Tac Chocrane

Assertive community treatment for people with severe
mental disorders (Review)
Marshall M, Lockwood A
This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library
2010, Issue 3
http://www.thecochranelibrary.com
Assertive community treatment for people with severe mental disorders (Review)
Copyright © 2010 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
TABLE OF CONTENTS
HEADER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
PLAIN LANGUAGE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
AUTHORS’ CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
CHARACTERISTICS OF STUDIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
DATA AND ANALYSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Analysis 1.1. Comparison 1 ACT vs STANDARD CARE, Outcome 1 Number lost to follow up. . . . . . . . 41
Analysis 1.2. Comparison 1 ACT vs STANDARD CARE, Outcome 2 Death (all causes). . . . . . . . . . . 42
Analysis 1.3. Comparison 1 ACT vs STANDARD CARE, Outcome 3 Admitted to hospital during study. . . . . 42
Analysis 1.5. Comparison 1 ACT vs STANDARD CARE, Outcome 5 Trouble with the police. . . . . . . . . 44
Analysis 1.6. Comparison 1 ACT vs STANDARD CARE, Outcome 6 Not living independently at end of study. . . 45
Analysis 1.7. Comparison 1 ACT vs STANDARD CARE, Outcome 7 Homeless during or at end of study. . . . . 45
Analysis 1.8. Comparison 1 ACT vs STANDARD CARE, Outcome 8 Mean days per month in stable accommodation. 46
Analysis 1.9. Comparison 1 ACT vs STANDARD CARE, Outcome 9 Unemployed at end of study. . . . . . . 46
Analysis 1.10. Comparison 1 ACT vs STANDARD CARE, Outcome 10 Mental state at about 12 months. . . . . 47
Analysis 1.11. Comparison 1 ACT vs STANDARD CARE, Outcome 11 Social functioning at about 12 months. . 48
Analysis 1.12. Comparison 1 ACT vs STANDARD CARE, Outcome 12 Satisfaction with care (Client Satisfaction
Questionnnaire, low score = poor). . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Analysis 1.13. Comparison 1 ACT vs STANDARD CARE, Outcome 13 Self esteem (Rosenberg Scale, low score = poor). 49
Analysis 1.14. Comparison 1 ACT vs STANDARD CARE, Outcome 14 Quality of life (general well-being in Quality of
Life Scale, low scores = poor). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Analysis 2.1. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 1 Numbers lost to follow up. 51
Analysis 2.2. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 2 Admitted to hospital during
study. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Analysis 2.4. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 4 Trouble with the police. 53
Analysis 2.5. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 5 Not living independently at
end of study. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Analysis 2.6. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 6 Unemployed at end of
study. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Analysis 3.1. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 1 Numbers lost to follow up. . . . . . 54
Analysis 3.3. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 3 Trouble with the police (imprisonment during
study). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Analysis 3.4. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 4 Mean days per month in stable
accommodation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Analysis 3.5. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 5 Mental state at about 12 months (Brief
Psychiatric Rating Scale, high score = poor). . . . . . . . . . . . . . . . . . . . . . . . 56
Analysis 3.6. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 6 Social functioning at about 12 months (Social
Adjustment Scale, low score = poor). . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Analysis 3.7. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 7 Satisfaction with care (Client Satisfaction
Questionnnaire, low score = poor). . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Analysis 3.8. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 8 Self esteem (Rosenberg Scale, low score =
poor). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
WHAT’S NEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
HISTORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Assertive community treatment for people with severe mental disorders (Review) i
DECLARATIONS OF INTEREST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
SOURCES OF SUPPORT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
NOTES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
INDEX TERMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Assertive community treatment for people with severe mental disorders (Review) ii
[Intervention Review]
Assertive community treatment for people with severe

mental disorders
Max Marshall1 , Austin Lockwood2

1
University of Manchester, The Lantern Centre, Preston., UK. 2 School of Psychiatry and Behavioural Sciences, University of Manchester,
Preston, UK
Contact address: Max Marshall, University of Manchester, The Lantern Centre, Vicarage Lane, Of Watling Street Road, Fulwood,
Preston., Lancashire, UK. max.marshall@manchester.ac.uk. max.marshall@lancashirecare.nhs.uk.
Editorial group: Cochrane Schizophrenia Group.

Publication status and date: Edited (no change to conclusions), published in Issue 3, 2010.
Review content assessed as up-to-date: 24 February 1998.
Citation: Marshall M, Lockwood A. Assertive community treatment for people with severe mental disorders. Cochrane Database of
Systematic Reviews 1998, Issue 2. Art. No.: CD001089. DOI: 10.1002/14651858.CD001089.
ABSTRACT
Background
Assertive Community Treatment (ACT) was developed in the early 1970s as a response to the closing down of psychiatric hospitals.
ACT is a team-based approach aiming at keeping ill people in contact with services, reducing hospital admissions and improving
outcome, especially social functioning and quality of life.
Objectives
To determine the effectiveness of Assertive Community Treatment (ACT) as an alternative to i. standard community care, ii. traditional
hospital-based rehabilitation, and iii. case management. For each of the three comparisons the main outcome indices were i. remaining
in contact with the psychiatric services, ii. extent of psychiatric hospital admissions, iii. clinical and social outcome and iv. costs.
