, Random, 95% CI)0.64 [0.51, 0.79]
4 Service use: 1b. Re-hospitalisation (5-7 year follow-up)2156Risk Ratio (M-H, Random, 95% CI)0.82 [0.49, 1.36]
4.1 training delivered to patient only1108Risk Ratio (M-H, Random, 95% CI)1.04 [0.79, 1.36]
4.2 training delivered to both patient and carer/health professional148Risk Ratio (M-H, Random, 95% CI)0.62 [0.42, 0.92]
5 Service use: 2. Time to re-hospitalisation61149hazards ratio (Random, 95% CI)0.58 [0.33, 1.04]
5.1 training delivered to patient only3409hazards ratio (Random, 95% CI)0.37 [0.07, 1.95]
5.2 training delivered to both patient and carer/health professional3740hazards ratio (Random, 95% CI)0.62 [0.46, 0.83]
6 Mental state: general 1a. Average score (BPRS, high = bad)10 Mean Difference (Random, 95% CI)-4.55 [-8.21, -0.90]
7 Mental state: general 1b. Average score (PANSS, high = bad)3385Mean Difference (IV, Random, 95% CI)-5.89 [-15.86, 4.09]
8 Mental state: general 1c. Average score (KGV, high = bad)1 Mean Difference (Fixed, 95% CI)-2.6 [-4.07, -1.13]
9 Mental state: specific 2. Average score in positive symptoms (various scales)6 Mean Difference (IV, Random, 95% CI)Subtotals only
9.1 BPRS (high = bad)164Mean Difference (IV, Random, 95% CI)-0.90 [-3.52, 1.72]
9.2 PANSS positive (high = bad)4508Mean Difference (IV, Random, 95% CI)-0.41 [-2.70, 1.87]
9.3 SAPS (high = bad)1104Mean Difference (IV, Random, 95% CI)-2.04 [-3.22, -0.86]
10 Mental state: specific 3. Average score in negative symptoms (various scales)7 Mean Difference (IV, Random, 95% CI)Subtotals only
10.1 PANSS negative (high = bad)4508Mean Difference (IV, Random, 95% CI)-0.74 [-3.13, 1.64]
10.2 SANS (high = bad)3194Mean Difference (IV, Random, 95% CI)-3.95 [-15.25, 7.35]
11 Social functioning: 1a. Not employed2185Risk Ratio (M-H, Fixed, 95% CI)0.68 [0.57, 0.82]
12 Social functioning: 2a. Average scores (various scales)7 Mean Difference (IV, Fixed, 95% CI)Subtotals only
12.1 GAF (high = good)288Mean Difference (IV, Fixed, 95% CI)2.51 [-3.84, 8.85]
12.2 GAF-DIS (high = good)161Mean Difference (IV, Fixed, 95% CI)-6.10 [-12.34, 0.14]
12.3 ILSS (high = good)184Mean Difference (IV, Fixed, 95% CI)0.35 [-0.70, 1.40]
12.4 Time budget interview (high = good)1224Mean Difference (IV, Fixed, 95% CI)1.61 [-4.04, 7.25]
12.5 SOFAS (high = good)1236Mean Difference (IV, Fixed, 95% CI)1.21 [-2.71, 5.13]
12.6 SLOF (high = good)164Mean Difference (IV, Fixed, 95% CI)27.5 [17.03, 37.97]
12.7 SAS (high = good)180Mean Difference (IV, Fixed, 95% CI)-0.43 [-2.20, 1.34]
13 Social functioning: 2b. Average scores (various scales)6 Mean Difference (Random, 95% CI)Subtotals only
13.1 SFS (high = good)1 Mean Difference (Random, 95% CI)4.4 [-3.16, 11.96]
13.2 GAS (high = good)3 Mean Difference (Random, 95% CI)1.92 [-0.54, 4.38]
13.3 NOSIE (high = good)3 Mean Difference (Random, 95% CI)25.12 [5.75, 44.48]
14 Quality of life: Average scores (various scales)4 Mean Difference (IV, Fixed, 95% CI)Subtotals only
14.1 QLS (high = good)146Mean Difference (IV, Fixed, 95% CI)-5.05 [-16.12, 6.02]
14.2 LQLP (high = good)148Mean Difference (IV, Fixed, 95% CI)-0.40 [-1.30, 0.50]
14.3 EuroQoL (high = good)1206Mean Difference (IV, Fixed, 95% CI)-4.80 [-10.79, 1.19]
14.4 LQoLS (high = good)180Mean Difference (IV, Fixed, 95% CI)0.09 [-0.24, 0.42]
15 Satisfaction with care: 1a. Average endpoint scores (various scales)2 Mean Difference (Fixed, 95% CI)Subtotals only
15.1 CALPAS (high = good)1 Mean Difference (Fixed, 95% CI)0.3 [0.12, 0.48]
15.2 IAPSRS (high = good)1 Mean Difference (Fixed, 95% CI)0.1 [-0.11, 0.31]
16 Satisfaction with care: 1b. Average change scores (various scales)1 Mean Difference (Fixed, 95% CI)2.15 [-9.66, 13.96]
16.1 VSSS (high = good)1 Mean Difference (Fixed, 95% CI)2.15 [-9.66, 13.96]
17 Adverse events: 1. Lost to follow-up171495Risk Ratio (M-H, Fixed, 95% CI)0.96 [0.72, 1.28]
18 Adverse events: 2. Death4462Risk Ratio (M-H, Fixed, 95% CI)0.91 [0.29, 2.89]
19 Knowledge: 1a. Average scores (various scales)4 Mean Difference (IV, Random, 95% CI)Subtotals only
19.1 KASQ (patients, high = good)161Mean Difference (IV, Random, 95% CI)1.60 [-0.84, 4.04]
19.2 IS (patients, high = good)140Mean Difference (IV, Random, 95% CI)0.96 [-0.48, 2.40]
19.3 ITAQ (patients, high = good)2166Mean Difference (IV, Random, 95% CI)7.21 [0.94, 13.48]
20 Knowledge: 1b. Average scores (various scales)1 Mean Difference (IV, Random, 95% CI)Subtotals only
20.1 SUMD (patients, high = bad)181Mean Difference (IV, Random, 95% CI)-1.18 [-2.46, 0.10]
21 Knowledge: 1c. Average scores (various scales)2 Mean Difference (Fixed, 95% CI)Subtotals only
21.1 DAI (patients, high = good)1 Mean Difference (Fixed, 95% CI)1.4 [-0.97, 3.77]
21.2 KASI (relatives, high = good)1 Mean Difference (Fixed, 95% CI)2.83 [0.93, 4.73]
22 Burden of care: 1. Average scores (FBIS, high = bad)164Mean Difference (IV, Random, 95% CI)-2.40 [-7.10, 2.30]
23 Medication compliance: Not compliant to medication4374Risk Ratio (M-H, Fixed, 95% CI)0.57 [0.42, 0.77]
Analysis 1.1.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 1 Relapse: 1. Relapse.

Analysis 1.2.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 2 Relapse: 2. Time to first relapse.

Analysis 1.3.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 3 Service use: 1a. Re-hospitalisation.

Analysis 1.4.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 4 Service use: 1b. Re-hospitalisation (5-7 year follow-up).

Analysis 1.5.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 5 Service use: 2. Time to re-hospitalisation.

Analysis 1.6.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 6 Mental state: general 1a. Average score (BPRS, high = bad).

Analysis 1.7.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 7 Mental state: general 1b. Average score (PANSS, high = bad).

Analysis 1.8.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 8 Mental state: general 1c. Average score (KGV, high = bad).

Analysis 1.9.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 9 Mental state: specific 2. Average score in positive symptoms (various scales).

Analysis 1.10.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 10 Mental state: specific 3. Average score in negative symptoms (various scales).

Analysis 1.11.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 11 Social functioning: 1a. Not employed.

Analysis 1.12.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 12 Social functioning: 2a. Average scores (various scales).

Analysis 1.13.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 13 Social functioning: 2b. Average scores (various scales).

Analysis 1.14.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 14 Quality of life: Average scores (various scales).

Analysis 1.15.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 15 Satisfaction with care: 1a. Average endpoint scores (various scales).

Analysis 1.16.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 16 Satisfaction with care: 1b. Average change scores (various scales).

Analysis 1.17.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 17 Adverse events: 1. Lost to follow-up.

Analysis 1.18.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 18 Adverse events: 2. Death.

Analysis 1.19.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 19 Knowledge: 1a. Average scores (various scales).

Analysis 1.20.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 20 Knowledge: 1b. Average scores (various scales).

Analysis 1.21.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 21 Knowledge: 1c. Average scores (various scales).

Analysis 1.22.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 22 Burden of care: 1. Average scores (FBIS, high = bad).

Analysis 1.23.

Comparison 1 EARLY WARNING SIGNS TRAINING + TAU versus TREATMENT AS USUAL (TAU), Outcome 23 Medication compliance: Not compliant to medication.

Appendices

Appendix 1. Details of contact with authors

Details regarding intervention in the study were provided by Turkington 2006: confirmed that all participants received one hour at least on training in recognising early waring signs; Hogarty 1997a: suggested clarifying the process of recognising early waring sings of schizophrenia in the book ‘ A Guide to Individualised Treatment’ by Gerard E Hogarty; Vreeland 2006: wrote and explained that their Team Solutions study involved approximately  five to six hours of recognition of early warning signs. Two hours in Workbook No.1 and three to four hours in Workbook No.7. The manual is free and can be downloaded from www.partners4excellence.org ; Kopelowicz 1998a: confirmed that the first eight sessions of the Community Re-entry module included teaching patients early warning signs of psychotic relapse; Cunningham Owens 2001: confirmed that their study did not focus on early relapse symptom recognition but compliance; Hahlweg 1999: confirmed that the behavioural family intervention in their study had recognition of early warning signs as part of the psychoeducation and it lasted for about one session including the patient and family. This was assessed at the beginning of every family session over the course of about 20 sessions (one year).

Details regarding diagnoses were provided by Jumnoodoo 2008: confirmed that ICD-10 criteria were not used for diagnoses and less than 50% of patients had a diagnosis of schizophrenia, also, separate data for schizophrenia were not available to hand; Nordentoft 1998: OPUS STUDY: clarified that it was a first episode study, so none had a previous diagnosis in the schizophrenia spectrum.

Contributions of authors

IV rated all studies for inclusion after both sets of searches, contacted authors, extracted secondary data with MF, conducted analysis and wrote the review.

MF rated random studies for inclusion, extracted primary data with AJ and secondary data with IV and conducted analysis.

RM had the original idea for the review, devised methodology, checked random studies for inclusion and helped with the final version of the review.

CB and JMC were involved in the protocol development.

