Eur Arch Psychiatry Clin Neurosci (2007) 257:23–30 DOI 10.

1007/s00406-006-0677-z

ORIGINAL PAPER

Giuseppe Carrà Æ Cristina Montomoli Æ Massimo Clerici Æ Carlo Lorenzo Cazzullo

Family interventions for schizophrenia in Italy: randomized

controlled trial

Received: 28 April 2006 / Accepted: 18 July 2006 / Published online: 17 August 2006

j Abstract Objective To evaluate the effectiveness when baseline high EE was again predictive of pa-
of multiple group family treatment for Schizophrenia. tient’s admission and relatives were more vulnerable
Method Relatives were randomly provided with an to objective burden. Baseline illness severity variables
informative programme (n = 50), or allocated to re- predicted a number of medium and long-term poor
ceive an additional support programme (n = 26). clinical outcomes. Conclusions Although family psy-
Patients did not attend the programme to overcome choeducation has been tested in a wide range of An-
cultural and organizational implementation barriers. glo-Saxon settings, there remains need to assess
The 12 and 24 months clinical and family outcomes outcomes more internationally. Effective family
were assessed. Results Patients’ compliance with interventions for people with schizophrenia probably
standard care was greater at 12 months in the more require continued administration of key-elements or
intensive behavioural management group over a ongoing informal support to deal with the vicissitudes
control group receiving treatment as usual (TAU) of illnesses.
(n = 25). A reduction in levels of expressed emotion
(EE), significantly more frequent in those receiving j Key words schizophrenia Æ psychoeducation Æ
the additional support programme than just the family Æ expressed emotion Æ randomized controlled
informative, occurred after treatment completion. trials
Other clinical and family outcomes did not differ.
However, treatment benefits declined at 24 months,

Introduction
G. Carrà (&)
Department of Mental Health Sciences The influence of the family emotional climate on the
Royal Free and University College Medical School course of schizophrenia has been empirically and
48 Riding House Street
London W1W 7EY, UK
repeatedly demonstrated over the past four decades
Tel.: +44(0)20/7679-9428 since Rutter and Brown’s (1966) pioneering work. A
Fax: +44(0)20/7679-9426 number of observational studies have indicated that
E-Mail: g.carra@ucl.ac.uk the Expressed Emotion (EE) index, as rated by the
C. Montomoli Camberwell Family Interview (CFI-Vaughn and Leff
Department of Health Sciences 1976), is predictive of relapse after hospital discharge
Section of Medical Statistics (e.g., Vaughn et al. 1984). The predictive ability and
University of Pavia
Via Bassi 21 the clinical utility of the family EE have been dem-
27100 Pavia, Italy onstrated in a variety of international community
M. Clerici settings (Bebbington and Kuipers 1994a, b; Butzlaff
Department of Psychiatry and Hooley 1998; Wearden et al. 2000), but some
San Paolo’s Hospital Medical School individual studies report different results in Northern
University of Milan (McCreadie and Phillips 1988) and Southern Europe
Via A. Di Rudinı̀ 8
20144 Milan, Italy
(Montero et al. 1992) and in ethnic minority popu-
lations in the United States (Kopelowicz et al. 2002).
EAPCN 677

C.L. Cazzullo Æ G. Carrà Æ M. Clerici As a result, a number of family intervention strategies

Association for Research on Schizophrenia (ARS)
Via Tamagno 5 have been developed to augment standard care for
20124 Milan, Italy schizophrenia by reducing the level of EEs within
24