Search strategy
Electronic searches of CINAHL (1982-1997), the Cochrane Schizophrenia Group’s Register of trials (1997), EMBASE (1980-1997),
MEDLINE (1966-1997), PsycLIT (1974-1997) and SCISEARCH (1997) were undertaken. References of all identified studies were
searched for further trial citations.
Selection criteria
The inclusion criteria were that studies should i. be randomised controlled trials, ii. have compared ACT to standard community care,
hospital-based rehabilitation, or case management and iii. have been carried out on people with severe mental disorder the majority
of whom were aged from 18 to 65. Studies of ACT were defined as those in which the investigators described the intervention as
“Assertive Community Treatment” or one of its synonyms. Studies of ACT as an alternative to hospital admission, hospital diversion
programmes, for those in crisis, were excluded. The reliability of the inclusion criteria were evaluated.
Data collection and analysis
Three types of outcome data were available: i. categorical data, ii. numerical data based on counts of real life events (count data) and
iii. numerical data collected by standardised instruments (scale data). Categorical data were extracted twice and then cross-checked.
Peto Odds Ratios and the number needed to treat (NNT) were calculated. Numerical count data were extracted twice and cross-
checked. Count data could not be combined across studies for technical reasons (the data were skewed) but all relevant observations
Assertive community treatment for people with severe mental disorders (Review) 1
based on count data were reported in the review. Numerical scale data were subject to a quality assessment. The validity of the quality
assessment was itself assessed. Numerical scale data of suitable quality were combined using the standardised mean difference statistic
where possible, otherwise the data were reported in the text or ’Other data tables’ of the review.
Main results
ACT versus standard community care
Those receiving ACT were more likely to remain in contact with services than people receiving standard community care (OR 0.51,
99%CI 0.37-0.70). People allocated to ACT were less likely to be admitted to hospital than those receiving standard community care
(OR 0.59, 99%CI 0.41-0.85) and spent less time in hospital. In terms of clinical and social outcome, significant and robust differences
between ACT and standard community care were found on i. accommodation status, ii. employment and iii. patient satisfaction. There
were no differences between ACT and control treatments on mental state or social functioning. ACT invariably reduced the cost of
hospital care, but did not have a clear cut advantage over standard care when other costs were taken into account.
ACT versus hospital-based rehabilitation services
Those receiving ACT were no more likely to remain in contact with services than those receiving hospital-based rehabilitation, but
confidence intervals for the odds ratio were wide. People getting ACT were significantly less likely to be admitted to hospital than
those receiving hospital-based rehabilitation (OR 0.2, 99%CI 0.09-0.46) and spent less time in hospital. Those allocated to ACT were
significantly more likely to be living independently (OR (for not living independently) 0.19, 99%CI 0.06-0.54), but there were no
other significant and robust differences in clinical or social outcome. There was insufficient data on costs to permit comparison.
ACT versus case management
There were no data on numbers remaining in contact with the psychiatric services or on numbers admitted to hospital. People allocated
to ACT consistently spent fewer days in hospital than those given case management. There was insufficient data to permit robust
comparisons of clinical or social outcome. The cost of hospital care was consistently less for those allocated to ACT, but ACT did not
have a clear cut advantage over case management when other costs were taken into account.
Authors’ conclusions
ACT is a clinically effective approach to managing the care of severely mentally ill people in the community. ACT, if correctly targeted
on high users of in-patient care, can substantially reduce the costs of hospital care whilst improving outcome and patient satisfaction.
Policy makers, clinicians, and consumers should support the setting up of ACT teams.
PLAIN LANGUAGE SUMMARY
Assertive community treatment for people with severe mental disorders
Synopsis pending.
BACKGROUND
patients and failing to meet their complex psychiatric and social
In the 1960s liberal democracies adopted a policy of caring for the needs (Melzer 1991, Audit Commission1986).
severely mentally ill in the community. Large psychiatric hospitals
were closed down and patients were treated in out-patient clinics, Assertive Community Treatment (ACT) was developed in the
day centres or community mental health centres. Sharply rising early 1970s to address the difficulties of caring for severely men-
readmission rates soon indicated that this type of community care tally ill people in the community (Thompson 1990). Initially
was less effective than anticipated (Rossler-Mannheim1, Ellison ACT was conceived as an alternative to acute hospital admission
1995). Community services were seen to be losing contact with (Stein-Madison, Hoult-Sydney) but it was soon more widely ap-
plied as a means of caring for patients who did not require im- The initial intention of this review was to make two main com-
mediate admission (Thompson 1990). The main goals of ACT parisons: i. ACT versus standard community care; and ii. ACT
were to: i. keep ill people in contact with services (Thornicroft versus case management. However, commentators on earlier drafts
1991); ii. reduce the extent of hospital admissions (and hence of the review pointed out that there were two distinct types of
costs) (Rossler-Mannheim2); and iii. improve outcome, especially ’standard community care’ against which ACT was being com-
social functioning and quality of life (Holloway 1991). pared. The first type consisted of care provided by a combination
of outpatient clinics and community mental health teams. The
Assertive Community Treatment should be practised according to second consisted of management by hospital-based rehabilitation
a defined and validated model, based on the consensus of an inter- services (which would tend to admit control patients initially for
national panel of ACT experts (McGrew 1994, McGrew 1995). stabilisation and then discharge them to the community). In trials
A key aspect of this model is that ACT is a team-based approach where the control treatment was ’rehabilitation’ there was clearly
(Burns 1995). Characteristically a multi-disciplinary team, includ- a high likelihood of people allocated to the control group being
ing social workers, nurses and psychiatrists, cares exclusively for admitted to hospital (because a planned period in hospital was an
a defined group of patients (McGrew 1995, Olfson 1990). Team integral part of the treatment). Therefore it was essential to split
members share responsibility for their clients, so it is common the first comparison into: i. ACT versus standard community care
for several members to work together with the same person. ACT (out-patients and CMHTs); and ii. ACT versus hospital-based re-
teams attempt to provide all the psychiatric and social care that habilitation.