Declarations of interest

None known.

Sources of support

Internal sources

  • Paula Williamson, University of Liverpool Centre for Medical Statistics, UK.

External sources

  • Mersey Care NHS Trust, UK.

Differences between protocol and review

The protocol aimed to compare effectiveness of intermittent medication used on recognition of early warning signs in people not taking maintenance neuroleptic medication with treatment as usual (TAU) involving maintenance neuroleptic medication. We have not been able to do this because the studies that were identified used early warning signs intervention in both control and treatment groups and had to be excluded.

Cochrane methodology has been updated since initial publicationo of our protocol, we have updated our methods section of the protocol to reflect these advances.

For the study Garety 2008, which had multiple treatment arms, data were entered into the analysis for each pair-wise comparison separately, with shared intervention groups divided among the comparisons, according to the methods in section 16.5.4 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011); the total number of participants was divided and the means and standard deviations remained the same. This differs from the protocol which stated that for continuous data we would combine data from multiple treatment groups.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Anzai 2002

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 9 weeks.

Design: parallel.

Setting: secondary care.
Country: Tokyo.

Participants

Diagnosis: schizophrenia (ICD-10 and DSM-IV).

N = 32.
Age: average ˜47 years (SD 11).
Sex: not reported.
History: persistent and refractory symptoms of psychosis and poor insight into their illness. Mean duration of the current hospitalisation 4 ± 4.3 years, mean duration of illness 20.5 ± 11.6 years.

Interventions

1. Psychoeducation + TAU: CREP (Community Re-entry Program) includes identifying and keeping track of warning signs of relapse with a relapse prevention plan, group therapy, 18 one-hour sessions, twice a week (N = 16).

2. TAU + other: occupational rehabilitation activities (N = 16).

Outcomes

Social functioning: REHAB scale.

Unable to use:

Service use: re-hospitalisation (number not reported).
Mental state: PANSS (no mean and SD).

Knowledge: related to self management of illness, HOW (mean and SD not reported for the TAU group).

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"Randomised" no further details provided.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNo information provided.
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk"Before training started, two participants in the training group and one in the control group were discharged from the hospital. These patients were not included in calculations of the discharge rates". No further details provided.
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskSmall sample size. Protocol not available. Source of funding not reported.

Bauml 2007

Methods

Allocation: randomised.
Blinding: single-blind.
Duration: 2 years*.

Design: parallel.
Setting: secondary care.
Country: Germany.

Participants

Diagnosis: Schizophrenia (DSM-III-R and ICD-9).

N = 101.
Age: average ˜34 years (SD 11).
Sex: M 51, F 50.
History: indication for an antipsychotic relapse prevention for a period of at least 12 months.

Interventions

1. Psychoeducation + TAU: Patients had 4 weekly sessions of 60 minutes, then 4 more monthly sessions. Relatives were separately invited to 8 bi-weekly sessions of 90 to 120 minutes. Sessions were conducted by psychiatrists. Comprehensive information given about symptoms, aetiology, treatment, relapse prevention, etc. and individual crisis plans drawn up (N = 51).

2. TAU: maintenance therapy with antipsychotic medication, outpatient appointments (N = 50).

Outcomes

Service use: re-hospitalisation rate, (2-year follow-up reported in review, 7-year follow-up data available for a subgroup but not reported in review), time to re-hospitalisation**.
Mental state: BPRS.
Social functioning: GAF.

Quality of life: Lancashire quality of life profile (German).
Adverse events: death***.

Notes

Results based only on Technical University of Munich study centre.

*48 participants also followed up at 7 years.

**Time to relapse was imputed using method 7 from Tierney 2007 (report presents P value and events on each arm and randomisation ratio is 1:1).

***1 suicide at 12-month follow-up in the intervention group.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskBlock randomisation of 8 to 12 patients, "the randomisation list for each study centre was generated by a computerised random sampling in the study evaluation centre".
Allocation concealment (selection bias)Low risk"The study centres could learn to which treatment condition the group was assigned via telephone".
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low risk"The treating study psychiatrists were blind to the randomisation".
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNineteen patients excluded prior to index discharge and 34 patients dropped out during the 7-year follow-up. Missing patients were not included in the final analysis.
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available.

Bradley 2006

Methods

Allocation: randomised.
Blinding: single-blind.
Duration: 12 months.

Design: parallel.
Setting: secondary care.
Country: Australia.

Participants

Diagnosis: Schizophrenia (DSM-IV).

N = 59*.
Age: Average ˜34 years (SD 8).
Sex: M 15, F 35.
History: number of previous episodes: treatment group mean 0.6 (SD 1.5), control group 0.29 (SD 0.86).

Interventions

1. Psychoeducation + TAU: 26 sessions of multiple family group intervention for patients and families which involved recognising early warning signs, 26 sessions over 12 months (N = 25).

2. TAU: Regular appointments with case manager/doctor and individual psychosocial rehabilitation. Family contact (direct or on telephone) focused on psychoeducation, monitoring mental state and general support (N = 25).

Outcomes

Lost to follow-up.

Relapse**: rate of relapse, time to relapse.

Mental state: BPRS, SANS.

Social functioning: HONOS.

Quality of life: Quality of life scale.
Adverse events: death***.

Burden of care: FBIS (Family Burden Interview Schedule).

Unable to use -

Relapse**: time to relapse (hazard ratio not reported), duration (no mean and SD).

Notes

*59 consumer-caregiver pairs. 34 pairs were English speaking and 25 were Vietnamese speaking.

**Relapse is defined as the reemergence of frank psychotic symptoms after a period of remission of such symptoms, persisting continuously for a minimum of 7 days and requiring intensive community treatment or hospital admission. Time to relapse was imputed using method 4 from Tierney 2007 (report presents HR and events on each arm and randomisation ratio is 1:1).

***1 participant in the treatment group died from a heroin overdose.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"randomised" - participants divided into English speaking and Vietnamese speaking, randomly allocated within their group by drawing names form a canister.
Allocation concealment (selection bias)Unclear risk"A staff member drew names from a canister and without looking at the names, assigned them to experimental and control groups".
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low risk"Independent researchers, who were blind to study condition, conducted assessments".
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk"Nine pairs did not complete the data collection procedure after treatment or at 18 months (four in the control group and five in the treatment group". Missing patients were not included in the final analysis.
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available.

Buchkremer 1997

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 2 years*.

Design: parallel.
Setting: secondary care.
Country: Germany.

Participants

Diagnosis: schizophrenia (DSM-III-R).

N = 191.
Age: average ˜ 31 years (SD 7).
Sex: M 111, F 80.
History: mean number of hospitalisations 4.7 (SD 3.6); at least two acute psychotic episodes within the past 5 years; at least 4 weeks of psychopathological stabilization.

Interventions

1. Psychoeducational medication management (PMT) + TAU: all patients given detailed individualised information about schizophrenia and treatment; trained to recognize early symptoms, individual crisis schedules drawn up and medication management skills promoted. 10 sessions, first 5 at weekly intervals and 5 at fortnightly intervals (N = 32).

2. PMT + TAU + Key counselling (KC): Key-person counselling: relatives and carers were given information about schizophrenia and recognition of impending relapses, with coping strategies. 20 sessions (N = 35).

3. PMT + TAU + cognitive psychotherapy (CP): Cognitive psychotherapy: mediation of problem solving skills and improve coping strategies. 15 sessions, 7 sessions at weekly and 8 sessions at fortnightly intervals (N = 34).

4. PMT + TAU + CP + KP: (N = 33).

5. TAU: regular leisure time group activities. (N = 57).

Outcomes

Mental state: BPRS.

Service use: re-hospitalisation rate, time to re-hospitalisation**.
Social functioning: Global Assessment Scale (GAS).

Unable to use -

Mental state: SANS (no mean and SD).

Knowledge: Illness related attitudes - KK-Skala scale (no mean and SD).

Notes

Multi-centre study: 7 centres in Germany.

* 126 participants also followed up at 5 years.

**re-hospitalisation is reported as 36 hour full hospitalisation or 5-day partial hospitalisation.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"randomised" - initial homogeneity of the groups was ensured by applying a randomisation technique in which the prognostic factors of sex, Strauss-Carpenter prognosis score and medication compliance were balanced by preliminary matching; then randomly assigned to one of five groups with an allocation ratio of 1:1:1:1:2.
Allocation concealment (selection bias)Low risk"Randomisation was carried out by an independent institution".
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk

For the outcome re-hospitalisation there was blinding: "Where necessary, rehospitalisation data were obtained directly from the pertaining psychiatric institutions with the consent of the patient. As the project staff had no influence on the re-hospitalisation decision, this is an independent outcome criterion. Referral to hospital was accordingly 'blind' with respect to allocation to one treatment condition".

Other outcome measurements were not blind: "data were recorded by trained project staff who were not blind with respect to the group to which the patients were assigned".

Incomplete outcome data (attrition bias)
All outcomes
Low risk"In a modified intention-to-treat approach all patients who had attended at least one group session were included in the main analysis"; "44 patients who were re-hospitalised or dropped out for other reasons during the period between randomization and the start of group treatment (2 to 6 months) had to be excluded. However, these patients were followed up as completely as possible."
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available.

Chan 2007

MethodsAllocation: randomised.
Blinding: no details.
Duration: 12 months.
Design: parallel.
Setting: inpatient secondary care.
Country: Hong Kong.
Participants

Diagnosis: schizophrenia or schizo-affective disorder (DSM-IV).

N = 81.

Age: 18-63 years, mean: 35.82 years.

Sex: M 81.

History: not reported.

Interventions1. Psychoeducation + TAU: Transforming Relapse and Installing Prosperity (TRIP) programme to improve illness insight and health, ten 50-minute sessions over two weeks, includes relapse prevention plan development and symptom management (N = 44).
2. TAU + other: traditional ward occupational therapy of equivalent frequency and length, including a routine of work, rest and leisure activities (n = 37).
Outcomes

Service utilisation: Number of re-hospitalisations*.
Knowledge: Scale of Unawareness of Mental Disorder (SUMD)**.

Unable to use -

Short Form Health Survey (SF-36) questionnaire** (overall score not reported).

Time to relapse (unable to read data from curve).

Notes*Number of re-hospitalisations was classified as relapse in the study, but as service utilisation in our review.
**Translated to Chinese.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"Randomly assigned", no further details reported.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNo information provided.
Incomplete outcome data (attrition bias)
All outcomes
High riskNumber of patients randomised not reported. Data reported on treatment completers only.
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available. Source of funding not reported.

Chen 2003

Methods

Allocation: randomised.

Blinding: double-blind.
Duration: 10 weeks, follow-up 1 year.

Design: parallel.
Setting: secondary care.
Country: China.