families (Mari and Streiner 1994; Pitschel-Walz et al. Methods

2001; Pilling et al. 2002), thereby improving clinical
outcomes as well as the burden of care (Dixon et al. j Setting
2000; Pharoah et al. 2003). Family psychoeducation
has been widely tested in a range of Anglo-Saxon The study was carried out in a non-profit, family advocacy and
support agency, the Association for Research on Schizophrenia
settings (Bustillo et al. 2001) and a few from China (ARS), which is funded by a charity (Legrenzi Foundation) in
(Xiong et al. 1994; Ram et al. 2003; Chien and Chan Milan, Italy. Lombardy Health System encourages non-statutory
2004). However some evidence from Spanish-speak- charities, funded by the National Health Service (NHS), to com-
ing immigrants (Telles et al. 1995) and Dutch (Lins- plement existing teams by providing not otherwise available
zen et al. 1996) samples suggest that cultural treatments. Due to organizational and cultural barriers to patient’s
participation ARS provides programmes for relatives of people with
differences may be an impediment to successfully schizophrenia, in the absence of the latter. NHS community mental
applying existing family therapy interventions to di- health centres in the metropolitan catchment area refer the rela-
verse cultural groups. Furthermore, despite the tives. The different therapeutic options consist of two elements. The
extensive documentation of its basic benefits, family first involves weekly meetings with an information group (IG)
composed of 16–18 relatives for 24 sessions (1.75 h per session)
psychoeducation is still rarely offered because of large using an informative approach. Contents and goals are mainly
attitudinal, knowledge, practical, and systemic derived from the model of relatives groups (Leff et al. 1989) but the
implementation obstacles (Dixon et al. 2001). The preliminary in-home individual family sessions. Curricula include:
World Schizophrenia Fellowship (1998) has delin- aetiology, positive symptoms, negative symptoms, mood disorders,
problem behaviours, medical and psychiatric treatment, denial and
eated the core components of a successful family non-compliance, interpersonal and social issues, relationship with
intervention but the minimal ingredients are still family, education, independence and dependence, resources and
uncertain (McFarlane et al. 2003) and it is worthwhile benefits. Educational tools include lectures, videos and leaflets. The
to look for selected combinations, which can be second element comprises weekly meetings for 48 sessions (1.5 h
realistically implemented and accepted within stan- per session) over 2 years with a support group (SG), made up of 8–
9 relatives who have previously attended the IG. The general
dard care. In Italy, since the deinstitutionalization of character of the SG programme can be summarized as consisting of
the ‘70s (de Girolamo and Cozza 2000), the key issue two components that roughly correspond to the phases of the
in implementing effective treatment systems for se- group. The first phase involves training on communication and
vere psychiatric disorders has been balancing com- coping skills, stress identification and management, and multiple
family group-based problem solving, basically derived from the
munity- and hospital-based mental health care second stage of the psychoeducational multiple family group ap-
(Thornicroft and Tansella 2004). There has been an proach (McFarlane et al. 2002). This usually occurs during the first
increasing acknowledgment of the importance of year. The second phase emphasizes mutual support and consists of
families in terms of the care they give, their thera- deliberate efforts to mould the group into a social network that can
persist for an extended period and satisfy family needs for social
peutic potential and the burden they carry, and as contact, support, and ongoing monitoring. Expansion of the fam-
political lobbying groups (Cazzullo et al. 1994). Al- ilies’ social networks occurs through problem solving, direct
though the predictive ability of the EE index has been emotional support, and out-of-group socializing, all involving
confirmed in an Italian urban setting (Bertrando et al. members of different families in the group. Both the IG and SG
1992), family interventions have been implemented programmes are co-led by two specifically trained psychiatrists not
involved in patients’ community standard care. Different clinicians
slowly within mental health services. This has been lead the different types of intervention. Greater details about the
due to an under appreciation on the part of mental interventions used are given in Cazzullo et al. (1989).
health care providers of the utility and importance of
this treatment approach and to some cultural barri-
ers—stigma and sense of hopelessness—to patient’s j Participants
participation at the level of the individual consumer
and members of his or her family. The main aim of From those who had been referred to ARS consecutively from 1995
this study was to carry out a randomized controlled to 2000 (n = 320), relatives were selected with the following
inclusion criteria (n = 205):
trial to assess the medium (12 months) and long-term
outcomes (24 months) of two programmes of family • they were living with someone suffering from schizophrenia and
intervention for the care of schizophrenia, compared had not attended family groups or other support services before
the study intervention;
to the standard community mental health care in
Italy. The primary measures of outcome were hospital • the patient was clinically stable (having had no psychiatric
hospitalization or any relapse for six months prior to study
admission, relapse, compliance with standard com- entry) and was not receiving any psychosocial or rehabilitative
munity care, and employment status. Outcome was treatment other than standard care;
also assessed measuring relative’s subjective burden • the patient did not have a primary diagnosis of alcohol or drug
as well as the objective one regarding the patient’s dependence or organic disease.
self-sufficiency and social functioning. Finally, there is Fifty relatives were randomized, using a random numbers table,
the need to study correlates of outcomes in a multi- to enter the study to complete only the information group pro-
dimensional perspective (Hofer et al. 2006), thus the gramme (IG). By the same method of random sequence generation
26 relatives were allocated to receive subsequent additional treat-
secondary aim was to evaluate the predictive utility of ment through the SG programme (IG + SG). Both relatives and
baseline patients’ demographic and clinical assess- clinicians in the IG groups programme were blind as to successive
ment, and the relatives’ EE, in relation to outcome. participation to the SG. Finally, a further group of relatives from
 25