their clients require, rather than obtain care from other agencies.
ACT teams try to provide care at home or in places of work (Olfson The effectiveness of case management versus standard care is re-
1990, Solomon 1992, Scott 1995, Thompson 1990). ACT teams viewed elsewhere on this CD-ROM (Marshall 1998). The present
practice ’assertive outreach’, meaning that they continue to con- review will refer to the case management review in its discussion
tact and offer services to reluctant or uncooperative people. ACT and conclusions, so it is recommended that the case management
teams also place particular emphasis on medication compliance review be read first.
(McGrew 1995).
ACT is often confused with case management, a superficially sim- OBJECTIVES

ilar approach that evolved at the same time in response to the
same problem. There are two key differences between ACT and The objectives are to determine the effectiveness of ACT as an
case management. The first is that ACT emphasises team working alternative to: i. standard community care; ii. traditional hospital-
and team responsibility - the vital link being between the team based rehabilitation; and iii. case management. For each of the
and the client group (not between individual team members and three comparisons the main outcome indices were: i. numbers
particular clients) (Taube 1990, Olfson 1990, Teague 1995). By remaining in contact with the psychiatric services; ii. extent of
contrast, case management emphasises professional autonomy and psychiatric hospital admissions; iii. clinical and social outcome;
individual responsibility - the vital link being between a single case and iv. costs.
manager and his or her ’case load’ of ’clients’ (Chamberlain 1991,
Solomon 1995). In other words, ACT team members share respon-
sibility for the sick individuals in their care, whereas case managers METHODS
carry individual case loads (Essock-Connecticut, Bond-Chicago1,
Anthony 1988, Harris 1987, Kanter 1989, Modcrin 1985). The
second difference is that ACT teams attempt to remain faithful to Criteria for considering studies for this review
a specified model (see above) whereas case management practice is
guided only by broad theoretical concepts (Ellison 1995). Whilst
some case managers may adopt elements of the ACT model, case
Types of studies
management, as generally practised, has little in common with
ACT (Essock-Connecticut, Teague 1995). Case management is All relevant randomised controlled trials. Only trials using an in-
far more widely practised than ACT (Ellison 1995) and reviewers tention-to-treat analysis were considered acceptable.
are generally agreed that it is effective (Scott 1995, Solomon 1992,
Holloway 1995, Thornicroft 1991); however, a systematic review
Types of participants
has shown that this consensus is questionable (Marshall 1998).
Proponents of ACT have argued that only the ACT model is ef- A majority of participants were required to be:i. within the age
fective (Stein 1992). range 18 to 65; and ii. suffering from severe mental disorder (de-
fined as: a. schizophrenia and schizophrenia-like disorders; b. bipo-
Aims of this review lar disorder; or c. depression with psychotic features). Substance
abuse was not considered to be a severe mental disorder in its own 4.1 General symptoms
right, however studies were eligible if they dealt with people with 4.2 Specific symptoms
both diagnoses, that is those with severe mental illness plus sub- 4.2.1 Positive symptoms (delusions, hallucinations, disordered
stance abuse. thinking)
4.2.2 Negative symptoms (avolition, poor self-care, blunted affect)
4.2.3 Mood - depression
Types of interventions 5. Quality of life
For an intervention to be accepted as ACT it must have been 5.1 No substantial improvement in quality of life
described in the trial report as: Assertive Community Treatment, 5.2 Patient satisfaction
Assertive Case Management or PACT; or as being based on the 5.3 Self-esteem
Madison, Treatment in Community Living, Assertive Community 6. Adverse effects
Treatment or Stein and Test models. 6.1 Death (suicide and non-suicide)
Trials of case management that did not meet the criteria for ACT
are considered in the case management review (Marshall 1998). 7. Economic
The review did not consider the use of ACT as an alternative 7.1 Cost of care
to acute hospital admission. The review also excluded studies of Measures of cost were expressed as mean weekly costs per patient
’Home-Based Care’ (which involves a multi-disciplinary team as- in the trial. Three main types of cost were reported:
sessing and treating urgent psychiatric referrals at home). Home- 7.2 Costs of psychiatric in-patient care
based care is a form of crisis intervention which deals with those 7.3 Costs of all health care (including the above plus the costs of
who are usually acutely ill, and should not be classified with either all other medical & psychiatric care such as: out-patient care and
ACT or case management as these are long-term interventions for Assertive Community Treatment)
severely and persistently ill people. 7.4 Total costs (including types of costs above plus the costs of
accommodation and transfer payments and minus benefits, such
as earnings).