Participants

Diagnosis: schizophrenia (ICD-10 and CCMD-3).

N = 64.
Age: 15-55 years, average ˜31 years (SD 9).
Sex: M 35, F 28.
History: average length of illness: 7.12 ± 5.45 years (N = 32, intervention group)vs. 8. 03 ± 5.95 years (N = 32, control group).

Interventions

1. Psychoeducation + TAU: teamwork education, medication management skills training module, symptom management skills training module. (N = 32).

2. TAU: antipsychotic therapy (N = 32).

Outcomes

Relapse: relapse rate.

Service use: re-hospitalisation rate (hospital stay ≥ 4 weeks).
Mental state: PANSS;
Social functioning: SDSS (Social dysfunction screening scale), NOSIE, not employed (no work after discharge ≥ half a year).

Adverse events: lost to follow-up.

Knowledge: ITAQ (Insight and treatment attitude questionnaire)

Not used in the review:

MRSS (Morningside Rehabilitation Status Scale)

NotesAssessment of curative effect: based on PANSS scale score.
Relapse: score of items P2, P3 , P6 , G5 or G9 ≥ 5, or two of which ≥ 4.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomised using draw lots.
Allocation concealment (selection bias)Unclear riskNot stated.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk"Double blind."
Incomplete outcome data (attrition bias)
All outcomes
Low riskOne person dropped out of the control group due to myocardial infarction. The reason for this is unlikely to be related to the true outcome.
Selective reporting (reporting bias)Low riskAll measured outcomes were reported.
Other biasLow riskNone obvious.

Chien 2004

Methods

Allocation: randomised.
Blinding: single-blind.
Duration: 18 months.

Design: parallel.
Setting: secondary care.
Country: Hong Kong.

Participants

Diagnosis: schizophrenia (DSM-IV).

N = 96*.
Age: average ˜42 years (SD 7).

Sex: M 128, F 72.

History: the relative with schizophrenia had no other mental illness, and the duration of the schizophrenia was 3 years or less at the time of recruitment.

Interventions

1. Psychoeducation + TAU: 12 bi-weekly 2-hour sessions over 6 months by trained psychiatric nurses to educate, improve illness management including recognising early warning signs and improve coping skills, multiple family group intervention (N = 33).

2. TAU: routine psychiatric outpatient and family support services (N = 31).
3. TAU + mutual support: peer-led group for information, support and coping skills for caregiving, 12 bi-weekly 2-hour sessions over 6 months, and did not include the patients (N = 32).

Outcomes

Lost to follow-up.
Mental state: BPRS (Positive symptoms).
Social functioning: SLOF (Specific level of functioning scale).

Burden of care: FBIS (Family Burden Interview Schedule).

Not used-

Service use: mean number of re-hospitalisations.

Notes

*96 families of out-patients with schizophrenia.

Data used only from interventions 1. and 2.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"Randomised" - computer-generated random numbers table.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskClinic staff were blinded, blinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low risk"An independent trained assessor (research assistant) undertook measurements at baseline (Time 1), 6 months (Time 2) and 18 months (Time 3), using a set of questionnaires. Both assessor and clinic staff were masked to treatment allocation".
Incomplete outcome data (attrition bias)
All outcomes
Low riskAnalysis of data was on an intention-to-treat basis.
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available.

Drury 1996

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 9 months*.

Design: parallel.
Setting: secondary care.
Country: UK.

Participants

Diagnosis: schizophrenia (ICD-9).

N = 62**.
Age: average ˜30 years (SD 9)
Sex: M 25, F 15.

History: mean number of previous episodes: treatment group 3.5 (SD 3.5), control group 3.0 (SD 2.7).

Interventions

1. Psychoeducation + TAU: four individual and group cognitive therapy, and family engagement. Cognitive therapy involved individual and group procedures including early warning signs recognition and strategies for coping (N = 30).

2. TAU: care of psychiatrist + recreation and informal support (N = 32).

Outcomes

Lost to follow-up*.

Mental state: PAS (Psychiatric assessment scale)*.

Unable to use:

Mental state: self report measure of delusional conviction (no mean and SD).

Notes

*Usable data reported after 12 weeks of treatment.

**Demographic data given only for participants completing the study.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"randomised" - no further details provided.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low risk"Prescribing clinicians were blind to group allocation", blinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk"All patients were rated by the first author and a random subset of patients were blindly rated by MB and JFM".
Incomplete outcome data (attrition bias)
All outcomes
High risk22 patients were excluded after randomisation and not included in the final analyses.
Selective reporting (reporting bias)High riskNot all expected outcomes were reported.
Other biasUnclear riskSmall sample size. Protocol not available. Source of funding not reported.

Garety 2008

Methods

Allocation: randomised.

Blinding: single blind.
Duration: 24 months.
Design: parallel.

Setting: secondary care.

Country: UK

Participants

Diagnosis: non-affective psychoses (ICD-10 or DSM-IV).

N = 301.

Age: 18-65 years.

Sex: M 211 F 90.

History: Patients with recent relapse, whether or not they had been admitted.

Interventions

There were two randomisation pathways:

Patients without carers:

1. CBT + TAU: Cognitive behavioural therapy focused on relapse prevention including pragmatic relapse prevention plan and identification of early signs, 12-20 sessions over nine weeks,  (N = 106)
2. TAU (N = 112)

Patients with carers:

1. CBT + TAU: Cognitive behavioural therapy focused on relapse prevention including pragmatic relapse prevention plan and identification of early signs, 12-20 sessions over nine weeks,  (N = 27)
2. TAU (N = 28)

3. Family Intervention + TAU: Family intervention focused on relapse prevention including how family members might understand warning signs, 12-20 sessions over nine weeks, (N = 28)

Outcomes

Rate of relapse*.
Mental state: PANSS.
Social functioning: Time-budget interview, SOFAS (Social and Occupational Assessment Scale)
Quality of Life: EuroQol
Adverse events: death, suicide attempts, violent incidence, lost to follow-up.

Not used-

Mental health: BDI (Beck Depression Inventory), BAI (Beck Anxiety Inventory).

Days on hospital and number of admissions.
Unable to use -

PSYRATS (Psychotic Symptom Rating Scale) (reported on two subscales).

Notes

*Measured in a subgroup of participants that relapsed after partial or full remission.

Protocol: ISRCTN83557988.

For the patients with carers pathway, data for the TAU group were split in two in the analyses: total number of participants were divided and the means and standard deviations remained the same.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"Randomisation schedules were independently generated by a trial randomisation service in a separate location from all trial centres using randomised permuted blocks"
Allocation concealment (selection bias)Low riskCentral randomisation "Randomisation schedules were independently generated by a trial randomisation service"
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low risk"Trial research assessors were independent of treatment delivery and every effort was made to ensure they were kept masked to allocation" "Patients were reminded by the assessors not to talk about treatment allocation". "The primary outcome variable, relapse, was assessed by masked panel evaluation"
Incomplete outcome data (attrition bias)
All outcomes
Low risk96% followed up for primary outcome data, secondary outcome data were available for 82% of the total sample at 12 months and for 80% at 24 months. No reasons for losses reported.
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasLow riskProtocol pre-published (ISRCTN83557988). Source of funding: Wellcome Trust.

Granholm 2005

Methods

Allocation: randomised.
Blinding: single blind.

Duration: 6 months.

Design: parallel.
Setting: secondary care.
Country: USA.

Participants

Diagnosis: schizophrenia (DSM-IV).

N = 76.
Age: average ˜54 years (SD 7).
Sex: M 56, F 20.

History: not reported.

Interventions

1. Psychoeducation + TAU: 24 weekly 2 hour group cognitive behavioural therapy sessions, including early detection and management of symptoms (N = 37).

2. TAU: ongoing care as usual (N = 39).

Outcomes

Lost to follow-up.

Unable to use -

Mental state:PANSS (Positive and negative syndrome scale); HDRS (Hamilton Depression Rating Scale); Beck Cognitive insight scale (mean and SD not reported).

Social functioning: Independent Living Skills Survey; UCSD performance based skills assessment; coping skills (mean and SD not reported).

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"randomised" - sequential list of random numbers.
Allocation concealment (selection bias)Unclear risk"The project coordinator assigned participants to treatments in the order that they consented, and the coordinator was the only staff person other than therapists with knowledge of group membership".
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low riskThe assessors were blinded to treatment group.
Incomplete outcome data (attrition bias)
All outcomes
Low risk"Intent-to-treat analyses were used to examine all outcome variables".
Selective reporting (reporting bias)High riskNot all expected outcomes were reported.
Other biasUnclear riskProtocol not available.

Gumley 2003

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 12 months.

Design: parallel.
Setting: secondary care.
Country: UK.

Participants

Diagnosis: schizophrenia (DSM-IV).

N = 144.
Age: average ˜36 years (SD 10).
Sex: M 66. F 134.

History: receiving antipsychotic medication, prone to relapse.

Interventions

1. Psychoeducation + TAU: clinical psychologist provided CBT, 5-session engagement phase and intensive targeted phase of 2-3 sessions per week at the appearance of relapse. Individualised case formulation of the cognitive behavioural factors associated with relapse, an idiosyncratic early signs questionnaire incorporating these factors was developed (N = 72).

2.TAU: Treatment overseen by consultant psychiatrist for ongoing medication, regular reviews, follow-up from a community mental health nurse (N = 72).

Outcomes

Lost to follow-up.

Relapse*: relapse number, time to relapse.

Service use: Re-hospitalisation rate.
Mental state: PANSS positive, PANSS negative.

Unable to use -

Service use: time to re-hospitalisation.

Mental state: Personal beliefs about Illness questionnaire (overall score not reported); Brief symptom inventory (only baseline data reported).

Social functioning: Social Functioning Scale (overall score not reported).

Notes*Increase in participant's self-reported early signs, failure to return questionnaire more than once, treating team reported symptom changes or circumstances/stressors suggestive of relapse. Time to relapse data was log transformed.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"Patient randomised according to predetermined envelopes containing the treatment group, devised by researcher, which was unbeknown to the assessors, therapist or participants".
Allocation concealment (selection bias)Unclear risk"A member of the research team opened an envelope that informed as to which group individual participants were to be allocated. Another member of the team witnessed this procedure, and the envelope was placed in the participant's case file".
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
High risk"Two research nurses, who were not blind to the treatment allocation of participants, conducted follow-up assessments".
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll data analysed on an intention-to-treat basis.
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available.

Harris 2009

Methods

Allocation: Cluster randomised, community health professional unit of randomisation.

Blinding: Not blinded.

Duration: 9 months.

Design: Parallel.
Setting: Mental health trusts.
Country: England.

Participants

Community mental health professionals (CMHPs) with caseloads of service users (SU) with a diagnosis of schizophrenia (DSM-IV).

