Family interventions for bipolar disorder

Luis Justo; Bernardo Garcia de Oliveira Soares; Helena Calil

doi:10.1002/14651858.CD005167.pub2

. 2007 Oct 17;2007(4):CD005167. doi: 10.1002/14651858.CD005167.pub2

Family interventions for bipolar disorder

Luis Justo ^1,^✉, Bernardo Garcia de Oliveira Soares ², Helena Calil ³

Editor: Cochrane Common Mental Disorders Group

PMCID: PMC6669265 PMID: 17943843

Abstract

Background

Pharmacological treatments are the principal intervention for bipolar disorder. Alone, however, they are not sufficient to control symptoms and maintain psychosocial functioning. Adjunctive psychosocial interventions may help to improve the patient's condition and the course of the illness. Family interventions are deserving of special attention, since they may help to relieve the burden of care borne by relatives and caregivers, which in turn may facilitate the task of supporting the patient.

Objectives

The objective of this review was to investigate the effectiveness of family interventions in the treatment of bipolar disorder compared with no intervention and other forms of intervention.

Search methods

We searched the electronic databases CCDANRCT‐Studies and CCDANCTR‐References on 1/8/2007, CENTRAL (2006‐3), MEDLINE (2006), EMBASE (2006) and LILACS (2006), and searched the reference lists of included studies. We also made personal contact with authors.

Selection criteria

We included randomised controlled trials (RCTs) and quasi‐randomised trials. Participants were people with bipolar disorder and their relatives or caregivers; family psychosocial interventions of any type were considered; primary outcomes were changes in the status of symptoms and relapse rates.

Data collection and analysis

Data were independently extracted by two review authors. Quality assessment of included studies was carried out. The findings were presented descriptively. Where there were sufficient studies, dichotomous data were combined using relative risk, and continuous data were combined using weighted mean difference, with their 95% CIs.

Main results

Seven RCTs were included in the review, involving a total of 393 participants. All of the included studies assessed psychoeducational methods, and one study also assessed a type of systems psychotherapy. In all trials, participants continued to receive pharmacotherapy treatment. Due to the diversity of interventions, outcome measures and endpoints used across studies, it was not possible to perform meta‐analyses for primary outcomes. Five studies compared a variety of family interventions, involving carers, families or spouses, against no intervention, with individual findings indicating no significant added effect for family interventions. Three studies compared one type or modality of family intervention against another family intervention, with inconsistent findings.

Authors' conclusions

To date there is only a small and heterogeneous body of evidence on the effectiveness of family oriented approaches for bipolar disorder, and it is not yet possible to draw any definite conclusions to support their use as an adjunctive treatment for bipolar disorder. Further well designed RCTs should be a research priority.

Plain language summary

Family interventions for bipolar disorder

Drug treatments are the primary treatment for bipolar disorder. Alone, however, they are not sufficient to manage the disorder. Studies on psychosocial interventions for mental disorders such as schizophrenia and anxiety show that they are effective treatments. Reports in the literature suggest that they may be useful for people with bipolar disorder as well. The role of the family is important in the care of people with bipolar disorder, with effective family functioning helping to maintain a person's psychological balance. This systematic review investigated the effectiveness of any psychosocial family intervention for people with bipolar disorder and/or their families and carers. Seven randomised controlled trials (393 participants) were included in the review, all of which evaluated psychoeducational interventions. Five studies compared family interventions against no treatment, and three studies compared one type or delivery of family intervention against another family intervention. Differences in the interventions, outcome measures and end points used in the trials did not allow us to perform a meta‐analysis. Whilst results from individual studies did not suggest a significant effect for family interventions when added to drug therapy, the studies provide insufficient evidence to draw conclusions which can be generalised to everyday practice. Further research using appropriate randomised controlled trial methodology and evaluating family interventions other than psychoeducation is called for in this under‐researched and important topic.

Background

Bipolar disorder is a group of illnesses characterised by the presence of symptoms of pathologic variations of mood. According to DSM‐IV‐TR, people suffering from bipolar disorder type I have at least one episode of mania, commonly presenting with more than one episode, with or without depressive episodes. In bipolar disorder type II there is at least one episode of depression and one or more episodes of hypomania (never one episode of mania) associated. People may also exhibit mood symptoms that do not meet the criteria for bipolar disorder I or II, leading to a diagnosis of bipolar disorder not otherwise specified. Cyclothymic disorder may be diagnosed in people who do not meet the criteria for manic, mixed, or major depressive episode but exhibit periods of depressive symptoms and periods of hypomaniac symptoms for at least two years (adults) and no symptom‐free period longer than two months. Although bipolar disorder is usually described as a periodic disease, mood liability may occur between episodes and become an important cause of impairment in general abilities with potential harmful changes in the lives of affected people (Gitlin 1995; Kalbag 1999; Judd 2005). Consequently it is associated with significant morbidity, psychosocial and laborative maladjustment and high suicide risk (Judd 2005).

The lifetime prevalence of bipolar disorder I ranges from 0.4% to 1.6% and the lifetime prevalence of bipolar disorder II is 0.5%, in the adult population (APA 2002). Different ethnic groups do not show differences in these rates (APA 2002 ). Bipolar I disorder affects the same proportion of men and women, but bipolar II disorder is more common in women (APA 2002 ). The first episode in men is more likely to be manic, but for both genders the first episode is more frequently a depressive one (APA 2002). Divorce rates are two or three times higher among people with bipolar disorder than the general population, and occupational status is twice as likely to deteriorate (APA 2002). It is estimated that 25% to 50% of all people suffering from bipolar disorder will attempt suicide in their lives (Jamison 2000).

Bipolar and unipolar mood disorders may be distinct entities (Goodwin 1990), although this concept lacks consensus. Currently the diagnosis is mostly made according to ICD‐10 and DSM‐IV‐TR. The concept of bipolar disorder has now been broadened and so this illness is identified more often (Akiskal 2006). Aetiology seems to be strongly associated to genetic and biologic factors, but the involvement of psychosocial factors is increasingly gaining attention (Alloy 2005).

The prognosis for bipolar disorder, in spite of continual pharmacological maintenance treatment, is not always favourable. There is evidence suggesting that the course is less benign than previously thought. One study demonstrated a 73% relapse risk over a period of five years post recovery of an episode, and that morbidity appeared to be a more sensitive correlate of psychosocial functioning than the number of relapses (Gitlin 1995). Psychosocial impairments may persist even when mood symptoms are controlled by medications (Zaretsky 2003 ). Treatment consists predominantly of pharmacological agents such as mood stabilisers, antidepressants and antipsychotics when necessary, sometimes benzodiazepines and more rarely electroconvulsive therapy. Compliance is often a significant problem.

Pharmacological treatments are the fundamental tool in managing the illness, however they are not sufficient to control all the problems associated with the course of the disorder and its consequences in a person's life condition (Hilty 1999). It is especially difficult to achieve satisfactory prevention of relapse through the use of medication alone ( Gelemberg 1989; Gitlin 1995 ). In addition, the sophistication and complexity of questions linked to human psychological functioning and social performance demand more specific interventions in order to address subtle aspects of the ill person. It is also important to remember that subsyndromal symptoms persisting after the acute episode may be very difficult to manage, even with appropriate pharmacological treatment, and may cause non‐compliance, thus impairing restoration of former capacities and wellness (Coryell 1993). Additionally, it seems to be important with life events and correlated stress in order to prevent the negative impact they may have on the illness, specially regarding depressive symptoms and episodes (Johnson 2005; Johnson 2006). There is evidence that stressful events inside family milieu are connected to symptoms of bipolar disorder and expressed emotion is an important predictor of symptoms severity (Miklowitz 2005; Kim 2007). Bipolar disorder may represent a significative psychologic burden for family members and other caregivers (Perlick 1999). Family psychosocial interventions appear to be useful for people with other mental illnesses like schizophrenic disorders (Falloon 1984; Tomaras 2000).