those on the agency waiting list (i.e., those referred, but for which or moved out of Lombardy, one committed suicide and two refused
available resources did not allow immediate provision) were as to be re-interviewed), giving a resulting overall sample of 95 rela-
randomly allocated to a control group and the corresponding pa- tives. However, since the patients’ data regarding hospital admis-
tients received only treatment as usual (TAU). Participants were sion, relapse, compliance with standard community care and
intentionally allocated in unequal numbers to IG or IG + SG and current employment were available from community-based service
TAU groups with a randomization ratio 2:1:1. This decision was managers also for drop-outs, an intention-to-treat analysis was
taken in order to focus on the basic intervention effectiveness and applied to that extent in both follow-up years. The inability to
to limit costs of the trial. Allocation concealment was ensured by follow up relatives’ drop-outs at 24 months rendered the intention
the external involvement of a statistician (C.M.), who was not in- to treat methodology problematic by this time point. Therefore, all
volved in enrolling participants, and was responsible for the the relatives entered the statistical analysis with the burden out-
method of sequence generation. In total, 101 relatives agreed to come variables at 12 months but drop-outs were excluded at
participate, gave informed consent and completed the assigned 24 months, adopting analysis per-protocol as the most applicable
treatment as appropriate. Both family programmes involved only design.
one relative from each patient’s family, and all patients received
just standard care, which entailed key worker’s management and
consistent pharmacological interventions monitored by a consul- j Statistical analysis
tant psychiatrist in community mental health centres of the Milan
metropolitan area. We used regional unpublished data for the power calculation.
About one-quarter of patients suffering from schizophrenia are
admitted to hospital each year in the Milan metropolitan area,
j Measures and procedures actually confirmed by recent published figures in Northern Italy
(Ruggeri et al. 2004). We estimated that 25 participants per group
Community-based service managers were contacted personally by would yield 75% power to detect a 10% difference in admission
research assistants not involved in the treatment to check the fol- rates as primary outcome, at a two sided 5% level of significance.
lowing criteria: (a) patients’ DSM-IV diagnoses of schizophrenia Participants selected on the basis of inclusion criteria were com-
(American Psychiatric Association 1994); (b) current satisfactory pared to non-participants as regards patients’ main baseline soci-
functioning, as measured by a Global Assessment Scale-GAS score odemographic and clinical characteristics, as well as on family
of 30 or more (Endicott et al. 1976); (c) compliance with standard variables using one-way ANOVA for continuous variables, and
care, with a specifically designed 3 point scale defining non-com- Fisher’s exact test for nominal variables. The equivalence of IG,
pliance as a rating of 3 (refusal of every proposed treatment), and IG + SG and TAU groups in terms of participating patients’ and
(d) consistency of prescribed pharmacological treatment, with all relatives’ baseline characteristics was assessed, using the same tests.
but 3 patients receiving standard doses (300–1,000 mg chlor- Fisher’s exact test for nominal variables was also used to compare
promazine equivalents). In addition, at induction each relative was the changes from baseline high EE between the IG and IG + SG
given a standardized questionnaire on clinical and social charac- groups. The equivalence of treatment groups for clinical and family
teristics of patient and family. Clinicians from the community outcomes at 12 and 24 months was first tested by Fisher’s exact
settings supplied missing information on treatment variables. Fi- test. However, multiple logistic regression models with a stepwise
nally, the relatives’ EE was evaluated by the Camberwell Family procedure were used to analyze the association between clinical and
Interview-CFI (Vaughn and Leff 1976). Every interview was tape family outcomes at 12 and 24 months and the following indepen-
recorded. The two evaluators had been trained by Dr. C.E. Vaughn. dent variables: treatment group, baseline demographic and clinical
Relatives were defined as high EE if they made six or more critical variables, and EE. The clinical dependent variables in the regres-
comments, expressed hostility, or were rated as four or more on the sion models were dichotomized according to the definitions pro-
EOI in the course of the interview. The latter is in accordance with vided. The family outcome variables were analyzed yielding odds
the Italian field study on predictive value (Bertrando et al. 1992) ratios with respect to improved or unchanged outcome. The sta-
and not with the classical (>3) scoring criteria (Leff and Vaughn tistical package Stata 7.0 (Stata Corporation 2001) was used for
1985). Positive remarks (a frequency count) and warmth (a 6-point these analyses, with a level of significance of 0.05.
scale: 0–5) were rated as well. At the end of the assigned pro-
gramme, the relatives’ EE was evaluated again to determine whether
the levels had indeed been reduced by family intervention. Follow-
up assessments were carried out by research assistants blind about Results
the treatment assigned. Each community-based service manager
was contacted 12 and 24 months after the relatives completed their j Descriptive analysis
assigned programme to fill in a follow-up questionnaire that re-
corded: (a) hospital admission; (b) the patient’s relapse by means of Comparison of trial participants and non-participants
an operational definition that included a GAS score of less than 30;
(c) compliance with standard community care, where improvement on all measures used in the study showed no significant
consisted of scores lower than at baseline; and (d) employment in differences on sociodemographic, clinical, or family
the past 12 months. At the same time periods, the relatives were variables at baseline. Patients’ and family main baseline
contacted to ascertain the objective burden in terms of patients’ (i) sociodemographic and clinical characteristics did not
self-sufficiency and (ii) social functioning in the past year and their
own subjective burden. Measure of elements of objective burden differ between the IG, IG + SG and TAU groups (Ta-
included disruption of household routines, disruption of relatives’ ble 1). The overall mean age of patients was less than
leisure time and career (self-sufficiency), and strain on family 30 years, and 28% were women. Furthermore, medium
relationships, and reduction of social support (social functioning). levels of education did not support consistent regular
Subjective burden was measured in terms of emotional exhaustion
referring to feelings of being emotionally overextended and de-
employment status (21/101) and only a few (9/101) had
pleted of one’s emotional resources. These outcomes were mea- stable intimate relationships, with most patients still
sured on a 2 point scale (0 = improved/unchanged, 1 = worsened). living with their family of origin. The clinical profile
Objective burden ratings were checked through a contact with the corresponded to that usually reflected in studies of this
relevant community-based service manager. Inter-observer reli- type in terms of onset age, duration of illness, and
ability was evaluated using Cohen’s kappa with kappa values
ranging from 0.82 for patient’s self-sufficiency and 0.91 for social number of previous hospitalizations. Most of the key-
functioning. At 24 months, six relatives were lost to follow up relatives were parents (80/101) with a corresponding
(three IG, two IG + SG, and one control relative as three left home mean age, gender bias (71% of mothers overall) and
26