CMHP
N = 28 pairs; service users N = 169.
Age: CMHP – Experimental : 39, Control: 40; SU -  Experimental : 44, Control: 41.4  

Sex: CMHP : 20 male, 36 female; SU:  96 male, 73 female.

History: SU length of illness – Experimental : 17.25 years, Control: 12 years  

Interventions

1. Experimental medication management training programme, including early warning signs work and relapse prevention, (N = 28 CMHP; N = 88 SU).

2. Waiting list (N = 28 CMHP; N = 81 SU).

Outcomes

Mental state: KGV (M) Version 6.

Social functioning scale: SFS.

Knowledge: Drug attitude inventory (DAI).

Satsifaction with care: The Californian Psychopharmacology Alliance Scale (CALPAS).

Unable to use-

Leaving the study early (number lost from each group not reported).

Not used-

Liverpool University Neuroleptic Side Effect Rating Scale (LUNSERS).

NotesResults adjusted for clustering: "Based on experience of similar studies with process outcomes a value of 0.05 was assumed for the ICC."
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"Service users were randomly selected from each CMHP’s sample frame to form their ‘study caseload’ before CMHPs were randomised to either the training programme or training waiting list." Does not report method of randomisation.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
High risk“There was no ‘blind’ assessment of service user level outcomes.”
Incomplete outcome data (attrition bias)
All outcomes
Low risk

6 CMHPs withdrew from the study, 3 that had been randomised to the intervention and 3 randomised to the control group. "34 [service users] withdrew from the study and 12 were lost to follow-up"; 16 from the intervention group and 30 from the control group.

Intention-to-treat analysis used.

Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasLow riskNone obvious.

Herz 2000

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 18 months.

Design: parallel.
Setting: secondary care.
Country: USA.

Participants

Diagnosis: schizophrenia (DSM-III-R).

N = 82.
Age: average ˜29 years (SD 9)
Sex: M 53, F 29.

History: increased risk for relapse (at least 1 hospitalisation in past 3 years or 2 or more lifetime hospital admissions).

Interventions

1. Psychoeducation + TAU: 5 components of PRP (Program for relapse prevention): education for patients and family about relapse and recognizing prodromal symptoms and behaviours, active monitoring of prodromal symptoms by team, immediate clinical intervention within 24 to 48 hours, supportive group/individual therapy improving coping skills, multifamily psychoeducation (N = 41).

2. TAU: individual supportive therapy and medication management biweekly for 15-30 min (N = 41).

Outcomes

Relapse*: relapse rate, time to relapse.

Service use: re-hospitalisation rate.

Unable to use -

Mental state: PANSS; Early Signs Questionnaire (no mean and SD).

Social functioning: GAS (no mean and SD).

Notes*defined as increase in any PANSS to moderately severe or higher and a GAS score of 30 or less. Time to relapse data was log transformed.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"randomised" - computer-generated cards.
Allocation concealment (selection bias)Low riskSealed envelopes.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low risk"Assessments were done by research interviewers who were master's-level mental health professionals, blinded as to patient group assignment, not associated with clinical care and instructed not to inquire about a patient's treatment during interviews".
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskThe study did not address this outcome.
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available.

Hogarty 1997a

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 3 years.

Design: parallel.
Setting: secondary care.
Country: USA.

Participants

Diagnosis: schizophrenia.

N = 97.
Age: average ˜28 years (SD 7).
Sex: M 56, F 41.

History: not reported.

Interventions

1. Personal therapy + TAU: taught awareness of one's own individual prodromes of psychosis (N = 23).

2. Family psychoeducation/management (family therapy) + TAU: included the three broad phases of joining, survival skills training and reintegration within the home, and reintegration into the community (N = 26).
3. Personal therapy + family therapy + TAU: (N = 26).
4. TAU: supportive therapy, active listening, correct empathy, reassurance (N = 24).

Treatment sessions were 30-45 min weekly for about 3 years. Personal and family therapy patients also received 1.9 additional monthly medication management sessions in years 1 and 2, and an additional 1.3 sessions each month in year 3.

Outcomes

Relapse*: rate of relapse.

Unable to use -

Time to relapse: defined as also including early treatment termination.

Service use: re-hospitalisation rate (only reported for patients who had relapse episodes).
Mental state: BPRS; Raskin negative symptom scale; Covi anxiety scale; Wing negative symptom scale (no mean and SD).
Social functioning: GAS (no data for TAU group).

Notes

*Defined as mild or less to greater than mild on two or more of the four psychotic symptoms on the BPRS or a 2 point increase on one or more severe symptoms + decrease of 10 points or more on GAS.

Data for intervention groups 1, 2 and 3 were combined.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"randomised" - no further details provided.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNo information provided.
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot intention-to-treat: "Since missing data were relatively few and not different among treatment conditions, only the data collected were analysed."
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available.

Horan 2009

Methods

Allocation: randomised.

Blinding: no details.

Duration: 6 weeks.

Design: parallel.

Setting: outpatient secondary care.

Country: USA.

Participants

Diagnosis: schizophrenia or schizoaffective disorder (DSM-IV).

N = 34.

Age: mean age social cognition: 50.7; mean age control (relapse prevention): 45.9 years.
Sex: social cognition: 87% M; control (relapse prevention): 100% M.

History: Clinically stable outpatients with no psychiatric hospitalisations in the past 6 months and the same antipsychotic medication for previous 3 months.

Interventions1. Social cognition: social skills training (N = 17)
2. Control*: illness self-management and relapse prevention skills including identifying and managing warning signs of relapse (n = 7).
Both programmes consisted of 12 one-hour groups that met twice weekly.
Outcomes

Loss to follow-up
Unable to use -
Mental state: BPRS (total scores not reported)

Satisfaction with treatment (not validated scale)

Not used -

Facial Emotion Identification Test, Half-profile of Nonverbal Sensitivity, Ambiguous Intentions Hostility Questionnaire, The  Awareness of Social Inference Test , MATRICS Consensus Cognitive Battery (MCCB): cognitive performance

Notes*EWS training was the control group, social skills the intervention.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"randomly assigned", no further information.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNo information provided.
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskThree participants out of 34 dropped out.
Selective reporting (reporting bias)High riskNot all expected outcomes were reported.
Other biasUnclear riskProtocol not available. Source of funding:  US Mental Illness Research, Eduation and Clinical Center.

Kopelowicz 1998a

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 1 month.

Design: parallel.
Setting: secondary care.
Country: USA.

Participants

Diagnosis: schizophrenia (DSM-IV).

N = 70*.
Age: average ˜35 years (SD 11).
Sex: M 42, F 17.
History: mean duration of illness 12 years (SD 9).

Interventions

1. Psychoeducation + TAU: (CREP) community re-entry programme: eight sessions each 45 minutes, in groups, twice a day, four days a week schedule, over 2 weeks.; including teaching patients early warning signs of relapse (N = 28).

2. TAU+ other: occupational therapy sessions conducted by certified therapists (N = 31).

Outcomes

Lost to follow-up.

Unable to use -

Knowledge and performance of skills (no mean and SD).

Notes*Data were reported on 59 subjects that completed participation, defined as attending at least four sessions.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"Randomised" -no further details provided.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNo information provided.
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskData reported on treatment completers only.
Selective reporting (reporting bias)High riskNot all expected outcomes were reported.
Other biasUnclear riskProtocol not available. Source of funding not reported.

Kopelowicz 2003

Methods

Allocation: randomised.
Blinding: single-blind.
Duration: 9 months.

Design: parallel.
Setting: secondary care.
Country: USA.

Participants

Diagnosis: schizophrenia (DSM-IV).

N = 92.
Age: average ˜38 years (SD 11).
Sex: M 62, F 30.

History: not reported.

Interventions

1. Psychoeducation + TAU: skills training groups met for 90-minute sessions four times per week for 3 months. Skills training for patients and relatives included medication management and symptom management (identification and intervention of early warning signs) (N = 45).

2. TAU: monthly psychiatric visits for medication management and case management by social workers (N = 47).

Outcomes

Lost to follow-up.

Service use: re-hospitalisation rate.
Mental state: PANSS.
Social functioning: ILSS (Independent Living Skills Survey).

Medication compliance.

Unable to use -

Quality of life: Quality of life interview (no data).

Knowledge: Skill acquisition and generalisation (no overall scores).

Burden of care: FBIS (no SD).

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"randomised" -no further details.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low risk"All of the raters were blind to the treatment condition of the patients and relatives", "to ensure blindness the raters were instructed to remind patients and relative subjects at each visit to not disclose what kinds if treatment they were receiving".
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk"Baseline comparisons were conducted with all 92 subjects who were randomised, but all subsequent analyses included only those subjects who were assessed at all three time points".
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available.

Kuipers 1997

Methods

Allocation: randomised.
Blinding: not blind.
Duration: 9 months.

Design: parallel.
Setting: secondary care.
Country: UK.

Participants

Diagnosis: schizophrenia (DSM-IV).

N = 60.
Age: average ˜40 years.
Sex: M 38, F 22.
History: at least one current positive psychotic symptom that was distressing, unremitting, and medication resistant.

Interventions

1. Psychoeducation + TAU: 9 months of individual CBT weekly or fortnightly, individual CBT included discussion of relapse signatures, trigger identification and seeking help (N = 28).

2. TAU: case management and medication, including involvement of a key worker (N = 32).

Outcomes

Lost to follow-up.

Mental state: BPRS.

Economic burden (cost of care).

Unable to use -

Mental state: Insight; BDI; BAI; Beck Hopelessness Scale; NART; Quick test (no data).

Social functioning: SFS (Social Functioning Scale) (no data).

Satisfaction of care: Satisfaction with therapy questionnaire (results not reported for TAU group).

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomised permuted blocking, block size of six.
Allocation concealment (selection bias)Low riskCarried out by the trial statistician.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk"In a trial of psychological treatment it is extremely difficult to make assessments that are totally blind to the treatment condition and this was not attempted. However, all assessments were carried out by independent research workers who were not involved in the treatments".
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk'Decision to treat' analysis using the standard 'carry forward' method to impute missing values reported.
Selective reporting (reporting bias)High riskNot all expected outcomes were reported.
Other biasHigh riskProtocol not available. Charitable donation from Janssen Pharmaceutica.

Li 2003

Methods

Allocation: cluster-randomised (ward unit of randomisation).
Blinding: not reported.
Duration: 9 months.

Design: parallel.
Setting: secondary care.
Country: Hong Kong.

Participants

Diagnosis: schizophrenia (CCMD-II-R).