Psychosocial interventions have been used for bipolar disorder since the pre‐pharmacological treatment era. In the early days they mostly appeared not to be useful therapies for people with this disorder, probably due to the lack of concomitant medication. However, the course of bipolar disorder has changed, through use of suitable biological treatments, and several forms of psychotherapies and psychoeducational methods are now applied as adjunctive care, with more evident success ( Huxley 2000; Gonzales‐Pinto 2004; Jones 2004). Different types of individual, group, and family psychosocial interventions are now applied, using a non‐specific or a specially tailored design for bipolar disorder (Jones 2004). Cognitive‐behavioural therapy, interpersonal therapy, interpersonal and social rhythm therapy, psychoanalytically‐based psychotherapy, family therapies (based on different theoretical orientations including cognitive‐behavioural, psychoanalytical and systemic theories) (Mikolwitz 1990 ; Huxley 2000 ; Reinares 2002 ; Fristad 2003 ; Jones 2004), and several types of group therapies are examples of psychotherapies that can be used to treat bipolar patients. Psychoeducational methods may also use different formats according to individuals, groups of patients or family settings.

There is general consensus that bipolar disorder affects relationships in patients' families, and that family relationships also affect the course of bipolar disorder ( Reinares 2002 ). Family interventions seem to be useful in controlling the illness (Reinares 2002; Kim 2004). Under the name of family therapy one can find very different kinds of interventions with diverse concepts in foundation. They can operate as direct psychoeducational methods with the explicit objective of instructing patients and relatives or carers about vicissitudes of the illness, or they can be behavioural therapy models with more or less educational purposes, often with communication enhancement and problem‐solving training. Alternatively, they can signify efforts for modifications in the functioning of the whole family group, treating them as a system where none is isolated and none is the single owner of the illness, as in the case of systemic family therapy models.

The aim of this review was to obtain and summarise all relevant trials that evaluate the effectiveness of all possible different forms of family interventions, according to their differing theoretical basis, for bipolar disorder. Family interventions were understood here to be any type of psychosocial interventions for family members of people with biploar, with or without the participation of the individual with bipolar disorder. It also included therapies with groups of families.

Objectives

1) To investigate the effectiveness (improvement of symptoms or reduction of relapse rates) of family interventions for bipolar disorder as compared to:   a) no intervention   b) other family psychosocial interventions

Methods

Criteria for considering studies for this review

Types of studies

All relevant randomised controlled trials (RCTs) evaluating family psychosocial interventions for bipolar disorder. Specific relevant outcome data from quasi‐randomised studies were included. Cluster randomised trials were also considered.

Types of participants

People with a diagnosis of bipolar disorder based on DSM‐III, DSM‐III‐R, DSM‐IV, DSM‐IV‐TR, IDC 9 or ICD 10 criteria and their relatives or caregivers. There were no restrictions regarding gender, age, ethnical group, nationality, stage of the disease (remitted or in acute phase), presence of co‐morbidity, and use of concomitant medication (as long as these were balanced by design).

Types of interventions

We included any family psychosocial interventions used for treating bipolar disorder in acute phase, or as preventive intervention in stable participants. Family psychosocial interventions could include any type of psychological therapy or psychoeducational methods for the treatment of bipolar patients and their families or caregivers (couples or group of relatives of a bipolar patient, or group of families of different bipolar patients, with or without the attendance of the bipolar patient),drawing from diverse theoretical foundations.

The term psychoeducation means to teach patients, family and caregivers about the illness and possible coping strategies. Psychological therapies are more difficult to define due to the diversity of theoretical approaches, but generally speaking, mean methods oriented to address psychological problems using a therapeutic relationship. The types of family psychosocial interventions could be: family psychoeducation methods, cognitive‐behavioural family therapy, cognitive family therapy, behavioural family therapy, interpersonal family therapy, psychodynamic family therapy, systemic family therapy, a mixed modality between this types (e.g. an intervention mixing psychoeducational and cognitive‐behavioural techniques). Couples therapy and therapies with groups of families were also included. Family interventions could be administered by psychiatrists, psychologists or other health care professionals.

The meaning of 'family' in this review was an extended one that encompassed people who were closely related to the patient, living in the same home or not, a biological relative or a significant and closely connected person for the patient, including close caregivers. Spouses were considered family.

As specified in the Background section, family interventions were categorised according to three different theoretical orientations, as follows:   1) Cognitive Behavioural Family Therapy (CBFT) (to include cognitive‐behavioural family therapy, cognitive family therapy, behavioural family therapy and family psychoeducation methods)   2) Psychodynamic therapy (including psychoanalytic psychotherapy, object relations)   3) Systemic therapy (including structural and post‐Milan)

Main comparisons:   1) Family interventions (stratified by category) versus no intervention   2) Family intervention versus other family psychosocial intervention   Where further studies become available in future updates of the review, the following additional comparisons are planned:   3) CBFT versus psychodynamic therapy   4) CBFT versus systemic therapy   5) Systemic therapy versus psychodynamic therapy

Types of outcome measures

Primary outcome   Effectiveness of interventions measured by:   1) Changes in the status of the illness, as measured by standard scales, such as Young Mania Rating Scale‐YMRS, Montgomery Affective Disorders Rating Scale‐MADRS, Hamilton Rating Scale for Depression‐HRSD, Schedule for Affective Disorders and Schizophrenia‐SADS, Brief Psychiatric Rating Scale‐BPRS, Bech‐Raphaelson Manic Scale, or any other validated scale; for studies where subjects are in the acute phase of disease or have subsyndromic symptoms in maintenance phase;   2) Relapse rates, for studies where participants are stable (not in an acute phase of disease).

Secondary outcomes:   1) Hospitalisation   2) Length of remission   3) Suicide attempts   4) Treatment compliance (attendance at psychosocial treatment appointments)   5) Dropout rates at endpoint   a) Number of participants who dropped out because of lack of efficacy   b) Number of participants who dropped out because of symptoms worsening   6) Employment related events (Work Adjustment Scale, and any other validated scale or objective event like as loss of job, interruption or return to work)   7) Social and family functioning (UCLA‐Social Attainment Survey, Family Assessment Device‐FAD, Global Functioning Scale‐GFS, or any other validated scale)   8) Quality of life (SF‐36, World Health Organisation Quality of Life Scale ‐ WHOQOL, or any other validated scale)   9) Anxiety levels (post‐hoc outcome)

Outcomes would be grouped for analysis according to the duration of active treatment and follow up: less than 6 months, 6 to 12 months, 12 to 24 months. For the current version of the review, this was not performed due to limited number of studies and heterogeneity of data.