Types of outcome measures
Primary outcomes
Search methods for identification of studies
1. General
1.1 Hospital admission
Secondary outcomes Electronic searches

1. General a. CINAHL (January 1982 to May 1997) was searched using the
1.1 Converting to psychosis during follow-up period CSG’s terms for randomised controlled trials and the CSG’s terms
1.2 Overall functioning for schizophrenia combined with the phrase:
1.3 Duration of hospital stay [and ((case or care) near management) or CPA or (Care near1 Pro-
1.4 Relapse - as defined by each study gramme near1 Approach) or (Assertive near1 Community near1
1.5 Length of hospitalisation Treatment) or PACT or TCL or (Training near (community near1
1.6 Loss to follow up living)) or (Madison near4 model)]
1.7 Satisfaction with treatment - participant/carer b. The Cochrane Schizophrenia Group’s Register (1997) was
1.8 Remaining in contact searched using the phrase:
2. Behaviour [and ((case or care) and management) or CPA or (Care and Pro-
2.1 General behaviour gramme and Approach) or (Assertive and Community and Treat-
2.2 Specific behaviours (e.g. aggressive or violent behaviour) ment) or PACT or TCL or (Training and (community and living))
2.3 Occurrence of violent incidents (to self, others or property) or (Madison and model)]
3. Social functioning c. EMBASE (January 1980 to May 1997) was searched using the
3.1 Imprisonment (e.g. police contacts & arrests) CSG’s terms for randomised controlled trials and the CSG’s terms
3.2 Employment status (number unemployed at end of study) for schizophrenia combined with the phrase:
3.3 Accommodation status (number homeless or not living inde- [and ((case or care) near management) or CPA or (Care near1 Pro-
pendently, mean days homeless and mean days in stable accom- gramme near1 Approach) or (Assertive near1 Community near1
modation per month in study) Treatment) or PACT or TCL or (Training near (community near1
4. Mental state living)) or (Madison near4 model)]
d. MEDLINE (January 1966 to May 1997) was searched using 1997). Only trials in category A or B were included in this review.
the CSG’s terms for randomised controlled trials and the CSG’s It was required that all included trials were to be conducted on an
terms for schizophrenia combined with the phrase: intention-to-treat basis.
[and ((case or care) near management) or CPA or (Care near1 Pro- Data management
gramme near1 Approach) or (Assertive near1 Community near1 No individual patient data were sought at this stage of the review.
Treatment) or PACT or TCL or (Training near (community near1 Data management issues will be considered below under three
living)) or (Madison near4 model)] headings: i. Missing data; ii. extraction and quality of data; and
e. PsycLIT (January 1974 to May 1997) was searched using the iii. general issues.
CSG’s terms for randomised controlled trials and the CSG’s terms i. Missing data
for schizophrenia combined with the phrase: Data were excluded from studies where more than 50% of those
[and ((case or care) near management) or CPA or (Care near1 Pro- randomised were lost to follow up (with the exception of the out-
gramme near1 Approach) or (Assertive near1 Community near1 come ’number remaining in contact’). Amongst included studies,
Treatment) or PACT or TCL or (Training near (community near1 data was not reported on outcomes where less than 50% of those
living)) or (Madison near4 model)] assessed at baseline failed to be reassessed on the same outcome at
follow-up. Data were reported as presented in the original studies,
without making any assumptions about those lost to follow up.
Searching other resources The number remaining in contact was estimated by taking the
1 Reference searching number of patients who were reinterviewed at the final follow-up
Each of the randomised controlled trial identified was sought as assessment in each trial. When analysing loss of contact in trials
a citation on the SCISEARCH database. Reports of articles that where deaths had occurred, treatment and control groups were
had cited these studies were inspected in order to identify further reduced by the respective numbers of deaths, so that deaths were
trials. Reference lists of all included trials and identified reviews not counted as losses of contact. Insufficient data were available to
were scanned for evidence of trials missed by the computerised determine how many people not re-interviewed were of unknown
search. whereabouts, known whereabouts but not having psychiatric con-
It should be noted that in electronic searches the phrase ’ACT’ is tact, and known whereabouts and having contacts but refusing to
not feasible as this common word generates a very large number be re-interviewed. The estimate of numbers remaining in contact
of false positives. assumes that patients who were not re-interviewed were likely to
be refusing or resisting further contact with the psychiatric ser-
vices, but this may not be entirely correct.