N = 101.
Age: experimental group 16-61 years, mean age 32.9 (SD 9.7); control group 15-49 years, mean 32.2 (SD 8.9)
Sex: M 43, F 58 .
History: number of hospitalisations ranged from first to sixth time, duration of illness ranged from 0 to 21 years.

Interventions

1. Comprehensive patient/family education guide (CP/FEG): Nurse initiated patient/family education strategy. 8 hours with patients, 36 hours with families in hospital, then 2 hours per month for 3 months. Included early warning signs of relapse and responding to early warning signs training for families (N = 46, 4 wards).

2. TAU (N = 55, 4 wards).

Outcomes

Relapse rate.

Re-hospitalisation rate.

Pychotic symptoms of patient: Chinese version of the BPRS (Zhang 1998).

Social functioning: Chinese version of the NOSIE (measured at discharge) and GAS (Zhang 1998).

Knowledge (of relative): Chinese version of Knowledge about Schizophrenia Interview (KASI) (Tsang, Chan & Tam 1999).

Burden of care: FAS.

Medication compliance.

Notes

Results not adjusted for randomisation. ICC of 0.1 used to impute data.

Relapse defined as re-hospitalisation, recurrence of schizophrenic symptoms or a marked increase in the number or intensity of symptoms measured by a BPRS score of 5 or above on the relapse scale indicating exacerbation of the illness.

EWS part of training only delivered to family.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"Random assignment", no further details reported.
Allocation concealment (selection bias)Unclear riskNo details reported.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNo details reported.
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskThere were 10 dropouts in the experimental group and 22 dropouts in the control group.
Selective reporting (reporting bias)Unclear riskNumber of participants for the outcome burden of care (FAS) not reported.
Other biasUnclear riskNone obvious.

Liberman 1998

Methods

Allocation: randomised.
Blinding: single-blind.
Duration: 2 years.

Design: parallel.
Setting: secondary care.
Country: USA.

Participants

Diagnosis: schizophrenia.

N = 84.
Age: 37.1 years (SD 8.8).
Sex: M 84.
History: outpatients with persistent and unremitting forms of schizophrenia.

Interventions

1. Social skills training: UCLA Social and Independent Living Skills Program consisting of basic conversation, recreation for leisure, medication management, and symptom management, including Identifying warning signs of relapse and developing a relapse prevention plan, over 6 months, 3 hours per day, 4 days a week (N = 42)*.

2. Psychosocial occupational therapy: expressive, artistic, and recreational activities that mediated supportive therapy in groups and individually, over 6 months, 3 hours per day, 4 days a week (N = 42)*.

Outcomes

Mental state: BPRS**

Social functioning: Independent Living Skills Survey (ILSS)**

Social functioning: Global Assessment Scale (GAS)**

Adverse events: Lost to follow-up*

Quality of life: Lehman Quality of Life Scale (LQoLS)**

Not used in the review-

Profile of Adaptation to Life subscales

Rosenberg Self-Esteem Scale

Brief Symptom Inventory (BSI)

Social Activities Scale (SAS) (not a validated scale)

Notes

*Total included participants for each group was not clearly reported "assigned half to psychosocial occupational therapy and half to skills training", assumed this to be 42 in each treatment group.

**Data reported for 80 participants at 2 years follow-up, number in each group not reported, assumed to be 40 in each group.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskRandomised "Cohorts of 10–12 outpatients with persistent and unremitting forms of schizophrenia were entered into the study through a randomization procedure".
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low riskBlinding of outcome assessors "psychiatrists were blind to the psychosocial treatment assignment", "Blindness of the assessors was promoted by geographically distancing the treatment program from the locales for pharmacotherapy and ratings of outcome and by repeatedly instructing the subjects to refrain from mentioning their psychosocial treatment".
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskTotal included participants for each group was not clearly reported. "Attrition was minimal; only 14 of 84 subjects dropped out during the 2-year period (four in the control group and 10 in the social skills training group; x2 = 2.44, df = 1, P = 0.12). Almost all of the dropouts were accounted for by individuals moving to new geographical areas of the state or country and becoming unavailable for assessments."
Selective reporting (reporting bias)High riskNot all expected outcomes were reported.
Other biasLow riskFunded by Department of Veterans Affairs and NIMH.

McDonell 2006

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 3 years.

Design: parallel.
Setting: secondary care.
Country: USA.

Participants

Diagnosis: schizophrenia, schizoaffective disorder, other psychotic disorders (DSM-IV).

N = 97*.
Age: average ˜32 years (SD 9).

Sex: M 74, F 21.

History: mean illness duration ˜10 years (SD 8.5).

Interventions

1. Psychoeducation + TAU: multiple-family group treatment delivered by two clinicians to groups of 5-8 families over 2 years, including early warning signs work and relapse prevention, 24 sessions in year 1 and 12 sessions in year 2 (N = 53).

2. TAU: outpatient services, medication management and case management (N = 44).

Outcomes

Service use: re-hospitalisation rate*.

Unable to use -

Mental state: BPRS; MSANS (Modified scale for the assessment of negative symptoms) (only baseline data reported).

Burden of care: care givers' social support (no mean and SD).

Economic outcomes: Inpatient and outpatient service utilisation (no SD).

Notes

*Demographic data not reported for 2 participants.

**Community hospitalisation when imminent danger to oneself or others, grave disability, or medically necessary. State hospitalisation when need for long-term care not possible at a community hospital or direct referral from the justice system.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomisation of participants conducted by (1) separating participants on the basis of 'atypical v conventional medication at status; (2) assigning a number and (3) pulling a piece of paper with the number from a hat giving a 50% chance of being in either group.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNo information provided.
Incomplete outcome data (attrition bias)
All outcomes
Low riskData analysed on an intention-to-treat basis.
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available.

Merinder 1999

Methods

Allocation: randomised.
Blinding: single blind.
Duration: 1 year.

Design: parallel.
Setting: secondary care.
Country: Denmark.

Participants

Diagnosis: schizophrenia (ICD-10, OPCRIT).

N=46.
Age: average ˜36 years (SD 35).
Sex: M 24, F 22.

History: 8.2 years with a median of five earlier admissions.

Interventions

1. Psychoeducation + TAU: 8 session psychoeducation programme for patients and relatives including sessions called "Stress and early signs of relapse, emergency plan" and "What can you and your family do about it?", 8 weekly sessions separately for patients and relatives (N = 23).

2. TAU+ other: pharmacological intervention, psychosocial rehabilitation, supportive psychotherapy (N = 23).

Outcomes

Lost to follow-up.

Rate of relapse, time to relapse*.

Rate of re-hospitalisations**

Mental state: BPRS.
Social functioning: GAF (Global Assessment of Function).

Satisfaction with care: VSSS (Verona Service Satisfaction scale).

Knowledge: Insight scale***.

Unable to use -

Relapse*: time to (no hazard ratio).

Knowledge of schizophrenia scale (not validated scale).

GCSI (Global Clinical Assessment of Severity of Side Effects; results not reported for separately for treatment groups).

Notes

*3 levels rated: Type 1- aggravation of symptoms without change of treatment; Type 2- aggravation of symptoms with change in medication, Type 3- aggravation of symptoms with admission to hospital. Time to relapse was imputed using method 7 from Tierney 2007 (report presents P value and events on each arm and randomisation ratio is 1:1).

**Re-hospitalisation rate was added to the analysis as the number of participants with Type 3 relapse.
***Translated to Danish.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskPatients were block randomised, stratified for gender and for illness duration.
Allocation concealment (selection bias)Unclear riskRandomisation by an independent institution, no further details provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low risk"The ratings of psychopathology and psychosocial function were performed by researchers not involved in the intervention and not informed of the treatment allocation".
Incomplete outcome data (attrition bias)
All outcomes
Low riskData analyses on an intention-to-treat basis.
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasHigh riskProtocol not available. Part funded by Lunbeck A/S, Denmark.

Norman 2002

Methods

Allocation: randomised.
Blinding: single-blind.
Duration: 1 year.

Design: parallel.
Setting: secondary care.
Country: Canada.

Participants

Diagnosis: schizophrenia (DSM-III-R).

N = 130*.
Age: Average ˜34 years (SD 9).
Sex: M 81, F 40.

History: clinically stable.

Interventions

1. Psychoeducation + TAU: 12 weeks of group sessions+ 12 individual sessions emphasizing on three aspects of stress management. Stress management program included 12 aspects, one of which included recognition and reporting any increase in symptoms and learning to recognise and avoid situations that trigger symptoms. (N = 64).

2. TAU: social activities programme, including mutual support and recreational activities (N = 66).

Outcomes

Lost to follow-up.

Service use: Re-hospitalisation rate.

Unable to use -

Mental state: SAPS; SANS (no SD).

Social functioning: Life Skills Profile (no SD).

Notes* Demogaphic data reported for participants completing the study.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskParticipants were randomly assigned within matching by gender and number of previous hospitalisations, no further details provided.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low riskData was collected by psychiatrists who "were not responsible for the clinical care of the relevant clients and were kept blind as to the assignment of clients to the stress management or social activities conditions".
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk"Nine clients were lost to follow-up (five did not complete the intervention and four were unavailable for 1 year post-test assessments), leaving a net sample of 121 clients, 60 of whom were assigned to the stress management condition and 61 to the social activities. Because of sporadic instances of missing data, analyses are based on sample sizes varying between 112 and 121".
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available.

Shin 2002

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 10 weeks.

Design: parallel.
Setting: secondary care.
Country: USA.

Participants

Diagnosis: schizophrenia (DSM-IV).

N = 48.
Age: Average ˜38 years (SD 8).
Sex: M 20, F 28.
History: not reported.

Interventions

1. Psychoeducation + TAU: 90 minutes long, 10 weekly group therapy. Culturally sensitive psychoeducational group program (including early warning signs recognition and relapse prevention) and individual supportive therapy and family work. (N = 24).

2. TAU: individual supportive therapy (N = 24).

Outcomes

Lost to follow-up.

Mental state: BPRS.

NotesAll sessions and outcome measures were conducted in Korean.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"randomised" - no further details provided.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk"The BPRS was completed before and after the intervention in Korean by a bilingual Korean psychiatrist, who was blinded to participants' treatment modality [...] the other scales were verbally presented to the study participants by the first author in Korean and self-report answers were written down by the participants".
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll patients accounted for.
Selective reporting (reporting bias)High riskNot all expected outcomes were reported.
Other biasUnclear riskProtocol not available. Source of funding not reported.

Shon 2002

Methods

Allocation: randomised.
Blinding: unclear.
Duration: unclear, 12 sessions.

Design: parallel.
Setting: secondary care.
Country: Korea.

Participants

Diagnosis: schizophrenia, delusional disorders, mood disorders (DSM-IV).

N = 40*.
Age: average ˜32 years (SD 8).
Sex: M 22, F 16.
History: not reported.