Search methods for identification of studies

1)Electronic search   The CCDAN registers were searched using the following search strategies:

CCDANCTR‐Studies ‐ searched on 1/8/2007   Diagnosis = "Bipolar Disorder" or "Bipolar I Disorder" or "Bipolar II Disorder" or "Depression, Bipolar" or "Depression, Psychotic" or "Major Affective Disorders" or "Manic Disorder" or Mania or "Mood Disorders" or "Bipolar Not Otherwise Specified" or "Psychotic Disorders" or Psychoses or "Treatment resistant"   and   Intervention = Family or Marital or Spous* or Couples*

CCDANCTR‐References ‐ searched on 1/8/2007   Free‐text = "Bipolar III Disorder" or "Unipolar Mania" or "Rapid Cycling Disorder" or "Affective Disorders" or "Affective Psychosis, Bipolar" or "Bipolar Disorder " or "Bipolar Disorder" or "Bipolar I Disorder" or "Bipolar II Disorder" or "Cyclothymic Disorder " or "Depressive Psychosis" or "Excited Psychosis" or "Hypomania" or "Mania" or "Manic‐Depressive" or "Manic Disorder" or "Manic Episode" or "Melancholia" or "Mixed Depression" or "Mood Disorders" or "Bipolar Affective Disorder " or "Bipolar Not Otherwise Specified " or "Dysphoric Mania" or "Manic Episode" or "Manic Symptoms" or "Schizoaffective Disorder" or "Psychoses" or "Psychotic Disorders" or "Puerpal Psychosis " or "Reactive Depressive Psychosis"   and   Free‐text = family* or marital or couple* or spous*

The following databases were searched to identify randomised or quasi‐randomised controlled trials: The Cochrane Central Register of Controlled Trials (CENTRAL) (2006‐3), MEDLINE (1966‐2006) EMBASE (1980‐2006), and LILACS (1982‐2006). The "optimal" MEDLINE, EMBASE and LILACS sensitive search strategies for identification of controlled clinical trials (Castro 1997; Dickersin 1994) were combined with the following phrases:

#1 affective disorders OR mood disorder OR bipolar disorder OR bipolar psychosis OR manic disorder OR mania OR manic psychosis OR manic depression OR hypomanic OR hypomania OR mixed mania OR mixed states OR mixed episodes OR affective symptoms OR bipolar depression OR cyclothymic disorder OR cyclothymia

#2 family therapy OR family intervention OR family treatment OR family process therapy OR family management OR psychoanalytical family therapy OR psychodynamic family therapy OR supportive family therapy OR cognitive behavioral family therapy OR systemic family therapy OR continuation family treatment OR maintenance family treatment OR prophylactic family treatment OR interpersonal family therapy OR cognitive family therapy OR couples therapy OR marital therapy OR spouses treatment

2) Handsearches   Bibliography of the identified studies were checked.

3) Personal Communication   Attempts to contact the authors of included studies were made when any clarification was needed, and to find out about ongoing and unpublished studies.

Data collection and analysis

Selection of trials   LPJ and BGOS screened the abstracts of all publications obtained by the search strategy (a track record was kept). The articles potentially suitable for the review were obtained in full to assess their relevance, based on broad inclusion criteria (all potential RCTs in family interventions).

Quality assessment   The quality of each trial was based on the criteria of quality specified by Shultz 1995, which measure the following range of factors:   1) Minimisation of selection bias: a) was the randomisation procedure adequate? b) was the allocation concealment adequate?   2) Minimisation of attrition bias: a) were the withdrawals and dropouts completely described? b) was analysis by intention to treat?   3) Minimisation of detection bias: a) were outcome assessors blind to the intervention?   Based on these criteria, studies were classified according to the criteria described in the Cochrane Handbook (Higgins 2005)   A‐Low risk of bias (all quality criteria met)   B‐Moderate risk of bias (quality criteria partially met)   C‐High risk of bias (one or more criteria not met)   Trials were assessed independently by two review authors (LPJ, BGOS). In cases of disagreement the third review author (HMC) would be contacted.

Data extraction   Data were extracted independently by two review authors (LPJ, BGOS) using a standard data extraction form (Higgins 2005). This included data on: methods (generation of the allocation sequence, concealment of allocation, sample size estimation, length of follow‐up), participant characteristics (diagnostic procedures, age, gender, ethnic origin, criteria used to classify recurrence, number of patients randomised, reasons for withdrawal from the trial), interventions, and outcomes (specified previously, any other assessed outcomes, other events, length of follow‐up, reporting of outcomes quality).

Data analysis   Review Manager software developed by the Cochrane Collaboration was used to organise and process the results. Interventions were only grouped if clinically compatible, otherwise the findings were presented descriptively.

For dichotomous data, relative risks (RR) with 95% confidence intervals (CI) were estimated based on the fixed effects model, or on the random effects model when heterogeneity is present. An intention‐to‐treat analysis would be used in which the reviewers would assume that people who dropped out had a negative outcome, except in case there are death outcomes. For all statistically significant results, number needed to treat to benefit (NNTB) or number needed to harm (NNTH) were calculated with 95% confidence intervals (CI).

Continuous outcomes were analysed if the mean and standard deviation of endpoint measures were presented in the original articles. For the meta‐analysis of continuous outcomes, mean differences (MD) between groups were estimated. The weighted mean difference (WMD) would be used if the data were obtained from the same measurement scales, and the standardised mean difference (SMD) would be used if the measurement scales differed. Data on continuous outcomes are frequently skewed, the mean not being the centre of distribution. To avoid this potential pitfall, the following standards were applied to all data before inclusion: a) standard deviations and means would be obtained from authors, and b) for data with finite limits, such as endpoint scale data, the standard deviation (SD), when multiplied by two, was less than mean. Otherwise the mean is unlikely to be an appropriate measure of the centre of the distribution (Altman 1996). Only non‐skewed data would be used in meta‐analyses.

In order to describe the percentage of the variability in effect that occurs due to heterogeneity rather than to chance, it was planned to perform an I‐squared calculation (values greater than 50% to be considered substantial heterogeneity).Heterogeneity would also be assessed by the chi‐square test, and was assumed to be present when the significance level was lower than 0.10 (p<0.10). When significant heterogeneity was present, and a sufficient number of studies were available for comparisons, an attempt would be made to explain the differences, through use of subgroup analyses according to clinical characteristics of the included studies, or through the use of sensitivity analyses according to methodological differences in study design or degree of control over bias (e.g. allocation concealment).

Subgroup analyses   The following subgroup analyses were planned:   1) Family interventions directed at spouse and patient (couples therapy) versus family interventions for relatives and patient   2) Individual family versus group family interventions   3) Family interventions conducted with the individual versus family interventions conducted with the family   4) Family interventions with family members only versus family interventions with family members and patients   5) Family psycho‐educative methods versus other family intervention (CBT, psychodynamic, systemic) methods   6) Structural therapy versus post‐Milan therapy

Sensitivity analyses   Randomised versus quasi‐randomised trials

Publication bias   To assess potential publication bias, trial data (trial effect versus trial size) would be used to produce a funnel graph.

Where it was not possible to pool data from studies due to heterogeneity of interventions and outcomes, and a small number of studies for inclusion, the findings were presented descriptively under the appropriate comparison headings.