Table 1 Baseline comparison of

patients’ and relatives’ characteristics IG IG + SG TAU P
N = 50 N = 26 N = 25

PATIENTS
Age: Mean (SD), y 29.9 (8.9) 29.6 (5.8) 29.9 (10.6) 0.99a
Gender: Male No (%) 35 (70) 22 (85) 16 (64) 0.23b
Education: Mean (SD), y 12.2 (3.6) 12.3 (2.4) 10.9 (2.9) 0.19c
Ordinary employed, No. (%) 12 (24) 3 (12) 6 (24) 0.45b
Married/cohabiting: No. (%) 7 (14) 0 (0) 2 (8) 0.18b
Residence: No. (%)
In parental home 40 (80) 24 (92) 17 (68)
In conjugal home 5 (10) 0 (0) 2 (8)
Alone 5 (10) 2 (8) 6 (24) 0.16b
Onset age: Mean (SD), y 21.1 (7.7) 18.7 (4.2) 19.9 (6.4) 0.33d
Duration of illness: Mean (SD), y 9.6 (8.1) 11.3 (7.6) 10.3 (9.2) 0.68e
Previous hospitalizations: Mean (SD), No. 2.7 (3.0) 4.8 (8.0) 3.0 (4.3) 0.23f
RELATIVES
Age: Mean (SD), y 53.2 (10.7) 58.5 (8.5) 53.6 (11.5) 0.10g
Gender: No. Male (%) 17 (34) 6 (23) 12 (48) 0.20b
Education: Mean (SD), y 9.5 (3.9) 10.3 (3.7) 10.1 (4.1) 0.65h
Relative’s h per week spent in contact 40 (80) 21 (81) 17 (68) 0.43b
with the patient > 35: No. (%)
Worsened family’s social contacts after 16 (32) 8 (31) 9 (36) 0.88b
illness onset: No. (%)
High EE: No. (%) 19 (38) 10 (38) 10 (40) 1.0b
High warmth (4–5): No. (%) 14 (28) 3 (12) 1 (4) 0.34b
High EE relatives moved to low: No. (%) 9 (47) 10 (100) 0.005b

IG indicates Information Group, SG Support Group, and TAU Treatment As Usual

ANOVAs: a F = 0.01; c F = 1.69; d F = 1.15; e F = 0.36; f F = 1.34; g F = 2.4; h F = 0.43 (for all ANOVAs df = 2.98)
b
Fisher’s exact test

rate (77%) of high contact dichotomized as more than groups. With regard to the family outcome variables at
35 h face-to-face contact per week. A third of the key- the 12 and 24 month follow-up there were no signifi-
relatives reported that the family’s social contacts had cant differences reported in the relatives’ objective
worsened after illness onset. The distribution of high burden in relation to the patient’s self-sufficiency or in
EE baseline rates did not show any significant differ- the subjective burden. Interestingly, there was a sig-
ence between the three groups. This was also the case nificant difference at 24 months in the objective bur-
for the high warmth rates. However, assessing EE again den in relation to the patients’ social functioning, with
at the end of the family program for all participants, no the IG + SG group faring less well. However, negative
low EE relative became high, and there were signifi- trends in terms of proportions of worsened outcomes
cantly more frequent downward changes from baseline as compared to the past year mostly occurred within
high EE in the IG + SG (10/10) than in the IG (9/19) the first 12 months after treatment completion for all
group (P = 0.005). the groups, with a relative stabilization at 24 months.
Table 3 shows the OR and the corresponding 95%
j Follow-up at 12 and 24 months Confidence Intervals (95% CI) for the logistic regres-
sion models testing potential predictors of patients’
The univariate analysis of patients’ clinical outcomes and family outcomes. According to multivariate anal-
(hospital admission, relapse, compliance with stan- ysis, none of the independent variables were signifi-
dard community care, and current employment) over cantly associated with patient’s compliance at
12 and 24 months did not show any significant dif- 24 months or relapse at 12 and 24 months. The number
ference between the three groups (Table 2). However, of previous hospitalizations assessed at baseline was a
controlling for baseline demographic and clinical significant predictor of further hospital admissions at
variables, and EE levels the stepwise multiple logistic both 12 and 24 months and of unfavourable employ-
regression model (Table 3) showed that compliance ment status at 1-year follow-up. Relatives’ high EE be-
with standard care was significantly greater at 1-year came again predictive of readmission at 24 months,
follow-up in the IG + SG group than in the TAU one. while the duration of illness predicted the patients’
Most of the hospital admissions and relapses in each occupational status at 24 months. As regards the rela-
family program group occurred within the first year tive’s outcomes, at 12 months subjective burden ap-
after treatment ended, while compliance with standard peared to be associated per se with the number of past
care benefits of the IG + SG group compared to TAU hospitalizations at baseline, while the objective
(54% vs. 32%) was fairly reduced at 24 months. Rates dimensions of social functioning at 12 months and self-
of currently employed clients—overall between 30% sufficiency at 2-year follow-up were both associated
and 50%—slightly increased over time in all the with a high EE level at entry to the programme.
 27