Interventions

1. Psychoeducation + TAU: medication compliance and symptom management program consisting of 12 sessions, each lasting 70 minutes and included developing skills to recognize symptoms and effective methods of preventing relapse (N = 18).

2. TAU: standard psychiatric treatment post discharge (N = 20).

Outcomes

Lost to follow-up.

Unable to use -

Mental state: Relapse symptom measurement (not validated scale).

Medication compliance: Medication use scale; Scale for family support in medication compliance (not validated scales).

Notes*Two participants were excluded from the experimental group during the randomisation process.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"Randomised" - no further details provided.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNo information provided.
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskThe study does not address this outcome.
Selective reporting (reporting bias)High riskNot all expected outcomes were reported.
Other biasUnclear riskProtocol not available.

Tait 2002

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 10 months*.

Design: parallel.
Setting: secondary care.
Country: UK.

Participants

Diagnosis: schizophrenia (DSM-IV).

N = 20.
Age: average ˜39 years.
Sex: M 14, F 6.
History:

Interventions

1. Psychoeducation + TAU: initial engagement and formulation for 5 one-hour sessions, early signs monitoring fortnightly for 10 months, targeted cognitive therapy intensively for a brief period, about 2-3 sessions per week (N = 9).

2.TAU: routine appointments with psychiatrist and key worker (CPN, social worker or OT) (N = 11).

Outcomes

Relapse**: relapse rate.

Unable to use -

Mental state: Early signs monitoring; PANSS (no data).

Notes

*Average follow-up 10 months (range from 5-13 months).

**Relapse defined as i) significant increase in early signs; ii) presence of relevant psychosocial factors associated with previous relapses; iii) non return of early sings monitoring questionnaires.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"Randomised" - no further details.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNo information provided.
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk"The mean postal response rate of ESM questionnaires [...] is 83%". No further details provided.
Selective reporting (reporting bias)High riskNot all expected outcomes were reported.
Other biasUnclear riskSmall sample size. Protocol not available.

Turkington 2006

Methods

Allocation: randomised.
Blinding: single-blind.
Duration: 1 year.

Design: parallel.
Setting: secondary care.
Country: UK.

Participants

Diagnosis: schizophrenia (ICD-10).

N = 422.
Age: Average ˜40 years.
Sex: M 325, F 97.
History: ongoing positive and/or negative symptoms or at risk of relapse.

Interventions

1. Psychoeducation + TAU: community psychiatric nurse delivered 6 hour-long sessions over 2-3 months, carers offered 3 sessions. CBT including symptom management and relapse prevention (N = 257).

2. TAU: care of community mental health teams (N = 165).

Outcomes

Time to first relapse*.

Unable to use -

Mental state: Comprehensive psychopathological rating scale; Schizophrenia change scale, Montgomery and Asberg rating scale (no SD).
Insight: Insight rating scale (no SD).

Burden of care: Burden of care questionnaire (no SD).

Notes*Time to relapse was imputed using method 3 from Tierney 2007 (report presents HR and CIs).
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomisation conducted by computer-generated blocks of 6 random numbers and stratified by site.
Allocation concealment (selection bias)Low riskResults of the randomisation placed in sealed envelopes and only opened at the time of treatment allocation.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low riskRaters were independent. Patients were "asked not to reveal or even hint at their treatment allocation to the raters. Raters were informed that some study material would be left with TAU patients to help preserve the blindness of the ratings".
Incomplete outcome data (attrition bias)
All outcomes
Low riskData analysed on an intention-to-treat basis.
Selective reporting (reporting bias)High riskNot all expected outcomes reported.
Other biasHigh riskProtocol not available. Trial funded by a research grant from Pfizer.

Van Meijel 2006

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 1 year.

Design: parallel.
Setting: secondary care.
Country: Netherlands.

Participants

Diagnosis: schizophrenia (DSM-IV).

N = 95.
Age: average ˜35 years.
Sex: M 65, F 30**.
History: mean number of previous episodes: treatment group 4.4, control group 3.8.

Interventions

1. Psychoeducation + TAU: Relapse prevention plan with patient and carers, early signs are systematically inventoried and recognition assisted, followed by an action plan (pure early warning signs work). 4 phases: preparatory phase, listing of early warning signs, monitoring of the early signs, preparation of an action plan (N = 51).

2. TAU: care as usual (N = 44).

Outcomes

Relapse*: relapse rate, time to relapse.

Unable to use -
Mental state: PANSS (no SD); CGI (no data).
Knowledge: Insight scale (no mean and SD).

Satisfaction with care: Working Alliance Inventory (no mean and SD).

Notes

*Relapse defined as a significant increase in i) delusions, ii) hallucinations, iii) disorganisation of thinking, iv) chaotic or aggressive behaviour; 6 or higher on CGI; symptoms had to be present for at least 7 days. Time to relapse was imputed using method 7 from Tierney 2007 (report presents P value and events on each arm and randomisation ratio is 1:1).

**Demographic data only reported for participants completing the study.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskRandomisation of nurses (therapists) and patients, nurses were divided at random per department between experimental and treatment conditions. A researcher determined at random the order in which the patients would be approached for participation.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNo information provided.
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk"11 patients dropped out from the experimental group and 2 from the control group. The reasons for dropping out of the experimental group were: premature discharge (n = 1); stress caused by preparing the relapse prevention plan (n = 2); psychotic relapse (n = 1); lack of motivation to prepare the relapse prevention plan (n = 5); and lack of time of the nurses (n = 2). The reason for dropping out of the control group was completion of the treatment (n = 2).
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasHigh riskProtocol not available. Financial support in part from Eli Lilly The Netherlands.

Vreeland 2006

Methods

Allocation: randomised.
Blinding: single-blind.
Duration: 24 weeks.

Design: parallel.
Setting: secondary care.
Country: USA.

Participants

Diagnosis: schizophrenia (DSM-IV).

N = 74.
Age: average not reported, range 35 to 64 years.
Sex: M 33, F 41.
History: not reported.

Interventions

1. Psychoeducation + TAU: (Team Solutions Program): groups met twice a day, two days per week, for 24 weeks including education and management of illness, one session covering preventing relapse and crisis resolution (N = 40).

2. TAU + other: Day treatment: prevocational work areas, recreational groups, social skills training, psychoeducational groups (N = 34).

Outcomes

Mental state: PANSS positive and PANSS negative.
Social functioning: GAF-DIS (Global Assessment of Functioning-Disability Scale).

Satisfaction with care: IAPSRS (International association of psychosocial rehabilitation services).

Knowledge: KASQ.

Unable to use -

Mental state: CGI (no overall scores).

Quality of life: PGWB (Psychological general well-being scale) (no overall scores).

Knowledge: TSCKAS (not validated scale).

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomisation of participants based on a table of random numbers.
Allocation concealment (selection bias)Unclear riskNo details.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low risk"Trained raters who were blind to the study group assignment conducted interviews with participants to complete outcome measures at baseline, eight weeks and 24 weeks".
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskThree participants were not included in the final analysis "because of missing data for the primary outcomes at time 1".
Selective reporting (reporting bias)High riskNot all expected outcomes were reported.
Other biasHigh riskProtocol not available. Funded in part by Eli Lily and Company.

Weng 2005

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 2 months.

Design: parallel.
Setting: secondary care.
Country: China.

Participants

Diagnosis: schizophrenia (DSM-III-R).

N = 124.
Age: not reported.
Sex: not reported.
History: not reported.

Interventions

1. Psychoeducation + TAU: Multimodal rehabilitation program including medication management and symptom management delivered in groups of 8-10 patients over 2 months (N = 62).

2. TAU: standard inpatient services (N = 62).

Outcomes

Relapse: number.

Service use: re-hospitalisation number.

Social functioning: employment.
Unable to use -

Mental state: BPRS (mean change from discharge to follow-up).
Social functioning: SDSS (Social disability screening scale); NOSIE-30 (Nurses' observation scale for inpatient evaluation) (mean change from discharge to follow-up).

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"Randomly assigned" - no further details provided.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk"Double-blind assessments", no further details provided.
Incomplete outcome data (attrition bias)
All outcomes
Low risk"61 patients in each group were successfully followed up one year after discharge", no further details about participants lost to follow-up.
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskNone obvious.

Xiang 2001

Methods

Allocation: randomised.
Duration: 20 weeks, follow-up 1 year.

Design: parallel.
Setting: secondary care.
Country: China.

Participants

Diagnosis: schizophrenia (CCMD-2-R/ ICD-10).

N = 133.
Age: average ˜38 years (SD 21).

Sex: male and female.
History: patients in remission, average length of illness: median˜10.5 years

Interventions

1. Psychoeducation + TAU: With reference to "Social and independent living skills" by Liberman: 1) medication self-management, including information on antipsychotic medication, how to self-manage medication, assess the effect of medication, recognise adverse effect of medication, communicate the effect and adverse effect of medication to medical staff. 2) early warning sign training, including training on how to recognise and monitor signs of relapse, deal with symptoms of relapse, resist intake of alcohol and narcotics. Training was provided by a psychiatrist and a psychologist. Frequency 90˜120 minutes per session, 3 sessions a week for 20 weeks. (N = 66).

2. TAU: standard psychological intervention (N = 67).

Outcomes

Relapse: Relapse rate, time to relapse*.

Service use: re-hospitalisation rate, time to re-hospitalisation*.
Mental state: BPRS.

Adverse events: leaving the study early.

Unable to use-

Number of missed doses (skewed data).

Score of patient compliance with medication (not a validated scale).

Notes

In Chinese.

*Time to relapse and time to re-hospitalisation were imputed using method 7 from Tierney 2007 (report presents P value and events on each arm and randomisation ratio is 1:1).

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomised using random number table.
Allocation concealment (selection bias)Unclear riskNot stated.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low riskAssessor blind.
Incomplete outcome data (attrition bias)
All outcomes
Low risk2 people in the intervention group left the study early and 3 in the control group. Incomplete outcome data was excluded from the final analysis.
Selective reporting (reporting bias)Unclear riskInsufficient information to permit judgement.
Other biasLow riskNone obvious.

Xiong 1994

Methods

Allocation: randomised.
Blinding: unclear.
Duration: 18 months.

Design: parallel.
Setting: secondary care.
Country: China.

Participants

Diagnosis: schizophrenia (DSM III R).

N = 63*.
Age: average ˜ 31 years (SD 10).
Sex: M 43, F 20.
History: mean duration of illness 7.5 years (SD 6.8); mean number of hospitalisations 4 (SD 5.1).