Results

Description of studies

Results of the search   Electronic searches from all databases generated 1892 references, most of which were excluded because they did not cover the subject of our study, based on scrutiny of the title alone or after reading the abstract. From the search, 20 studies were considered potentially relevant, and articles were obtained and scrutinised. Nine studies did not meet the inclusion criteria, and are described in the Characteristics of excluded studies table. Two studies are ongoing (Miklowitz 2004, Miklowitz 2006a). Two studies are awaiting assessment to see if they are dual publications of included studies (Keitner 1996b, Miller 2000). The remaining seven studies were included in the review. These are described below and in the Characteristics of included studies table.

Excluded Studies   Five studies were excluded because they were not randomised or quasi‐randomised controlled trials. Two studies did not examine the effectiveness of family interventions on bipolar disorder as a primary or secondary outcome. Two studies were secondary reports from a trial already included in the review.

Studies awaiting assessment   There are two studies awaiting assessment (Keitner 1996b, Miller 2000)..

Included Studies   Participants   The total number of participants in included trials was 393, of whom 84 participants (Reinares 2004 and van Gent 1991), were spouses or caregivers (generally relatives). All participants were adults, with ages ranging from 18 to 62 years. Further information on characteristics of the participants is presented in the Characteristics of included studies table.

Interventions   Experimental family interventions were largely psychoeducational treatments for patients with their families, or for the families without participation of the patient. One intervention, Problem Centered Systems Therapy for the Family, was not specifically psychoeducational (Ryan 2003). Patients in experimental and control groups received concurrent pharmacological treatment.

1) Family Focused Therapy (FFT): Two studies (Miklowitz 1996, Goldstein 1996) used FFT as the experimental treatment. FFT was adapted for bipolar patients (Miklowitz 1997) from a previous psychosocial intervention model created for schizophrenic patients (Falloon 1984), and is a type of education for patients and their families. FFT consists of three components: psychoeducation about bipolar disorder, communication enhancement training and problem‐solving skills training. FFT was administered in 21 one‐hour session (12 weekly, 6 biweekly and 3 monthly) in 3 consecutive modules: psychoeducation (7 sessions), communication enhancement training (7‐10 sessions) and problem‐solving skills (4‐5 sessions), administered in the 9 month period following an episode of bipolar illness. One of the studies (Goldstein 1996) compared FFT with an individual intervention, named individually focused patient treatment, in which the goals were to educate the patient about the illness, conduct crisis interventions and reduce ongoing life stress. The other (Miklowitz 1996) used a less intensive psychoeducational family intervention called crisis management (CM) as the control condition, with two 1‐hour home‐based sessions of family psychoeducation within the first 2 months after study entry and crisis intervention sessions as needed during the remaining 9‐month treatment period. Both studies were performed with outpatients.

2) Problem Centered Systems Therapy of the Family: This is a short‐term, problem‐focused, semi‐structured family intervention that is manualised and is based upon the McMaster Model of family functioning (Epstein 1990). The therapy focus is directed to clinically relevant dimensions of family functioning as problem solving, communication, roles, affective responsiveness, affective involvement and behaviour control. This therapy relies on a core set of therapeutic principles tailored to the individual family's problems. The specific problems of the family are determined by the therapist and the family together. The therapy is performed through a progressive series of stages: assessment, contracting, treatment and termination. The number of sessions varies depending on the needs of the family ranging from 6 to 10 fifty‐minute sessions. This intervention was studied in Ryan 2003 in a comparison with two other groups of treatment that were multifamily psychoeducation and pharmacotherapy alone. Inpatients, partial hospital patients and outpatients were included.

3) Multifamily Psychoeducational Group Therapy: This is a manual based semi‐structured intervention developed for the study conducted by Ryan 2003, which included 4 to 6 patients and their family members above the age of 12 years. The sessions provided information about the bipolar disorder and taught members different coping strategies for common problems. The psychotherapists encouraged patients and family members to share their perspectives on family interactions. It was conducted in weekly 90‐minutes 6 sessions, each of them focusing on a specific topic. Inpatients, partial hospital patients and outpatients were included.

4) Couple Psychoeducational Intervention: This is a manual‐based psychoeducational intervention for patients and their partners, administered by social workers trained in family therapy. Couples in the experimental group received 25 (10 weekly and 15 bimonthly) sessions. In Clarkin 1998, this intervention was compared to standard medication for patients only. Inpatients and outpatients were included.

5) The psychoeducational family intervention studied in Reinares 2004 consisted of 12 ninety‐minute group sessions for patients' caregivers, and took place in a hospital setting. Relatives of 10 mood stabilised patients were included in each group and the patients themselves did to attend the group sessions. Oral information about the illness and guidelines on its management were provided. Encouraging discussion between participants was facilitated. After each session the participants received a written summary about the topic of the day. The same psychologist conducted all the groups. In the control group patients received pharmacological treatment using the same algorithms of the Barcelona Bipolar Disorders Program that were administered to patients in the experimental group, but the patients' relatives did not receive psychoeducational intervention. Similarly, in the study by van Gent 1991, only patients' partners received a psychoeducational intervention. The partners attended five sessions in which information about the illness, pharmacological treatment and practical advice was provided, and written material was given to participants. Partners' experiences were solicited and discussed in the sessions. The group sessions were conducted by a psychiatrist and a social worker.

6) Inpatient family intervention: this type of intervention was used in Clarkin 1990. It was designed for inpatients and their families to be performed during the period that the patient was hospitalised through at least six 45 min to one hour family intervention sessions. It was a manual based psychoeducational method, whose goals were acceptance and understanding by the patient and family of the reality of the illness, identification of precipitating stresses for present episode and future, within and outside of the family, elucidation of family interactions, planning strategies for managing stresses and acceptance of the need for continued treatment after discharge.

Outcomes   Primary   1) Patients' affective symptoms: measured by standard rating scales as Hamilton Depression Rating Scale, Young Mania Rating Scale, Schedule for Affective Disorders and Schizophrenia‐Change version, Brief Psychiatric Rating Scale, Bech‐Rafaelsen Mania Scale.   2) Recovery (numbers of patients that have recovered in the comparison groups): using symptom rating scales and transforming them into dichotomous measures (yes/no).

Secondary   1) Hospitalisations or rehospitalisations.   2) Adherence to pharmacologic treatment measured through self‐report scales provided by patients and their family, or by laboratory tests.   3) Dropouts (after randomisation): dropout data were reported in all studies.   46) Relationship within the family: measured through Family Environment Scale.

Settings   Two of the included studies were performed in Europe, one in Spain (Reinares 2004) and another in Netherlands ( van Gent 1991). The remaining five included studies were conducted in USA.

Risk of bias in included studies

Allocation   All studies included in this review were described as randomised by their authors, nevertheless only three studies (Reinares 2004, Miklowitz 1996, Clarkin 1998) provided some specific information regarding randomisation processes. Only one study mentioned allocation concealment (Miklowitz 1996), with randomisation sequence concealed until assignments had been made. Regarding allocation concealment, the study by Miklowitz 1996 was classified as A ,and all the others were classified as B, for the reason that only the first study described concealment, and the others did not give sufficient information.

Blindness to Evaluation of Outcomes   Four of the included studies reported blindness for outcome measurements (Reinares 2004, Miklowitz 1996, Goldstein 1996, Clarkin 1990). One of these studies (Miklowitz 1996) provided secondary blinded evaluators to analyse information through video‐tapes of interviews, and the authors compared the primary and secondary outcome rates because they considered it very difficult to keep the patients' psychosocial group assignments blind. The other studies (Reinares 2004, Goldstein 1996, Clarkin 1990) simply utilised evaluators blinded to the type of patients' psychosocial treatment group.