Table 2 Clinical and family

outcomes up to 12- and 24 months IG IG + SG TAU Pa
after psychoeducational treatment N = 50 N = 26 N = 25
completion
PATIENTS
12 months
Hospital admission No. (%) 13 (26) 6 (23) 7 (28) 0.92
Relapse No. (%) 13 (26) 7 (27) 9 (36) 0.65
Compliance with standard community care No. (%) 14 (28) 14 (54) 8 (32) 0.07
Current employment No. (%) 22 (44) 8 (31) 12 (48) 0.40
24 months
Hospital admission No. (%) 10 (20) 4 (15) 3 (12) 0.66
Relapse No. (%) 13 (26) 9 (35) 9 (36) 0.59
Compliance with standard community care No. (%) 18 (36) 11 (42) 8 (32) 0.68
Current employment No.(%) 23 (46) 8 (31) 12 (48) 0.45
FAMILY
12 months
Objective burden related to Self-sufficiency No. (%)
Improved/unchanged 37 (74) 21 (81) 16 (64) 0.41
Worsened 13 (26) 5 (19) 9 (36)
Objective burden related to Social functioning No. (%)
Improved/unchanged 39 (78) 21 (81) 20 (80) 1.0
Worsened 11 (22) 5 (19) 5 (20)
Subjective burden No.(%)
Improved/unchanged 39 (78) 17 (65) 19 (76)
Worsened 11 (22) 9 (35) 6 (24) 0.53
24 monthsb
Objective burden related to Self-sufficiency No. (%)
Improved/unchanged 42 (89) 22 (91) 23 (96) 0.89
Worsened 5 (11) 2 (8) 1 (4)
Objective burden related to Social functioning No. (%)
Improved/unchanged 46 (98) 18 (75) 22 (92) 0.007
Worsened 1 (2) 6 (25) 2 (8)
Subjective burden No.(%)
Improved/unchanged 40 (85) 21 (87) 19 (79)
Worsened 7 (15) 3 (13) 5 (21) 0.76

IG indicates Information Group, SG Support Group, and TAU Treatment As Usual

a
Fisher’s exact test
b
Number of relatives at the 24 months follow-up for IG was 47; for IG + SG 24; and for TAU 24

Table 3 Logistic regression models

for patients’ and family outcome Independent variable OR P 95% CI
variables
PATIENTS
12 months
Hospital admission Number of hospitalizations 1.20 0.012 1.04–1.39
Compliance with standard care IG + SG group versus TAU group 2.80 0.027 1.12–7.03
Current employment Number of hospitalizations 0.83 0.034 0.70–0.98
24 months
Hospital admission Number of hospitalizations 1.26 0.008 1.06–1.49
High EE 3.05 0.033 1.09–8.5
Current employment Duration of illness 0.93 0.012 0.87–0.98
FAMILY
12 months
Subjective burden Number of hospitalizations 1.12 0.025 1.01–1.23
Social functioning High EE 3.37 0.017 1.24–9.14
24 monthsa
Self-sufficiency High EE 5.42 0.046 1.03–28.5
Social functioning IG + SG group versus TAU group 7.55 0.007 1.72–33.2

IG indicates Information Group, SG Support Group, and TAU Treatment As Usual

a
Number of relatives at the 24 months follow-up for IG was 47; for IG + SG 24; and for TAU 24

more, as knowledge about empirical advantages of

Discussion family psychoeducation interventions may carry al-
most no weight in convincing clinicians to change
Although family psychoeducation has been tested in a their attitudes toward families and adopt them within
wide range of national settings, there remains a need standard practices (McFarlane et al. 2001), efforts are
to assess outcomes more internationally. Further- needed to identify the least intensive and smallest
28