Interventions

1. Psychoeducation + TAU: Family-based intervention, included discussion about detecting relapse. Introduction: two to three 45-minute sessions- family +/- patient; then monthly 45-minute counselling with patient (average 8.1 contacts per year) and family, and 90-minute family group; finally maintenance: families seen once in 2-3 months and monthly group meetings up to 18 months (N = 34).

2. TAU: prescription for antipsychotic medication given at discharge, sporadic renewals (N = 29).

Outcomes

Relapse**: relapse rate.

Service use: re-hospitalisation rate.
Adverse events: deaths***.

Employment: mean number of months of employment.

Unable to use -

Mental state: SAPS; SANS; BPRS (no mean and SD).

Social functioning: SDSS; GAF (no mean and SD).

Burden of care: overall burden on a 4 point scale (not validated scale).

Notes

*The number of patients in the groups at different time periods varies because the length of follow-up after hospitalisation differed.

**Relapse: hospitalisation or serious suicide attempt or deterioration on SAPS as marked or severe, symptoms on SANS marked or severe, BPRS items 4,7,11,12,and 15 rated severe; overall GAF score 40 or less; greater than 80% disability on SDSS.

***Suicides treated as relapses in the data, not entered as relapse on our analysis.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"Randomly assigned" - no further details.
Allocation concealment (selection bias)Unclear riskNo information provided.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk"As in all family intervention studies, there was some difficulty in maintaining blinding [...] this problem by instructing families not to state who had provided treatment and by ensuring that the evaluators were unaware of the purpose of the study".
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskOne patient from the experimental group and one patient in the control group committed suicide. One patient form the experimental group died from complications of inadequately treated diabetes. "These deaths were treated as relapses in the analysis". "Non-compliance does not confound the results because all subjects are included in the analysis".
Selective reporting (reporting bias)Low riskAll stated outcomes were reported.
Other biasUnclear riskProtocol not available.

Zhan 2003

Methods

Allocation: randomised.
Duration: 1 year.

Design: parallel.
Setting: secondary care.
Country: China.

Participants

Diagnosis: first episode or relapsing schizophrenia (CCMD-2-R).

N = 146.

Age: 37. 87 ± 10.12 years for intervention group and 36. 59 ± 11.12 years for control group.
History: average length of illness: 3. 45 ± 1.58 years for intervention group and 3. 57 ±1. 78 years for control group.

Interventions

1. Psychoeducation + TAU: pharmacotherapy, with reference to "social and independent living skills" by Liberman, intervention group received the following: 1) medication self-management, including information on antipsychotic medication, how to self-manage medication, assess the effect of medication, recognise adverse effect of medication, communicate the effect and adverse effect of medication to medical staff. 2) symptom self-monitoring, including training on how to recognise and monitor signs of relapse, learn to communicate with medical staff and prevent relapse, deal with persisting symptoms and develop good living habits. Training was provided by psychiatrists and psychologists. Frequency: 60˜90 minutes per session, 1 session a week for 12 weeks. (N = 74).

2. TAU: pharmacotherapy + standard psychological intervention (N = 72).

Outcomes

Relapse: relapse rate (occurrence of overt psychiatric symptoms, BPRS score ≥28).
Mental state: BPRS.

Social functioning: NOSIE.

Medication compliance.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomised by a tossing a coin.
Allocation concealment (selection bias)Unclear riskNot stated.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low riskAssessor blind.
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo incomplete outcome data.
Selective reporting (reporting bias)Low riskAll measured outcomes were reported.
Other biasLow riskNone obvious.

Zhang 2004

  1. a

    BAI: Beck Anxiety Inventory
    BDI: Beck Depression Inventory
    BPRS: Brief Psychopathology Rating Scale
    BSI: Brief Symptom Inventory
    CBT: cognitive behavioural therapy
    CG-I: Clinical Global Impression
    DSM: Diagnostic and Statistical Manual
    EWS: early warning signs
    FBIS: Family Burden Interview Schedule
    GAF: Global Assessment of Functioning
    GAF-DIS: Global Assessment of Functioning
    GAS: S Global Assessment Scale
    Hamilton Depression Rating Scale
    HoNOS: Health of the Nation Outcome Scale
    ICD: International Classification of Diseases
    ILSS: Independent Living Skills Survey
    ITAQ: Insight and Treatment Attitude Questionnaire
    LQoLS: Lehman Quality of Life Scale
    NOSIE: Nurses' Observation Scale for Inpatient Evaluation
    PANSS: Positive and Negative Syndromes Scale
    PAS: Psychiatric Assessment Scale
    PSYRATS: Psychotic Symptom Rating Scale
    REHAB: Rehabilitation Evaluation Hall and Baker
    SANS: Scale for the Assessment of Negative Symptoms
    SAS: Social Activities Scale
    SLOF: Specific Level of Functioning assessment scale
    SOFAS: Social and Occupational Functioning Assessment Scale
    SUMD: Scale of Unawareness of Mental Disorder
    TAU: treatment as usual

MethodsAllocation: randomisation.
Duration: 1 year.
Setting: secondary care.
Country: China.
Participants

Diagnosis: schizophrenia (CCMD-III) and their families.

N = 120.
Age: 26 ± 8 years for intervention group and 28 ± 7 years for control group.
Sex: male and female.
History: not reported.

Interventions

1. Psychoeducation + TAU: With reference to "social and independent living skills" by Liberman, intervention group received the following: 1) independent community living skill , 2) medication self-management, 3) symptom self-monitoring skills, including recognise early warning signs of relapse, symptom coping strategy, apply these strategy in a community setting. 16 sessions of group work by health professional, social and independent living skills training with elements of early warning signs training. Class learning (90˜120 minutes per session, 2 sessions a week for 8 weeks) Plus community follow-up and home-based intervention. (N = 60).

2. TAU: Conventional rehabilitation service (outpatient follow-up every 3 months, or telephone counselling as needed)
(N = 60).

Outcomes

Relapse: relapse rate.

Service use: re-hospitalisation rate.
Mental state: SAPS; SANS; BPRS.
Social functioning: SDSS (Social dysfunction screening scale).

Knowledge: ITAQ (Insight and treatment attitude questionnaire).

Compliance with medication.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomised using random number table.
Allocation concealment (selection bias)Unclear riskNot stated.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskBlinding of participants and personnel was not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias)
All outcomes
Low riskAssessor blind.
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk16 people left the study early, 6 from the intervention group and 10 from the control group. Incomplete outcome data were excluded from the final analysis.
Selective reporting (reporting bias)Low riskAll measured outcomes were reported.
Other biasLow riskNone obvious.

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
  1. a

    CBT: cognitive behavioural therapy
    CCMD-3: Chinese Classification of Mental Disorders Version 3
    DSM-IV: Diagnostic and Statistical Manual of Mental Disorders Version 4
    HARP: Health and Recovery Program
    fMRI: Functional magnetic resonance imaging
    ICD: International Classification of Diseases
    TAU: treatment as usual

Armstrong 1991Allocation: randomised.
Participants: 50% schizophrenia.
Interventions: does not include education to increase awareness of early signs of recurrence.
Beebe 2004

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Birchwood 2003

Allocation: randomised.

Participants: schizophrenia.

Interventions: early warning signs intervention is used for both control and experimental groups.

Buchkremer 1991Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups; mainly evaluated therapeutic relatives' group.
Carra 2007

Allocation: randomised.

Participants: relatives of people with schizophrenia.

Interventions: multiple group family treatment, does not include education to increase awareness of early signs of recurrence.

Cerniglia 1978

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Collins 2007

Allocation: randomised.

Participants: schizophrenia.

Interventions: cognitive behavioural social skills training.

Craig Thomas 2000Allocation: randomised.
Participants: population includes early psychosis.
Deng 2007

Allocation: randomised.

Participants: chronic schizophrenia (CCMD-3)

Intervention: social skills training vs routine care.

Dong 2008

Allocation: randomised. 

Participants: schizophrenia (CCMD-3), in remission.

Intervention: rehabilitation therapy (mainly social skills training and psycho-education) vs routine care.

Druss 2010

Allocation: randomised.

Participants: <5 0% schizophrenic, mostly bipolar disorder.

Interventions: Health and Recovery Program (HARP), no early warning signs training.

Falloon 1996

Allocation: randomised.
Participants: schizophrenia.

Interventions: early warning signs intervention is used for both control and experimental groups.

Foster 2008

Allocation: quasi-randomised.

Participants: serious and enduring mental illness, mostly depressive illness, < 50% schizophrenia.

Gaebel 2002Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Gleeson 2009

Allocation: randomised.

Participants: first episode of a DSM-IV psychotic disorder.

Interventions: early warning signs training versus TAU, which includes some early warning signs training.

Glick 1989

Allocation: randomised.
Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Glick 1990

Allocation: randomised.
Participants: schizophrenia

Interventions: does not include education to increase awareness of early signs of recurrence.

Glynn 2002Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Goldberg 1977Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Gong 2007

Allocation: Quasi-randomisation (randomised using hospital admission number: odds = intervention group, even = control group).

Participants: schizophrenia (CCMD-3).
Intervention: Family intervention vs routine care.

Grawe 1998Allocation: randomised.
Participants: population includes early psychosis and this is an exclusion criterion.
Gray 2004

Allocation: randomised.
Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Guohui 2003

Allocation: randomised.
Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Hahlweg 1999Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Haut 2007

Allocation: randomised.

Participants: schizophrenia.

Interventions: cognitive remediation versus control group skills training (includes relapse prevention skills).

Outcomes: working memory tasks, fMRI.

Hayward 1995

Allocation: randomised.
Participants: schizophrenia.

Interventions: do not include education to increase awareness of early signs of recurrence.

Herz 1989Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Herz 1991Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Hogarty 1988

Allocation: randomised.
Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Horan 2011

Allocation: randomised

Participants: schizophrenia, schizoaffective, delusional, disorder or psychosis

Interventions: Social Cognitive Skills Training, does not include education to increase awareness of early signs of recurrence.

Hou 2007

Allocation: Patients were randomly assigned according to hospital admission number.

Participants: first onset schizophrenia (CCMD-3) 

Intervention: insight and confidence training vs routine care.

Jolley 1990Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Jumnoodoo 2008Allocation: randomised.
Participants: ICD-10 criteria were not used for diagnoses and less than 50% of patients had a diagnosis of schizophrenia.
Kallert 2004aClarification from authors not available, no published papers relevant to the abstract from a conference.
Klosterkötter 2007

Allocation: randomised.

Participants: exclusion criteria was schizophrenia.

Interventions: clinical management and aripiprazole combined versus clinical management and placebo; CBT versus clinical management and placebo.

Kopelowicz 1998b

Allocation: randomised.
Participants: schizophrenia.

Interventions: do not include education to increase awareness of early signs of recurrence.

Leavey 2004Allocation: randomised.
Participants: population includes early psychosis.
Leclerc 2000

Allocation: randomised.
Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Leff 1984

Allocation: randomised.