Intention‐To‐Treat Analysis and Loss to Follow Up   Intention‐to‐treat analysis was performed in two studies (Ryan 2003, Goldstein 1996), in that all participants randomised were included in final analyses. One study (Miklowitz 1996) did not include patients who terminated prematurely in the final statistical analysis, but provided data for them. Another trial (Reinares 2004) mentioned the loss of two participants who failed to complete endpoint assessments, but the authors did not clarify whether or not their data were included in final analyses. In another study (van Gent 1991) five of the 19 participants randomised to the experimental group and eight of the 20 participants randomised to the control group were not allowed to be enrolled in the study by the patients (their partners) and they were considered dropouts. One study reported missing data, and there is a mention of an intention‐to‐treat analysis procedure related to composite outcomes, but not for individual measures (Clarkin 1990). With just one exception (Clarkin 1998), all reported studies included explanations on reasons for dropouts.

Effects of interventions

Comparison 1. Family intervention versus no intervention (see Graphs 01 ‐ 05)   We found five studies comparing any type of psychosocial family intervention (partners, caregivers or other members of family) (Ryan 2003, Reinares 2004, Clarkin 1998, van Gent 1991, Clarkin 1990) to no intervention. Clinical improvement outcomes (van Gent 1991, Clarkin 1990) and recovery (Ryan 2003) were reported, but were not combinable because they were measured in different ways. Secondary outcomes measured in individual studies included hospitalisation, medication compliance, relationships in the family environment and anxiety. Dropout outcome data were extracted from all five studies. Clarkin 1998 contributed dropout data only.

Graphs 01.01 to 01.09 present outcomes from all five studies. Graphs 02 to 05 present outcomes from four of the five studies individually (Clarkin 1990, Reinares 2004, Ryan 2003, van Gent 1991).

Primary outcome   1) Recovery: There was no significant difference in recovery rates between groups at 28 months' post‐treatment for all patients (62 participants, RR 0.87, 95% CI 0.52 to 1.47) or for manic patients only (45 participants, RR 0.82, 95% CI 0.43 to 1.55) (Ryan 2003).

2) Clinical improvement: In the study by Clarkin 1990, there was no significant difference in clinical improvement rates between groups at post‐treatment (26 participants, RR 0.49, 0.10 to 2.45) or at 6 months follow‐up (RR 0.73, 95% CI 0.05 to 10.49). Similarly, in the study by van Gent 1991, there was no significant difference in SCL‐90 total symptoms between groups at 12 months' post treatment (39 participants (RR, 0.03, 95% CI ‐0.59 to 0.66).

3) Relapse: No studies provided data on relapse rates.     Secondary outcomes   1) Hospitalisation: No patients were rehospitalised during the study by van Gent 1991.

2) Medication compliance: There were no significant differences between the family intervention and no intervention groups for medication compliance (van Gent 1991)

3) Relationships in the family environment: There were no significant differences between the family intervention and no intervention groups for relationships in the family environment (expressiveness, cohesion and conflict) (Reinares 2004).

4) Dropout: A meta‐analysis of dropout rates (5 studies, 214 participants) did not show any significant difference between the family interventions group and the no intervention group (RR 0.70 95%CI 0.43‐1.14) (Clarkin 1990, Clarkin 1998, Ryan 2003, van Gent 1991, Reinares 2004).

5) Anxiety (post‐hoc outcome): Patients in the family intervention group had increased levels of anxiety when compared to the no intervention group (WMD 0.69; 95% CI 0.05‐1.34) (van Gent 1991).     Comparison 2. Family intervention versus other family psychosocial intervention (See Graphs 06 ‐ 07)   One study compared a systems therapy family intervention called Problem Centered Systems Therapy of the Family to a multifamily group psychoeducation intervention (Ryan 2003).

Primary outcome   1) Recovery: There was no significant difference in recovery rates between the two intervention groups for all patients (63 participants, RR 1.72, 95% CI 0.91 to 3.25), or for manic patients only.

Secondary outcomes   There was no significant difference between the multifamily psychoeducative intervention and Problem Centered Systems Therapy of the family for medication compliance or dropout rates.

Subgroup analyses     1. Family intervention versus individual intervention ‐ subgroup analysis 3 (see Graphs 08)   One study compared a family psychosocial intervention to an individual psychosocial intervention (Goldstein 1996). Both interventions were based on similar principles. The main difference was the administration to participants alone or with their family.

Primary outcome   Relapse: There were no significant differences between the family and individual psychosocial intervention at post‐treatment (12 months) (53 participants, RR 0.89, 95% CI 0.52 to 1.54).

Secondary outcomes   There were no significant differences between family and individual interventions for rehospitalisation, medication compliance and dropout rates.     2. Psychoeducative methods versus other family psychosocial intervention ‐ subgroup analysis 5   One study compared Family Focused Therapy (FFT) with a more simple psychoeducative intervention called Crisis Management (CM), which was also administered to the family, but in a less intensive and complex way (Miklowitz 1996).     Primary outcome   Relapse: For relapse prevention, FFT was significantly superior to CM in preventing relapse (101 participants, RR 0.59, 95% CI 0.39‐0.88; NNT=3.1; CI 1.9‐8.3)

Secondary outcomes   There was no significant difference between CM and FFT for medication compliance or dropout rates.

Other data   1) Survival analysis   Data from survival analysis cannot be re‐analysed in Review Manager. Two studies presented primary outcomes using survival analysis. Miklowitz 1996 demonstrated superiority of the Family Focused Therapy group over the Crisis Management group for time to relapse (Wilcoxon x2 = 8.71, p = 0.03) and Ryan 2003 reported no significant statistical difference between the three groups for time to recovery (log rank x2 = 1.21, df = 2, p = 0.55).

2) ANOVA   Miklowitz 1996 reported results of a repeated‐measure mixed‐model ANOVA regarding effect of time on total affective symptoms scores (measured by SADS‐C). There was a statistically significant treatment group versus time interaction (F7,549=2.81, p=0.007) for symptom severity stabilised at lower level, which means that the FFT group presented significantly lower symptomatology scores from 6 months of treatment.

Data available for analysis   Further information about results can be found in the Characteristics of Included Studies table. Some outcomes were not suitable for meta‐analysis due to lack of complete information, such as standard deviations for means. Other outcome data could not be re‐analysed in this review because of the way they were measured or presented, for example, outcomes measured through survival analysis (although it is perhaps worth mentioning that the outcomes and measures are absolutely valid). We attempted to obtain additional information through making contact with the authors of studies, but it has not yet been possible to obtain this information.

Discussion

In the process of conducting this review we found a number of studies examining the usefulness of different types of psychosocial interventions as an adjunctive treatment for bipolar disorders, many of which reported on a modality of family psychosocial intervention. However, the majority of the papers retrieved from the literature were not randomised or quasi‐randomised controlled trials, therefore, only seven trials have so far been included in the review. Participants in all included studies were adults. All seven trials examined family psychoeducational methods, with only one trial assessing another specific type of psychotherapy, problem centered systems therapy of the family, in addition to psychoeducation (Ryan 2003 ).