‘‘dose’’ of family psychoeducation in terms of cost- focal format does not seem comparable to the bifocal
effectiveness (McFarlane et al. 2003). This study psychoeducational interventions as regards effective-
sought to examine the effectiveness of two pro- ness (Hornung et al. 1999) and more importantly in
grammes of family intervention for the care of terms of cost effectiveness if a relatively small extent
schizophrenia in a non Anglo-Saxon cultural context of psychosocial intervention, but including the patient
and with some modifications in content and form to (Pitschel-Walz et al. 2006), can reach positive out-
overcome obstacles related to their delivery on a comes. Unsurprisingly, variables that may be re-
regular basis in routine services. The family pro- garded as indicators of baseline severity of illness
grammes of the current study have already been (Üçok et al. 2006) predict unfavourable medium and
shown to improve knowledge about schizophrenia long term clinical outcomes: hospital admission is
(Cazzullo et al. 1989), and study findings show that significantly associated with number of previous
they are effective in lowering overall EE levels in hospitalizations, which predict also poor employment
relatives attending the information programme only status at 12 months, significantly associated in the
and more in those receiving the additional psycho- long term with baseline duration of illness. Regarding
educational multiple family group. However, the basis family variables, personal emotional exhaustion seem,
for evaluating the effectiveness of group family in any case, associated with the essential factor of
treatment has to comprise readmission and relapse illness severity as measured by baseline admissions
rates during intervals starting from the end of treat- (Baronet 1999). However, only high EE baseline rel-
ment, and effects on family burden (Pilling et al. atives cannot cope with the objective dimensions of
2002). Despite neither of the two programs showing burden suggesting that the two dimensions are actu-
any advantage in terms of relapse, readmission, or ally related and dependent on relatives’ appraisal of
employment in the medium or the long term, com- the patient condition rather than on his/her clinical
pliance with standard care was significantly greater at severity, thus emphasising the special needs of
1-year follow-up for patients of families having at- continuing treatment for this subpopulation (Sca-
tended the IG + SG than in the standard care group. zufca and Kuipers 1998; Möller-Leimkühler 2005). It
Behavioural management and multiple-family ap- is questionable whether these just partially positive
proaches seem to improve patient’s treatment results can be explained by the cultural, non Anglo-
adherence in his/her absence although not affecting Saxon context as would be suggested by similar evi-
relapses and hospital admissions. However, this effect dence from Greece (Tomaras et al. 2000) and Span-
appears to decline over time, being absent 2 years ish-speaking immigrants (Telles et al. 1995), with
after the end of the program, and at the same time comparable recruitment of relatives of patients in
high EE levels become predictive of hospital admis- remission, and receiving care in the community.
sion again (Bebbington and Kuipers 1994a). This Other possible explanations relate to the specific
confirms the mediating effect of reductions in EEs. characteristics of the programmes that aimed to
Furthermore, relatives attending the psychoeduca- overcome organizational and cultural barriers but
tional multiple family group report at the same have excluded some potentially basic components of
24 month follow-up greater objective burden related successful family treatments for schizophrenia: par-
with patients’ social functioning, suggesting that ticipation of the patient, recruitment at a relapse
previously learnt skills might be a risk factor if not episode, full inclusion in routine mental health ser-
regularly boosted. As a whole our findings seem to vices (McFarlane et al. 2003). The study program was
draw a picture where relatives attending an intensive carried out by a non-profit agency, which was not part
and prolonged programme, that over 3 years has of the statutory mental health services providing care
combined educational and behavioural elements, can for the patients. Access was based on referral by
beneficially affect the attitude of patients towards care community staff and such recruitment could have
in the medium term, though not their main clinical affected the generalizability of the findings. The rel-
outcomes. The same relatives seem more vulnerable ative’s motivation to accept family intervention not
to some form of objective burden in the long-term otherwise available could be similar to that in early
when the predictive value of high EE for relapse be- family programs (Grella and Grusky 1989) and could
came evident again. Neither the subjective nor have biased the results. On the other hand, the
objective burden of care are effectively reduced, de- involvement of a number of clinicians providing
spite a lowering in EE levels, even within the multiple standard care in different community mental health
family group treatment. This has been similarly found centres was a potential source of confounding, in spite
in some recent trials (McDonell et al. 2003; Stengard of our attempts to minimize this through the cross-
2003). However, the attempt of the program to over- checking procedure described. Our results should be
come consumer’s barriers to implementation (Dixon interpreted with caution because of these study limi-
et al. 2001)—by excluding the patient from the family tations. However, the impact of complex models of
programme—does not seem successful, confirming family treatments within non Anglo-Saxon cultures
the need for fidelity to the minimal components of a remains unclear. Further international research is
family intervention (McFarlane et al. 2003). The uni- needed assessing higher fidelity implemented family
 29