Data collection and analysis ii. Extraction of data and quality of data
Three types of outcome data were available: a. categorical data
Selection of studies
(such as number of patients admitted to hospital); b. count data,
The search for trials was performed independently by two review-
that is numerical data based on counts of real life events (such as
ers (AL, MM). Each read the abstracts of all publications detected
days in hospital or costs of care); and c. scale data, that is numerical
by their search (see search strategy above) and discarded irrelevant
data collected by standardised instruments (such as quality of life
publications, retaining only those trials in which some form of
interviews).
case management or ACT had been compared against a control
Categorical data were extracted twice and then cross-checked. Peto
treatment. The results of the two independent searches were then
Odds Ratios, and the number needed to treat (NNT), were cal-
merged to form a pool and copies were obtained of all papers
culated. A difference in outcome based on categorical data was
pertaining to trials in the pool. The reviewers together evaluated
considered to be robust if the possibility of the difference arising
the trials in the pool and decided which should be included in
by chance was less than 1 in 100 (that is 99% confidence intervals)
the systematic review of ACT. Subsequently, an independent rater
and the data on which the outcome was based were derived from
was asked to repeat the classification exercise on the pool of trials.
more than one study and did not show significant heterogeneity.
Inter-rater agreement between the reviewers and the independent
Numerical count data were extracted twice and cross-checked.
rater was 0.84 (Cohen’s kappa) indicating a high level of agree-
Count data could not be combined across studies for technical
ment on which trials should be included in this review. Included
reasons (the data were skewed) but all relevant observations based
trials were then allocated to the three comparisons by MM and
on count data were reported in the review. The reporting of all
AL. As yet no reliability study has been conducted for this part of
relevant count data (whether or not correctly analysed) was a de-
the procedure
parture from practice in earlier versions of the case management
Quality assessment
systematic review. Previously the procedure had been only to in-
Both reviewers rated the quality of all included trials. A rating
clude data that had been properly analysed. This practice had been
was given for each trial based on the three quality categories as
criticised by commentators on the grounds that the review then
described in the Cochrane Collaboration Handbook (Mulrow
failed to present clear information on the duration of in-patient iii. General issues
episodes (Kluiter 1997, Parker 1997). On reflection the reviewers In all cases the data were recorded on RevMan so that the area to
felt this criticism was reasonable - even though much of the data the left of the ’line of no effect’ indicated a ’favourable’ outcome
had been mis-analysed it nonetheless had an inherent meaning- for ACT. For binary outcomes (for example, ’admitted’ or ’not
fulness derived from reflecting the frequency of real-life events. admitted’) a standard estimation of the ’Odds Ratio’ (OR) and its
Nonetheless, the review continues to indicate whether reported confidence interval (99%) was calculated. The number needed to
observations based on count data were judged significant on the treat statistic (NNT), was also calculated. As well as inspecting the
basis of an appropriate statistical test. A difference in outcome graphical presentations, differences between the results of each in-
based on count data was considered robust if greater than 50% of cluded trial were checked using a test of heterogeneity. When het-
studies reporting this data showed a substantial difference (>33%) erogeneity was present, the data were re-analysed using a random
in favour of either treatment or control group and fewer than 25% effects model and efforts were made to identify the main source
showed a substantial difference in favour of the other group. of the heterogeneity.
Numerical scale data were subject to a quality assessment for two
reasons. a. It was felt they were prone to bias (see below); and b.
they lacked the inherent meaningfulness of count data, particu-
larly when reported without appropriate descriptive statistics. The
quality criteria used in the assessment were that scale data: i. were RESULTS
collected by an instrument described in a peer-reviewed journal; ii.
were elicited by self-report or an independent rater; iii. consisted
of a summary score for a broad area of functioning; and iv. were Description of studies
correctly analysed, and reported with descriptive statistics (that is
See: Characteristics of included studies; Characteristics of excluded
mean and standard deviation/error).
studies.
Skewed data is difficult to enter into a meta-analysis unless ’nor-
For substantive descriptions of studies please see Included Studies
malised’ by log transformation. Scale data with finite upper and
table.
lower limits can have an easy rule of thumb applied in order to test
Details of instruments used to collect continuous data.
for skewedness. If the standard deviation, when doubled, is greater
1. ACT versus Standard Care.
than the mean, the latter is not in fact the centre of the distribution
Continuous data collected by eight rating scales met criteria for
and should not be entered into the meta-analysis (Altman 1996).
inclusion in this comparison. Details of the scales that provided
Where continuous data has less obvious finite boundaries the situ-
these data are given below. Reasons for exclusion of data from other
ation is more problematic and may be matters of judgement. Rel-
instruments are given under ’outcomes’ in the ’included studies’
evant data that was skewed were presented in ’Other Data’ tables.
section.
The validity of the quality criteria was assessed by classifying all
Brief Psychiatric Rating Scale (Overall 1962).
observations (from the present review and the case management
A brief rating scale used by an independent rater to assess the sever-
review) into those based on ’poor quality’ data and those based
ity of a range of psychiatric symptoms, including psychotic symp-
on ’high quality’ data. The relative risk of ’being significant’ was
toms. The scale ranges from 24-168 with higher scores indicating
then calculated for observations based on ’poor quality’ data and
greater severity. Used in Audini-London.
observations based on ’high quality’ data. Observations based on
Brief Symptom Inventory (Derogatis 1983).
poor quality data were four times more likely to be significant
A brief rating scale used by an independent rater to assess the sever-
than those based on high quality data (N ’poor quality’ = 57, of
ity of psychiatric symptoms. Scores range from 0-4 with higher
which 17 significant; N ’good quality’ = 21, of which 2 significant;
scores indicating more symptoms. Used in Morse-St Louis1.