The seven studies included in the review presented characteristics of the samples at baseline, described diagnostic procedures, and set out inclusion/exclusion criteria. Moreover, they used relevant outcomes, and provided detailed descriptions of interventions. However, outcomes were diverse and data were often presented in a way that prevented their inclusion for the purposes of meta‐analysis, as in the case of those trials that used survival analysis (Ryan 2003; Miklowitz 1996; Goldstein 1996). A further problem was the lack of essential information, as in one study that did not present standard deviations for means of continuous data (Clarkin 1998). In that particular study, we were unsuccessful in obtaining further information from the authors ( Clarkin 1998 ). Some outcomes were not directly clinical, for instance, focusing on the illness outcome itself. Other outcomes were not directly linked to any modification in the manifestations or evolution of the disease, such as the patients' and their family members' knowledge of the disease, and therefore we did not include all their findings in analyses ( Reinares 2004; van Gent 1991).

Findings from individual studies appear inconclusive in demonstrating an effect for family interventions when compared with no family intervention. For studies comparing one family intervention against another type of family intervention, the Miklowitz 1996 study, a psychoeducational trial carefully conducted over a two year period, presented significant findings on relapse rates, from which the NNT shows that out of every three patients treated with family focused therapy, one less patient would relapse over time in comparison with patients undergoing crisis management intervention. For secondary outcomes, a result with clinical and statistical significance came from van Gent 1991, where participants attending for marital psychoeducation had a higher mean score in the anxiety scale than the control group (WMD 0.69; 95%CI 0.05‐1.34). This could be considered a side‐effect, which might be accounted for by the discomfort induced by the knowledge of the dificult aspects of the disorder, such as chronicity, and the fact that patients can relapse even with adequate medication. This increase in the levels of anxiety might be attributed to chance, and is also acknowledged as a post‐hoc finding. Further studies may elucidate this issue.

Results of our review are limited due to the small number of randomised controlled trials on this subject. Additionally, the heterogeneity of studies, specially because they assessed different outcomes, and the lack of trials for each type of comparison performed, were major limitations. Although a non‐randomised controlled study might produce useful information in this research field, we wished to follow the standards and criteria of the Cochrane Collaboration, in which the randomisation is a fundamental procedure in a health treatment study. In any case, we agree with the idea according to which the randomisation is really an important step in the trial process in order to minimise selection bias.

In general, there is increased acceptance of the need for adjuvant psychosocial interventions (psychotherapies and psychoeducational methods) added to standard medications in the treatment of bipolar disorder. It is suggested that the addition of psychosocial interventions may reduce symptoms, enhance social functioning, reduce hospitalisations and relapse rates and increase adherence to the treatment of bipolar disorder (Vieta 2005 b; Gutierrez 2004). Results of a systematic review on psychosocial interventions in bipolar disorder ( Vieta 2005 a ) found evidence of benefits of these interventions in combination with pharmacologic treatment. Furthermore, the same review considers the possibility that different types of psychosocial intevention have a distinct impact on the treatment of different types of bipolar patients. A related observation in another article suggests that mood fluctuations may have multiple distinct aspects in different patients, which might require different psychosocial intervention models in order for the health professional to manage the complexity of the illness (Jones 2005). In other words, bipolar disorder might present in different ways within the same phase or in the same patient, and possibly family intervention treatment could be more effective if these variations of clinical presentation were assessed at the point at which the intervention is initiated.

Family members or caregivers may play a very important role in detecting subtle mood fluctuations of the patient, and could act therapeutically if properly prepared. It is possible that improving the environment, in which family functioning plays a major role, may be one kind of help for patients. On the other hand, stressful conditions in the family context, such as excessive hostility or overinvolvement (also denominated "expressed emotion"), could result in increased risk for patients ( Miklowitz 1998 ). Among psychosocial interventions, family interventions may be as promising as other psychosocial interventions in improving therapeutic outcomes, and perhaps even more so, because they involve the patient's immediate world. Some important authors working in this field recommend the implementation of family interventions for the treatment of bipolar disorders (Miklowitz 2006a ). Nevertheless, given that studies with adequate methods to produce good quality evidence appear to be the exception rather than the rule in the literature, it will be necessary to gather more data yielded by good quality randomised controlled trials about effectiveness of family psychosocial interventions to provide a sound scientific basis for the use of these interventions. The fact that there are researchers working with samples of bipolar patient in child and adolescent populations is important (Fristad 2002), as these young people can present with different symptoms, behaviours and responses to psychosocial interventions, and in addition, may be more dependent on their family and its functioning.

The majority of interventions included in this review used psychoeducational methods, and the heterogeneity of interventions and outcomes, together with the small number of trials for inclusion, means that the evidence is not yet compelling. In the case of psychotherapies, the paucity of data is especially significant. Randomised controlled trials involving more specific family psychotherapies such as those based on cognitive behavioural, systemic family therapy and psychodynamic principles are needed. It is important to stress that although few significant results favouring psychosocial interventions were found in this review, it is not appropriate to conclude that family psychosocial interventions added to pharmacological treatment for bipolar disorders are not useful. Family interventions might be an important means of increasing the effectiveness of treatment for biolar patients, and of helping relatives and clinicians in the task of better dealing with the illness and its consequences. What is clear with this review is that it is difficult to make a precise evaluation of family psychosocial interventions' effectiveness, mainly due to the small and heterogeneous number randomised controlled trials and some of their methodological limitations. It is highly desirable that professionals working in this field concern themselves with carrying out further well designed studies.

Authors' conclusions

Implications for practice.

Whilst the addition of family psychotherapy or psychoeducation interventions to medications in the treatment of bipolar patients might be considered good practice, it is important to be aware that there is insufficient evidence in the literature to allow clear conclusions to be drawn on their effectiveness. Therefore, it is not possible to recommend without restriction that clinicians or health policy makers should use family interventions in the treatment of bipolar disorder or to recommend one family intervention as more effective than another. It is also important to bear in mind the cost/benefit ratio in this type of intervention, not only regarding the economic aspects, but also patients' and their families' affective investment and expectations.

Implications for research.

We suggest that researchers should carry out further randomised controlled trials to answer more clearly the questions about efficacy and effectiveness of family psychosocial interventions in the treatment of bipolar disorder. Use of rigorous methods of randomisation, larger sample sizes, standard outome assessments, measurement of adverse effects/acceptability and precise reporting of findings are called for, to produce reliable and applicable information on this important and under‐researched topic.