interventions in a wider range of cultures. Further- 12. De Girolamo G, Cozza M (2000) The Italian Psychiatric Reform.
more, the current claim for new and differently de- A 20-Year Perspective. Int J Law Psychiatry 23:197–214
13. Dixon L, Adams C, Lucksted A (2000) Update on family psy-
signed family programmes seems appropriate. The choeducation for schizophrenia. Schizophr Bull 26:5–20
enthusiasm for the ‘‘second generation’’ interventions 14. Dixon L, McFarlane WR, Lefley H, Lucksted A, Cohen M, Fal-
that incorporated the advantages of each of their loon I, Mueser K, Miklowitz D, Solomon P, Sondheimer D
sources should not be lost at this stage, although their (2001) Evidence-based practices for services to families of
people with psychiatric disabilities. Psychiatr Serv 52:903–910
effect might decline over time. Effective care for 15. Endicott J, Spitzer RL, Fleiss JL, Cohen J (1976) The Global
people with schizophrenia probably requires a ‘‘third Assessment Scale: a procedure for measuring the overall severity
generation’’ of family interventions that would of psychiatric disturbance. Arch Gen Psychiatry 33:766–771
administer the key-elements of training and man- 16. Grella CE, Grusky O (1989) Families of the seriously mentally ill
and their satisfaction with services. Hosp Commun Psychiatry
agement over time (Liberman and Liberman 2003). 40:831–835
What happens after a family has completed a family 17. Hofer A, Rettenbacher MA, Widschwendter ChG, Kemmler G,
psychoeducation programme? Families of patients Hummer M, Fleischhacker WW (2006) Correlates of subjective
with long-term problems and disability may need and functional outcomes in outpatient clinic attendees with
ongoing support and problem solving skills to deal schizophrenia and schizoaffective disorder. Eur Arch Psychiatry
Clin Neurosci 256:246–255
with the vicissitudes of illnesses. Programmes in 18. Hornung WP, Feldmann R, Klingberg S, Buchkremer G, Reker T
informal settings such as ongoing family support (1999) Long-term effects of a psychoeducational psychothera-
groups (Lefley 2001), open-ended multifamily group peutic intervention for schizophrenic outpatients and their key-
structure for families in need (McFarlane et al. 2002), persons-results of a five-year follow-up. Eur Arch Psychiatry
Clin Neurosci 249:162–167
offers of continuity in the NAMI support and educa- 19. Kopelowicz A, Zarate R, Gonzalez V, Lopez SR, Ortega P,
tional groups (Burland 1998), all represent examples Obregon N, Mintz J (2002) Evaluation of expressed emotion in
of how implementation of family psychoeducation schizophrenia: a comparison of Caucasians and Mexican-
could match long-term realities in the lives of po- Americans. Schizophr Res 55:179–186
20. Leff J, Berkowitz R, Shavit N, Strachan A, Glass L, Vaughn C
tential participants. (1989) A trial of family therapy v, a relatives group for
schizophrenia. Br J Psychiatry 154:58–66
21. Leff J, Vaughn C (1985) Expressed emotion in families: its sig-
j Acknowledgements We thank Paul Bebbington (Royal Free and nificance for mental illness. Guilford, New York
University College Medical School, London) for the critical, stim- 22. Lefley H (2001) Impact of mental illness on families and carers.
ulating and motivating comments on an earlier version of the In: Thornicroft G, Szmukler G (eds) Textbook of community
article. psychiatry. Oxford University Press, London, pp 141–154
23. Liberman DB, Liberman RP (2003) Rehab rounds: Involving
families in rehabilitation through behavioral family manage-
ment. Psychiatr Serv 54:633–635
24. Linszen D, Dingemans P, Van Der Does J, Nugter A, Scholte P,
References Lenior R, Goldstein M (1996) Treatment, expressed emotion
and relapse in recent onset schizophrenic disorders. Psychol
1. American Psychiatric Association (1994) Diagnostic and sta- Med 26:333–342
tistical manual of mental disorders, 4th edn. American Psy- 25. Mari JJ, Streiner DL (1994) An overview of family interventions
chiatric Association, Washington, DC and relapse in schizophrenia: meta-analysis of research find-
2. Baronet AM (1999) Factors associated with caregiver burden in ings. Psychol Med 24:565–578
mental illness: a critical review of the research literature. Clin 26. McCreadie RG, Phillips K (1988) The Nithsdale Schizophrenia
Psychol Rev 19:819–841 Survey. VII. Does relatives’ high expressed emotion predict
3. Bebbington P, Kuipers L (1994a) The predictive utility of ex- relapse? Br J Psychiatry 152:477–481
pressed emotion in schizophrenia: an aggregate analysis. Psy- 27. McDonell MG, Short RA, Berry CM, Dyck DG (2003) Burden in
chol Med 24:707–718 schizophrenia caregivers: impact of family psychoeducation
4. Bebbington P, Kuipers L (1994b) The clinical utility of expressed and awareness of patient suicidality. Fam Process 42:91–103
emotion in schizophrenia. Acta Psychiatr Scand Suppl 382:46–53 28. McFarlane WR, McNary S, Dixon L, Hornby H, Cimett E (2001)
5. Bertrando P, Beltz J, Bressi C, Clerici M, Farma T, Invernizzi G, Predictors of dissemination of family psychoeducation in
Cazzullo CL (1992) Expressed emotion and schizophrenia in community mental health centers in Maine and Illinois. Psy-
Italy. A study of an urban population. Br J Psychiatry 161:223–229 chiatr Serv 52:935–942
6. Burland J (1998) Family-to-family: a trauma and recovery 29. McFarlane WR, Hornby H, Dixon L, McNary S (2002) Psycho-
model of family education. New Dir Ment Health Serv 77:33–44 educational multifamily groups: research and implementation
7. Bustillo J, Lauriello J, Horan W, Keith S (2001) The psychosocial in the United States. In: McFarlane WR (eds) Multifamily group
treatment of schizophrenia: an update. Am J Psychiatry treatment for severe psychiatric disorders. Guilford, New York,
158:163–175 pp 43–60
8. Butzlaff RL, Hooley JM (1998) Expressed emotion and psychi- 30. McFarlane WR, Dixon L, Lukens E, Lucksted A (2003) Family
atric relapse: a meta-analysis. Arch Gen Psychiatry 55:547–552 psychoeducation and schizophrenia: a review of the literature. J
9. Cazzullo CL, Bertrando P, Clerici M, Bressi C, Da Ponte C, Marital Fam Ther 29:223–245
Albertini E (1989) The efficacy of an information group inter- 31. Möller-Leimkühler AM (2005) Burden of relatives and predictors
vention on relatives of schizophrenics. Int J Soc Psychiatry of burden. Baseline results from the Munich 5-year-follow-up
35:313–323 study on relatives of first hospitalized patients with schizophrenia
10. Cazzullo CL, Clerici M, Bertrando P (1994) Future strategies in or depression. Eur Arch Psychiatry Clin Neurosci 255:223–231
family research and intervention. Integr Psychiatry 10:20–23 32. Montero I, Gomez-Beneyto M, Ruiz I, Puche E, Adam A (1992)
11. Chien WT, Chan SW (2004) One-year follow-up of a multiple- The influence of family expressed emotion on the course of
family-group intervention for Chinese families of patients with schizophrenia in a sample of Spanish patients. A 2-year follow-
schizophrenia. Psychiatr Serv 55:1276–1284 up study. Br J Psychiatry 161:217–222
30