Relative Risk 4.0, 95% CI 1.26-12.7), suggesting the presence of
Colorado Symptom Index (Shern 1994).
bias and hence the validity of the quality criteria (see implications
A brief rating scale used by an independent rater to assess the
for research below).
severity of a range of psychiatric symptoms. A lower score indicates
more symptoms. Used in Lehman-Baltimore.
Numerical scale data of suitable quality were combined using the
Social Adjustment Sale (Weissman 1971).
standardised mean difference statistic where possible, otherwise
Measures social functioning in a number of life domains (work,
the data were reported in the text of the review. A difference in
social, extended family, marital, parental, family unit, and eco-
outcome based on scale data was considered robust if the possibility
nomic adequacy) on a scale of 1-7. Used in Audini-London
of the difference arising by chance was less than 1 in 100 (99%
and Jerrell-SCarolina (but with a different scoring system in
confidence intervals), and the data on which the outcome was
Jerrell-SCarolina, perhaps because this was an adapted version).
based were derived from more than one study and did not show
The Personality and Social Network Adjustment Scale (Clark
significant heterogeneity.
1968).
Rated on a scale of 0-4 with higher scores indicating better adjust- further problem was an apparent error in the reporting of numbers
ment. Used in Morse-St Louis1. admitted to hospital. In one table admission rates are reported as:
Client Satisfaction Questionnaire. (Larsen 1979). 31/52 experimental group and 33/45 control group; in another
Eight item patient-rated scale, each item rated 1-4. Higher scores table admission rates are reported as: 31/52 experimental and 25/
indicate greater satisfaction. Used in Audini-London and Morse-St 45 control. Further information is required before this trial can
Louis1. be included in the systematic review. The third problematic trial
Rosenberg Self-esteem Scale (Rosenberg 1979). was Mulder-Missouri. This early, unpublished, trial was classed
A short form of the Rosenberg Self Esteem Scale. Rated on a scale as awaiting assessment because data from randomised and non-
of 0-3 with higher scores indicating greater self-esteem. Used in randomised patients were not reported separately. Data from ran-
Morse-St Louis1. domised patients will be reported if they become separately avail-
Lehman’s Quality of Life Scale (General Well-Being) (Lehman able.
1983). 2. ACT versus hospital-based rehabilitation
Rated on a 7-point scale with higher scores indicating better quality Of 75 trials in the pool, three were eligible for inclusion in
of life. Used in Lehman-Baltimore. the second comparison. These were: De Cangas-Quebec, Lafave-
2. ACT versus hospital-based rehabilitation. Ontario, and Marx-Madison. De Cangas-Quebec was included
No continuous data met criteria for inclusion in this comparison. in the comparison despite posing some methodological problems.
3. ACT versus case management. First, the trial (reported in French) was hard to classify. It was
Continuous data collected by four rating scales met criteria for grouped with assertive case management studies on the grounds
inclusion in this comparison. of its title “le case management affirmatif ”, translated as “assertive
Brief Psychiatric Rating Scale (Overall 1962). case management” in the English language abstract. Caseload size
Details as above. Used in Morse-St Louis2. (the second inclusion criteria) was just inside the ACT borderline
Social Adjustment Scale (Weissman 1971). (20 people per team member), although actual follow up rates
Details as above. Used in Jerrell-SCarolina. indicate an active caseload of about 14. In assigning this trial to
Client Satisfaction Questionnaire (Larsen 1979). ACT the review is in agreement with Test 1992 who described it
Details as above. Used in Morse-St Louis2. as “a TCL like program”. However, it was not clear how far the
Self Esteem Scale (Short form, Rosenberg 1979) treatment offered (from a team of three nurses) really resembled
Details as above. Used in Morse-St Louis2. ACT. Second, the trial showed a number of unusual features: i. all
participants who were followed up were living with their families
(unusual in a study of the severely mentally ill); ii. complete data
Risk of bias in included studies collection across a wide range of variables was possible on all who
were followed up; and iii. cost data did not have the pronounced
1. ACT versus standard care.
positive skew normally expected in studies of this kind. In view
Of 75 trials in the pool, 17 met inclusion criteria for
of the unusual characteristics of this trial the comparison was re-
this comparison. The trials were Aberg-Stockholm, Audini-
peated without the De Cangas-Quebec data (see results section).
London, Bond-Chicago1, Bond-Indiana1, Chandler-California,
Godley-Illinois, Hampton-Chicago, Herinckx-Portland, Jerrell-
3. ACT versus case management
San Jose, Jerrell-SCarolina2, Lehman-Baltimore, Morse-St Louis1,
Of the 75 trials in the pool, six were included in the ACT versus
Mulder-Missouri, Quinlivan-California, Rosenheck-USA-10site,
case management comparison. These were Bush-Atlanta, Essock-
Solomon-Philadelphi2, and Test-Wisconsin. However three of the
Connecticut; Jerrell-SCarolina2, Morse-St Louis2, Quinlivan-
eligible trials have been classified as awaiting further assessment
California, and Solomon-Philadelphi2. Solomon-Philadelphi2,
pending clarification of methodological problems.