Participants: schizophrenia.

Interventions:do not include education to increase awareness of early signs of recurrence.

Leff 2001

Allocation: randomised.
Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Lenior 2001Allocation: randomised.
Participants: schizophrenia.
Interventions: Early warning signs intervention is used for both control and experimental groups.
Liang 2002

Allocation: randomised.

Patients: schizophrenia (CCMD-2-R).

Intervention: Psychoeducation vs routine care. 

Liberman 1986

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Liberman 1988Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Linden 2001

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Linszen 1994

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Linszen 1998

Allocation: randomised.

Participants: schizophrenia, recent onset.

Interventions: Early warning signs intervention is used for both control and experimental groups.

Linszen 1998a

Allocation: randomised.

Participants: schizophrenia, recent onset.

Interventions: Early warning signs intervention is used for both control and experimental groups.

Lopez-Luengo 2003

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Malm 2003Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Marder 1994Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Marder 2003Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Mausbach 2008Allocation: randomised.
Participants: schizophrenia.
Interventions: does not include education to increase awareness of early signs of recurrence.
McFarlane 1995Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
McGill 1983

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Merson 1992

Allocation: randomised.

Participants: not schizophrenia.

Monking 1994

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Montero 2001

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Moxon 2008

Allocation: Randomised

Participants: Schizophrenia & family members

Interventions: Psychoeductaion, does not include education to increase awareness of early signs of recurrence.

Munroe-Blum 1993Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Newton 2005

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Nordentoft 1998Allocation: randomised.
Participants: population includes early psychosis.
Owens 2001

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Papageorgiou 2002

Allocation: randomised.

Participants: sample not specifically schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Pietzcker 1993Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Power 2003Allocation: randomised.
Participants: population includes early psychosis.
Rogers 2003

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Rossotto 2003Allocation: randomised.
Participants: schizophrenia.
Interventions: early warning signs intervention is used for both control and experimental groups.
Rotondi 2005

Allocation: randomised.

Participants: Schizophrenia and family members.

Interventions: web-based tele-health psycho education, includes early warning signs training, but no face-to-face contact.

Salzer 2004

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Saren 2003

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Schonell 1995

Allocation: randomised.

Participants: schizophrenia.

Interventions: Clarification from authors not available regarding whether early warning signs of recurrence was the focus of intervention.

Schooler 1997Allocation: randomised.
Participants: schizophrenia.
Interventions: Early warning signs intervention is used for both control and experimental groups.
Smith 1999

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

So 2006

Allocation: randomised.

Participants: carers of people with first episode psychoses (schizophrenia, schizoaffective or schizophreniform disorder).

Interventions: family psycho-education, relapse prevention only included as one third of a 1.5 hour session.

Tamaras 2000

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Tsang 2001

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Van Der Krieke 2010Allocation: not randomised.
Velligan 2002

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Wallace 1985

Allocation: randomised.

Participants: schizophrenia.

Interventions: does not include education to increase awareness of early signs of recurrence.

Wei 2007Allocation: Quasi-randomisation. Randomised according to hospital admission order.
Woods 2002Allocation: randomised.
Participants: population includes early psychosis.
杜芳 2010Allocation: quasi-randomisation. Randomised according to hospital admission order.
王莉 2010

Allocation: randomised.

Patients: schizophrenia (CCMD-3), male patients only.

Intervention: self-management education vs routine care.

许玉芳 2008

Allocation: randomised.

Patients: schizophrenia (CCMD-3)

Intervention: comprehensive skills training vs routine care.

许玉芳 2008a

Allocation: randomised.

Patients: Schizophrenia.

Intervention: comprehensive skills training vs routine care.

Characteristics of ongoing studies [ordered by study ID]

Farhall 2001

Trial name or titleApplicability of Cognitive Behavior Therapy to Unselected Mental Health Service Clients With Psychotic Disorders
Methods

Allocation: Randomised.

Duration: 12 to 24 sessions over a maximum of 12 month period.

Setting: Community mental health services - unspecified.

Country: Australia

Participants

Diagnosis: A preliminary DSM-IV diagnosis of schizophrenia, schizoaffective disorder, mood disorder with psychotic features, or delusional disorder.

N = target 94.

Age: not stated.

History: A recovery "need" in one of the areas addressed by the therapy elements (below)

Interventions

1. Recovery therapy: A variant of CBT. Implementation, as needed by the client, of one or more of the following elements: i) introduction or enhancement of techniques for better coping with persisting positive symptoms ii) adoption of a more adaptive explanatory model of their disorder by clients iii) strengthening of positive beliefs of self within a reality framework iv) implementation of practical strategies for relapse prevention, v) progress in resolution of personal/emotional issues impeding adaptation.

2. Standard case management.

Outcomes

Mental state: PANSS

Readmissions,

Insight,

Depression.

Uptake and effectiveness of Recovery therapy- dropout rates,

Starting date1/06/2001
Contact informationMr John Farhall
Address: Academic Unit Whittlesea Community Mental Health
Service c/o The Northern Hospital 185 Cooper Street
Epping VIC 3076
Country: Australia
Tel: +61 3 94098779
Fax: +61 3 94089508
Email: john.farhall@mh.org.au
Notes 

Jones 2012

Trial name or titleInvestigation into the efficacy of cognitive training on cognition in adults with schizophrenia.
Methods

Allocation: not reported.

Duration: not reported.

Setting: Multi-centre.

Country: England

Participants

Diagnosis: Schizophrenia DSM-IV

N = 75

Age: not reported.

History: not reported.

InterventionsCognitive training
OutcomesCognition
Starting dateNot reported.
Contact information

Ms Jennifer Cook
University of Cambridge
Psychiatry
Herchel Smith Building
Robinson Way
Cambridge
Cambridgeshire

CB2 0SZ
UNITED KINGDOM
Tel: 07793386431
jc746@cam.ac.uk

Notes 

NCT00046085

Trial name or titleOnline Family Support and Education for Schizophrenia.
Methods

Allocation: Randomised.

Duration: 18 months

Setting: not reported.

Country: United States, California

Participants

Relatives of patients with schizophrenia.

Diagnosis: DSM-IV diagnosis of schizophrenia or schizoaffective disorder.

N = 72.

Age: 18-60.

History: Stable antipsychotic medication regimen for at least 1 month.

InterventionsFamily Psychoeducational Program: Relatives of patients with schizophrenia receive 18 months of customary care with access to the educational website for the first year. Families are provided with private, secure access to the website, which features family-to-family chat capabilities, video lectures on the management of schizophrenia, written materials on important issues in schizophrenia management, professionally facilitated online discussions of the material, and additional resource links.
Outcomes

Relatives:

Perception of the patient's symptoms

Knowledge of the illness

Illness's impact on the family member

Perception of the website intervention

Patients with schizophrenia or schizoaffective disorder:
Symptoms

Medication compliance
Side effects

Hospitalisations

Social functioning

Starting dateMarch 2001
Contact informationNational Institute of Mental Health (NIMH)
ClinicalTrials.gov Identifier: NCT00046085
Notes 

NCT00069433

Trial name or titleEnhancing the Outcome of Skills Training for People With Schizophrenia
Methods

Allocation: Randomised.

Duration: 12 months.

Setting: not reported.

Country: United States.

Participants

Diagnosis: schizophrenia or schizoaffective disorder for > 5 years.

N = 60.

Age: 18 = 55.

History: Use of stable dose risperidone, olanzapine, or quetiapine for at least 1 month with no planned medication changes; social Behavior Scale (SBS) score > 45

InterventionsSocial skills and symptom management training
OutcomesNot reported.
Starting dateNot reported.
Contact informationNCT00069433
NotesTrial suspended. http://clinicaltrials.gov/ct2/show/NCT00069433

NCT00169052

Trial name or titleHealth Care Management and Rehabilitation Skills Training for Treating Serious Mental Illness in Older People
Methods

Allocation: randomised.

Duration: not reported.

Setting: community.

Country: USA.

Participants

Diagnosis: schizophrenia,schizoaffective disorder,bipolar disorder, major depression with functional impairment.

N = 180.

Age: over 50 years.

History: older persons with serious mental illness.

Sex: female.

InterventionsHealth management (HM) and supported rehabilitation (SR) intervention (by nurses who monitor and facilitate routine preventive and acute health care) versus usual care.
Outcomes

Independent living skills.

Social skills.

Health behaviour skills.
Preventive health care.

Acute emergency, hospitalisation and long-term institution-based care.

Starting dateSeptember 2001
Contact informationPrincipal Investigator: Stephen J Bartels, MD Dartmouth.
NotesClinicalTrials.gov identifier: NCT00169052

NCT00435721

Trial name or titleThe Families Coping With Mental Illness Program.
Methods

Allocation: randomised.

Duration: not reported.

Setting: not reported.

Country: USA.

Participants

Diagnosis: Self identified family members of individuals with schizophrenia or schizoaffective disorder or psychosis NOS

N = 50.

Age: 18 years and over.

History: not reported.

InterventionsPractical, short-term support and education program for relatives of individuals with schizophrenia.
Outcomes

Family knowledge.

Family Burden.

Expressed emotion.

Social Support.
Coping Strategies.
Confidence in knowledge.
Use of community resources.

Starting dateJune 2002
Contact informationPrincipal Investigator: Michelle S. Friedman-Yakoobian, Ph.D, Massachusetts General Hospital, Beth Israel Deaconess Medical Center.
Notes

Protocol
Allocation: randomised.

Participants: relatives of individuals with schizophrenia or schizoaffective disorder.

Interventions: behavioural psycho-education, does not include education to increase awareness of early signs of recurrence.

NCT00515671

Trial name or titleIllness Management and Recovery for Veterans With Severe Mental Illness
Methods

Allocation: Randomised.

Duration:18 months (nine months course, follow-up at 18 months).

Setting: Day Treatment Program of Roudebush Veterans affairs.

Country: USA.

Participants

Diagnosis: SCID-confirmed diagnosis of schizophrenia or schizoaffective disorder.

N = 200.

Age: 18 and older.

History: Currently receiving (or newly admitted to) mental health services from the Day Treatment Program of Roudebush Veterans affairs.

InterventionsIllness Management and Recovery Training
Outcomes

Illness Management Ratings

Patient Activation

Functioning Level,
Perceived Recovery
Psychiatric symptoms

Substance abuse

Hope

Employment
Independent living

Starting dateMay 2008
Contact informationClinicalTrials.gov identifier: NCT00515671

Contact: Michelle P Salyers, MS PhD
(317) 988-4419
mpsalyer@iupui.edu
Contact: Gretchen Ricketts
(317) 988-2802
grickett@iupui.edu
Notes 

Ancillary