What's new

Date	Event	Description
1 November 2008	Amended	Converted to new review format.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 No recovery at end of study	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.1 all patients (28 months)	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.2 patients with mania (28 months)	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 No significant clinical improvement	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.1 at post‐treatment	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.2 at 6 month follow‐up	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Patients' total symptoms post‐treatment (SCL‐90)	1		Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5 Positive relationship within the family (high score=more positive)	1		Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.1 cohesion	1		Std. Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.2 expressiveness	1		Std. Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 Negative relationship within the family (low score=more positive)	1		Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
6.1 conflict	1		Std. Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7 Medication compliance	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
7.1 at end of study (12 months)	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
8 Patients anxiety post‐treatment (Trait Anxiety Inventory)	1		Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
9 Patients' anxiety post‐treatment (subscale SCL‐90)	1		Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
10 Dropouts at end of study	5	214	Risk Ratio (M‐H, Fixed, 95% CI)	0.70 [0.43, 1.14]
10.1 at 6 months	1	45	Risk Ratio (M‐H, Fixed, 95% CI)	0.5 [0.03, 7.45]
10.2 at 11 months	1	42	Risk Ratio (M‐H, Fixed, 95% CI)	0.15 [0.02, 1.10]
10.3 at 12 months	1	39	Risk Ratio (M‐H, Fixed, 95% CI)	0.66 [0.26, 1.66]
10.4 at 18 months	1	26	Risk Ratio (M‐H, Fixed, 95% CI)	1.1 [0.22, 5.51]
10.5 at 28 months	1	62	Risk Ratio (M‐H, Fixed, 95% CI)	1.05 [0.54, 2.07]

Outcome or subgroup title	No. of studies	Statistical method	Effect size
1 Positive relationship within the family (high score=more positive)	1	Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
1.1 cohesion	1	Std. Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.2 expressiveness	1	Std. Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 Negative relationship within the family (low score=more positive)	1	Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2.1 conflict	1	Std. Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Patients' total symptoms post‐treatment (SCL‐90)	1		Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2 No medication compliance	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.2 at end of study (12 months)	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Patients anxiety post‐treatment (Trait Anxiety Inventory)	1		Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4 Patients' anxiety post‐treatment (subscale SCL‐90)	1		Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5 Dropouts	1	39	Odds Ratio (M‐H, Fixed, 95% CI)	0.54 [0.14, 2.08]

Outcome or subgroup title	No. of studies	Statistical method	Effect size
1 No recovery for all patients	1	Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.1 at the end of study (28 months)	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 No recovery for patients with mania	1	Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.1 at the end of study (28 months)	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Dropouts	1	Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3.1 at month 6	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

Outcome or subgroup title	No. of studies	Statistical method	Effect size
1 No recovery at end of study (28 months)	1	Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.1 all patients	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.2 patients with mania	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 Dropouts	1	Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.1 at 6 months	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

Outcome or subgroup title	No. of studies	Statistical method	Effect size
1 Relapse rates	1	Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.1 at the end of treatment (12 months)	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 Rehospitalisation	1	Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.1 at the end of treatment (12 months)	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Medication compliance (high score is better)	1	Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.1 at the end of study (24 months)	1	Std. Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Dropouts	1	Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
4.1 at the end of study (24 months)	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

Outcome or subgroup title	No. of studies	Statistical method	Effect size
1 Relapse	1	Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.1 at the end of study (24 months)	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 Medication compliance (high is better)	1	Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2.1 during the follow up	1	Std. Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Dropouts	1	Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3.1 before 6 months	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.2 at end of study (24 months)	1	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