Jerrell-SCarolina2 and Quinlivan-California were also included in
The first of these problematic trials was Jerrell-San Jose. This par-
the ACT versus standard care comparison (Comparison 1 above) as
tially published trial was classified as awaiting further assessment
each had three arms (ACT, case management and control). There
because information is required on numbers of people excluded af-
were some problems extracting continuous data from Morse-St
ter randomisation (patients were excluded if they refused to partic-
Louis2, because the trial did not report the numbers of people
ipate after randomisation or if they had not been discharged from
in each group who provided data for the various outcome assess-
hospital within 6 months of entering the study). The trial may be
ments. An estimate of completers was made by dividing the total
included if these data become available. The second problematic
number of completers by three (the number of arms of the study).
trial was Godley-Illinois. This unpublished two-centre trial, was
classified as awaiting further assessment because it was not possible
to determine if the intervention was ACT or case management.
At one site the intervention appears to have been ACT, whereas
at the other it appears to have been intensive case management. A
Effects of interventions
The four main indices of outcome were: i. numbers maintaining ysed using ranks), fewer days homeless (Morse-St Louis2, anal-
contact with the psychiatric services; ii. extent of psychiatric hospi- ysed after transformation) and more days in stable accommoda-
tal admissions; iii. clinical and social outcome; and iv. costs. Each tion (Lehman-Baltimore). With respect to employment those ran-
index will be considered in turn for each of the three comparisons domised to ACT were less likely to be unemployed (OR 0.31,
1. ACT versus standard care. 99%CI 0.17-0.57, N=604, NNT 7.4, heterogeneity not signifi-
i. Numbers maintaining contact with the psychiatric services. cant). With respect to patient satisfaction, two trials showed that
Those receiving ACT were more likely to remain in contact with ACT resulted in a more satisfied clientele on the Client Satis-
services than those receiving standard community care (OR 0.51, faction Questionnaire (Weighted mean difference -0.56, 99%CI-
99%CI 0.37-0.70, N=1597, NNT 8.9). There was no significant 0.82 to -0.29, N=120). This improvement, of about one standard
heterogeneity on this outcome. deviation, probably represents a clinically significant difference.
ii. Extent of psychiatric hospital admissions. There was no clear difference between ACT and standard commu-
This outcome was assessed in two ways: a. likelihood of admission nity care on the outcomes of deaths, imprisonment/arrests/police
(the number of people in treatment and control groups admit- contacts, mental state, social functioning, self-esteem and quality
ted to hospital); and b. duration of admission (the mean days per of life. There was no evidence of clinically significant effect on
month in hospital for treatment and control groups). With respect mental state and social functioning because for both these out-
to likelihood of admission, those receiving ACT were significantly comes confidence intervals for the standardised mean difference
less likely to be admitted to hospital than those in the standard excluded a maximum effect of greater than half of one standard
community care groups (OR 0.59, 99%CI 0.41-0.85, N=1047, deviation. For the other outcomes confidence intervals were wide
NNT 10.3). There was, however, significant heterogeneity on this indicating that there was insufficient evidence to exclude a clini-
outcome (Chi Squared = 18.8, df 5). The effect remained signifi- cally significant effect in favour of either ACT or control.
cant at the 95% level after re-analysis with a random effects model, iv. Costs
but not at the 99% level. The heterogeneity relates to the size Three aspects of cost were considered: a. costs of psychiatric in-
rather than the direction of the effect (no trial found that ACT was patient care; b. costs of all health care; c. total costs. These data
inferior to standard community care on this variable). Count data are reported in ’other data tables’. Only five out of the fourteen
on duration of admission is summarised in an ’other data table’. included trials provided useful cost data and from these studies data
Nine of 14 trials in this comparison reported data on duration of were available on: a. costs of psychiatric in patient care (5 studies);
admission. Eight of these nine trials found a differences favouring b. costs of all health care (4 studies). No data were available on
ACT (in seven cases greater than 33%). All five significant differ- total costs. In all the cost observations the statistical significance
ences reported favoured ACT, although in three cases the data may of the difference between ACT and standard care was either not
not have been correctly analysed (the data is skewed and requires reported or was based on an incorrect statistical analysis (usually
transformation or use of a non-parametric test. the application of a parametric test to skewed untransformed data).
iii. Clinical and social outcome 2. ACT versus hospital-based rehabilitation
Across the 14 trials included in the comparison, 28 observations i. Numbers maintaining contact with the psychiatric services.
on clinical and social outcome were made using outcome scales. Those receiving ACT were no more likely to remain in contact
Eighteen of these observations were excluded because the data did with services than those receiving hospital-based rehabilitation.
not meet quality criteria (see Methods above) leaving 10 obser- The odds ratio had wide confidence intervals for this outcome in-
vations collected by nine scales. These observations were supple- dication that there was insufficient evidence to exclude a clinically
mented by categorical and count data. In summary, data were significant effect in favour of either ACT or control.
available on the following clinical and social outcomes: i. death; ii. Extent of psychiatric hospital admissions.