33. Pharoah FM, Rathbone J, Mari JJ, Streiner D (2003) Family 43. Telles C, Karno M, Mintz J, Paz G, Arias M, Tucker D, Lopez
intervention for schizophrenia. The Cochrane Database of S (1995) Immigrant families coping with schizophrenia.
Systematic Reviews, 3, Art. No.: CD000088. DOI: 10.1002/ Behavioral family intervention v. case management with a
14651858.CD000088 low-income Spanish-speaking population. Br J Psychiatry
34. Pilling S, Bebbington P, Kuipers E, Garety P, Geddes J, Orbach 167:473–479
G, Morgan C (2002) Psychological treatments in schizophrenia: 44. Thornicroft G, Tansella M (2004) Components of a modern
I. Meta-analysis of family intervention and cognitive behaviour mental health service: a pragmatic balance of community and
therapy. Psychol Med 32:763–782 hospital care. Overview of systematic evidence. Br J Psychiatry
35. Pitschel-Walz G, Leucht S, Bauml J, Kissling W, Engel RR (2001) 185:283–290
The effect of family interventions on relapse and rehospitaliza- 45. Tomaras V, Mavreas V, Economou M, Ioannovich E, Karydi
tion in schizophrenia-a meta-analysis. Schizophr Bull 27:73–92 V, Stefanis C (2000) The effect of family intervention on
36. Pitschel-Walz G, Bauml J, Bender W, Engel RR, Wagner M, chronic schizophrenics under individual psychosocial treat-
Kissling W (2006) Psychoeducation and compliance in the ment: a 3-year study. Soc Psychiatry Psychiatr Epidemiol
treatment of schizophrenia: results of the Munich Psychosis 35:487–493
Information Project Study. J Clin Psychiatry 67:443–452 46. Üçok A, Polat A, Çakır S, Genç A (2006) One year outcome in
37. Ram Ms, Xiang Mz, Chan Cl, Leff J, Simpson P, Huang Ms, Shan first episode schizophrenia. Eur Arch Psychiatry Clin Neurosci
Yh, Li S (2003) Effectiveness of psychoeducational intervention 256:37–43
for rural Chinese families experiencing schizophrenia-a 47. Vaughn C, Leff J (1976) The measurement of expressed emotion
randomised controlled trial. Soc Psychiatry Psychiatr Epidemiol in the families of psychiatric patients. Br J Soc Clin Psychol
38:69–75 15:157–165
38. Ruggeri M, Lasalvia A, Tansella M, Bonetto C, Abate M, 48. Vaughn CE, Snyder KS, Jones S, Freeman WB, Falloon IR (1984)
Thornicroft G, Allevi L, Ognibene P (2004) Heterogeneity of Family factors in schizophrenic relapse. Replication in Cali-
outcomes in schizophrenia. 3-year follow-up of treated pre- fornia of British research on expressed emotion. Arch Gen
valent cases. Br J Psychiatry 184:48–57 Psychiatry 41:1169–1177
39. Rutter M, Brown GW (1966) The reliability and validity of 49. Wearden AJ, Tarrier N, Barrowclough C, Zastowny TR, Rahill
measures of family life and relationships in families containing AA (2000) A review of expressed emotion research in health
a psychiatric patient. Soc Psychiatry 1:38–53 care. Clin Psychol Rev 20:633–666
40. Scazufca M, Kuipers E (1998) Stability of expressed emotion in 50. World Schizophrenia Fellowship (1998) Families as partners in
relatives of those with schizophrenia and its relationship with care: a document developed to launch a strategy for the
burden of care and perception of patients’ social functioning. implementation of programs of family education, training, and
Psychol Med 28:453–61 support. World Schizophrenia Fellowship, Toronto
41. Stata Corp (2001) Stata Statistical Software, Release 7.0. Stata 51. Xiong W, Phillips MR, Hu X, Wang R, Dai Q, Kleinman J,
Corp., College Station, TX Kleinman A (1994) Family-based intervention for schizophrenic
42. Stengard E (2003) Educational intervention for the relatives of patients in China. A randomised controlled trial. Br J Psychiatry
schizophrenia patients in Finland. Nord J Psychiatry 57:271–277 165:239–247