Methods	Allocation: randomisation without specification of methods. Duration: hospitalisation period (mean=36 days) plus18 month follow up.   Blinded assessments: reported   Analysis: Analysis of variance and analysis of covariance, correlation coeficients, chi‐square and t‐tests. ITT not used.   No of centres: 1   Country: USA
Participants	Diagnosis: current bipolar disorder mood episode ‐ mania (DSM‐III)   N: 26 (experimental group n=15; control group n=11)   Mean (sd) age: 32.3 (15.4)   Gender: 14 (%) female   Race: 17 white   Setting: Payne Whitney Clinic, New York, USA.   History: 10 of them had had no previous episodes, 8 had had 1 or 2 previous episodes and 4 had had 3 or more previous episodes.
Interventions	1 ‐ Standard multimodal hospital treatment (fixed drug regimen ‐ lithium 600‐2100mg/day, tricyclic antidepressants 100‐400mg/day, monoamine oxidase inhibitors 45‐90mg/day of phenelzine, antipsychotics in doses equivalent to 100‐800mg/day of chlorpromazine).   2 ‐ Standard multimodal hospital treatment plus the psychosocial family intervention for patient with his family Inpatient Family Intervention ( at least 6 45 minutes to 1 hour‐ manualized psychoeducational plus systemic‐dynamic interventions sessions, conducted by 2 therapists.
Outcomes	USED:   1 ‐ Dropouts at discharging   2 ‐ Improvement clinically significant at discharge, 6 months.   NOT USED:   2 ‐ Improvement clinicaly significant at 18 months.   3 ‐ Family functioning measures
Notes	26 patients were randomised but 5 withdrew and were not included in analysis
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	Allocation: randomised, according to an algorithm that took into account prior admissions, prehospital treatment compliance and level of functioning; no information about allocation concealment.   Blindness assessments: not reported.   Duration: 11 months.   Analysis: t‐tests and likelihood ratios; no ITT analysis.   No of centres: not mentioned   Country: USA
Participants	Diagnosis: bipolar disorder utilizing the SADS .   N: 42 (19 patients and their partners designated for esperimental group and 23 patients for control group).   Age: ranging between 21 ‐ 65 years (patients' average age was 47.7 years). Gender: distribution of patients was almost the same.   Setting: not mentioned.   History: Married or living with other of the opposite sex for at least 6 months (average of 17 years); no mention of illness duration or number of episodes.
Interventions	1 ‐ Structured Marital Psychoeducational Intervention:   25 sessions for the couple plus standard medication treatment for the bipolar patient (mood stabilizers, antidepressants and antipsychotics).   2 ‐ Standard medication (bipolar patient), no marital psychoeducation.
Outcomes	USED:   1 ‐ Dropouts at month 11   NOT USED:   2 ‐ Patients' symptoms (SADS‐C and BPRS) ‐ means without sd   3 ‐ Overall functioning ( SAS and GAS) ‐ means without sd   4 ‐ Adherence to medication treatment (Scale developed by the authors) ‐ means without sd
Notes	Forty six patients have fullfiled inclusion criteria and signed informed consent, but 4 of them dropped out before randomisation. So we have considered only 42 participants.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	Allocation: randomisation procedure using random number table. Randomisation sequence was concealed until assignments had been made. Duration: 9 months for therapy and 2 years of total follow up. Blinded assessments: reported. Analysis: Survival analysis using Kaplan‐Meier product‐limit formula and Cox proportional hazards models when incorporating covariates; mixed analysis of variance; ITT for survival analysis.   No of centres: not clear.   Country: USA
Participants	Diagnosis: bipolar disorder (post‐episode phases) DSM‐III‐R (SCID‐P). A posteriori confirmation by DSM‐IV showed that two patients of the sample were bipolar II while the others (99) were bipolar I.   N: 101 (experimental group n=31; control group n=70).   Age: 18‐62 years (mean 35.6 sd 10.2).   Gender: 64 women and 37 men.   Setting: Family interventions performed at patients' homes.   Race: not informed.   History: mania, depression or mixed episode within the past 3 months, living with or in regular contact with a caregiving family member, willingness to take medication, english speaking. Excluded disability or neurologic disorder, alcohol or substance abuse disorder within the previous 6 months.
Interventions	1 ‐ Family‐Focused Therapy: 21 sessions (psychoeducation, communication enhancement training, problem‐solving skills training; 12 weekly, 6 biweekly, 3 monthly ‐ one‐hour). 2 ‐ Crisis Management: 2 one‐hour home based sessionsof family psychoeducation (within 2 months after entry) and more crisis intervention sessions during the rest of 9 months if necessary. All patients in both groups were receiving pharmacologic treatment (mood‐stabilizers and/or antipsychotics, antidepressants).
Outcomes	USED:   1 ‐ Relapse at the end of treatment (24 months) ‐   symptoms: SADS‐C (ranging from 1 (absent) to 7 (very extreme).   2 ‐ Medication compliance during follow up: all available compliance data and laboratory results were assembled into a dosier and assigned a 3‐point adherence rating ‐ 1 is fully nonadherent, 2 is partially nonadherent and 3 is fully adherent.   3 ‐ Dropouts   NOT USED:   4 ‐ Rehospitalization   5 ‐ Time to relapse ‐ survival analysis
Notes	We have considered as relapsed the patients that have have been computed as dropouts during the study duration.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Methods	Allocation: randomisation in a 2:1 formula, without mention about concealment procedures.   Blinded assessments: reported.   Duration: 6 months   Analysis: chi‐squared test, Students' t test, Mann‐Whitney U test, mixed analysis of variance.   No of centres: 1   Country: Spain
Participants	Diagnosis: bipolar I and II disorder by SCID‐DSM‐IV, stabilized for at least 3 months (the participants of psychosocial intervention were the relatives and not the patients).   N: 45 (30 in experimental group and 15 in control group).   Age: Means (SD), for bipolar patients in experimental group 34.9(11.56) and in control group 35.9(7.52); for caregivers in experimental group 50.20(10.37) and in contrl group 45.07(16.25).   Gender: for bipolar patients in experimental group 12 women and 18 men and for control group 10 women and 5 men; for caregivers in experimental group 25 women and 5 men and for control group 9 women and 6 men.   Race: not mentioned.   Setting: hospital   History: Age of onset in means (SD) for experimental group 24.23((7.87) and for control group 23.00(7.17). Manic episodes 2.00(2.78) and 2.60(2.56) respectively. Hypomanic episodes 2.24(3.81) and 2.71(3.15) respectively. Depressive episodes 4.69(4.51) and 5.64(2.95) respectively. Mixed episodes 0.67(1.12) and 1.47(2.36) respectively. N of hospitalizations 1.67(1.32) and 2.36(1.78) respectively.
Interventions	1‐ Psychoeducational Family Intervention: 12 psychoeducational 90‐min weekly group sessions taking place at hospital. In each group there were a maximum of 10 patients' relatives. The psychoeducation was structured. All patients (experimental and control) received standard pharmacologic treatment . 2 ‐ The relatives of control group received no psychoeducational intervention. The psychoeducational intervention was administered only for caregivers without the presence of patients.
Outcomes	USED:   1 ‐ Relationships in the family invironment ‐ Cohesion, Expressiveness, Conflict ‐ (Family Environment Scale). 2 ‐ Dropouts of caregivers.   NOT USED:   3 ‐ Caregivers' knowledge of bipolar disorder (Bipolar Disorder Knowledge Questionnaire). 4 ‐ Caregivers' burden (Social Behavior Assessment Schedule).
Notes	Only the relationships in the family invironment was included in this review because it seems to us to be the one with direct clinical relevance.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	Allocation: randomisation without methods specification.   Blinded assessments: not reported.   Duration: 12 months.   Analysis: Analysis of Variance for repeated measuresand Wilcoxon test.   N of centres: 1   Country: Netherlands
Participants	Diagnosis: bipolar patients through DSM‐III‐R criteria, by two independent psychiatrists.   N: 39 (19 in experimental group and 20 in control group).   Age: Experimental group (partners) ‐ mean 42 (sd 12); control ‐ 56 (sd 8). Experimental group (patients) ‐ 44 (sd 11); control ‐ 55 (sd 8).   Gender: distribuition not mentioned.   Race: not mentioned.   Settingnot mentioned.   History: onset of illness in means (SD) of age for experimental group 26(10) and control 36(10). Duration of illness 18(10) years and 18(11) years respectively. Number of hospitalizations 4.4(4.5) and 4.4(2.8) respectively.
Interventions	1 ‐ Psychoeducation: Experimental group: requested to fill questionaires and take part in 5 sessions of structured psychoeducational intervention regarding the information about desease and medication plus practical advice.   2 ‐ Control group: only requested to fill questionaires. All patients were in pharmacological treatment with lithium.
Outcomes	USED:   1 ‐ Patients' total symptoms (SCL‐90‐total), before/after intervention. 2 ‐ Patients' anxiety (subscale SCL‐90‐Anxiety and Trait Anxiety Inventory), before/after intervention. 3 ‐ Medication compliance (compliant or not ‐ non compliance defined as a difference of more than 0.3mmol/l in serum lithium levels without changing the medication). 4 ‐ Rehospitalizations.   5 ‐ Dropouts   NOT USED:   6 ‐ Partners' knowledge of the illness.   7 ‐ Partners' social strategies.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Study	Reason for exclusion
Anderson 1986	The outcomes assessed were not clinical outcomes directly related to bipolar disorder ie they did not examine the impact of interventions on the disorder itself.
Catanzaro 1973	The study was not a RCT. It was a theoretical explanation of a type of family therapy.
Davenport 1977	The study was not a RCT.
Fitzgerald 1972	The study was not a RCT.
Fristad 2002	The outcomes considered in this study were not of direct clinical relevance. No data were presented on the possible impact of the intervention on the disorder.
Honig 1995	The study was not a RCT.
Kim 2004	This was not an primary RCT. Data were collected from two others studies, one of them included in the present review (Miklowitz 2003).
Miklowitz 2000	This paper reports on one year follow‐up data from the study by Miklowitz 2003, which is included in this review.
Simoneau 1999	This presents some results from the study reported by Miklowitz 2003, which is included in this review.

PERMALINK

Family interventions for bipolar disorder

Luis Justo

Bernardo Garcia de Oliveira Soares

Helena Calil

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Background

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Search methods for identification of studies

Data collection and analysis

Results

Description of studies

Risk of bias in included studies

Effects of interventions

Discussion

Authors' conclusions

Implications for practice.

Implications for research.

What's new

History

Acknowledgements

Data and analyses

Comparison 1. Family interventions versus No intervention.

1.1. Analysis.

1.2. Analysis.

1.4. Analysis.

1.5. Analysis.

1.6. Analysis.

1.7. Analysis.

1.8. Analysis.

1.9. Analysis.

1.10. Analysis.

Comparison 2. Psychoeducation for caregivers versus No intervention.

2.1. Analysis.

2.2. Analysis.

Comparison 3. Marital psychoeducation for partners versus No intervention.

3.1. Analysis.

3.2. Analysis.

3.3. Analysis.

3.4. Analysis.

3.5. Analysis.

Comparison 4. Multifamily psychoeducation group intervention versus No intervention.

4.1. Analysis.

4.2. Analysis.

4.3. Analysis.

Comparison 5. Couples psychoeducation for patient + partner versus No intervention.

5.1. Analysis.

Comparison 6. Problem centered systems therapy for the family versus Multifamily psychoeducational group therapy.

6.1. Analysis.

6.2. Analysis.

Comparison 7. Family intervention versus Individual intervention (subgroup analysis).

7.1. Analysis.

7.2. Analysis.

7.3. Analysis.

7.4. Analysis.

Comparison 8. Crisis management versus Family focused therapy (subgroup analysis).

8.1. Analysis.

8.2. Analysis.

8.3. Analysis.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Clarkin 1990.

Clarkin 1998.

Goldstein 1996.

Miklowitz 1996.

Reinares 2004.