ii. imprisonment/arrests/police contact; iii. accommodation status This outcome was assessed in two ways: a. likelihood of admission
(homelessness, living independently, days in stable accommoda- (the number of patients in treatment and control groups admitted
tion); iv. employment; v. mental state; vi. social functioning; vii. to hospital); and b. duration of admission (the mean days per
patient satisfaction; viii. self-esteem; and ix. quality of life. month in hospital for treatment and control groups). ACT patients
Significant and robust differences between ACT and standard were significantly less likely to be admitted to hospital than those
community care were found on three outcomes, accommodation receiving hospital-based rehabilitation care (OR 0.2, 99%CI 0.09-
status, employment, and patient satisfaction. With respect to ac- 0.46, N=185, NNT 2.6). This finding remains significant after
commodation status those allocated to ACT were more likely to exclusion of De Cangas-Quebec (see Methodological Quality for
be living independently (OR 0.46, 99%CI 0.25-0.86, N=362, reasons). Data on duration of admission were available from two
NNT=6.6, heterogeneity not significant) and less likely to be- of the three studies and are summarised in ’other data tables’.
come homeless (OR 0.24, 99%CI 0.08-0.65, N=374, NNT= iii. Clinical and social outcome.
10.2, heterogeneity not significant). Those receiving ACT spent Across the three trials included in the comparison, 15 observa-
more days in independent accommodation (Test-Wisconsin, anal- tions on clinical and social outcome were made using outcome
scales. All of these observations were excluded because the data did
not meet quality criteria (see methods section above). Categorical iv. Costs
data was available on the following clinical and social outcomes: Data on the relative costs of ACT and case management were avail-
i. imprisonment/arrests; ii. accommodation status (living inde- able from three of six included trials (Jerrell-SCarolina, Morse-St
pendently); and iii. employment. Significant differences between Louis2, Quinlivan-California). From these trials data were avail-
ACT and hospital-based rehabilitation were found on all three out- able on: a. costs of psychiatric in-patient care (3 studies); b. costs
comes, although those relating to imprisonment/arrests are from of all health care (3 studies); and c. total costs (1 study). These
De Cangas-Quebec only and are not considered robust. With re- are presented in the ’other data tables’ of this review. In all the
spect to accommodation, those in the ACT groups were signif- cost observations reported above the significance of the difference
icantly more likely to be living independently (OR (for not liv- between ACT and standard care was either not reported or was
ing independently) 0.19, 99%CI 0.06-0.54, N=106, NNT 2.4). based on an incorrect statistical analysis (usually the application
With respect to employment the ACT group were significantly of a parametric test to skewed untransformed data).
less likely to be unemployed (OR 0.3, 99%CI 0.13-0.7, N=161,
NNT 3.6), but this effect is no longer significant after exclusion
of data from De Cangas-Quebec so it is not considered robust.
iv. Costs DISCUSSION
Cost data was only available only from one problematic trial (De
The validity of findings
Cangas-Quebec, see methodological quality above) and only for
total costs of all care. This finding is not considered robust. These will be considered for each of the main outcome indices in
3. ACT versus case management turn.
i. Numbers maintaining contact with the psychiatric services i. Numbers maintaining contact with the psychiatric services
Of the six trials included in this comparison only one (Bush-
Atlanta) provided data on this outcome. The available data were The review found that ACT was clearly superior to standard care
insufficient to permit calculation of ORs. on this index, but there was insufficient data to compare ACT
ii. Extent of psychiatric hospital admissions directly with hospital-based rehabilitation or case management.
This outcome was assessed in by recording duration of admission The superiority of ACT over standard care is valid, as confidence
(the mean days per month in hospital for treatment and control intervals for the odds ratio are narrow and there is no evidence of
groups). No data were available on likelihood of admission. heterogeneity.
iii. Clinical and social outcome ii. Extent of psychiatric hospital admissions
Across the six trials included in the comparison, six observations
on clinical and social outcome were made using outcome scales. The review found that ACT was superior to standard care and hos-
Five of these six observations met quality criteria for inclusion in pital-based rehabilitation in terms of numbers admitted to hospi-
the review. These observations were supplemented by categorical tal, but there was insufficient data to compare ACT directly with
and count data. In summary, data were available on the following case management. In terms of mean days in hospital, ACT was
clinical and social outcomes: i. imprisonment; ii. accommodation consistently superior to standard care, hospital-based rehabilita-
(mean days per month in stable accommodation); iii. mental state; tion, and case management. The finding that ACT is superior to
iv. social functioning; v. patient satisfaction; and vii. self-esteem. standard care on numbers admitted to hospital may be questioned
Statistically significant differences between ACT and case manage- as there is significant heterogeneity on this outcome. Nonetheless,
ment were found on two outcomes (accommodation status and the finding is likely to be valid for three reasons: i. the heterogene-
patient satisfaction). With respect to accommodation status, one ity relates to size rather than direction of effect; ii. the difference
study (Morse-St Louis2) showed that people in the ACT group remains significant after analysis using a random effects model;
spend more time in stable accommodation. With respect to satis- and iii. ACT is also superior to standard care on mean days in
faction, one study (Morse-St Louis2) showed that those receiving hospital per month. The finding that ACT is superior on mean
ACT were more satisfied with their care than those in the case days in hospital per month might itself be questioned as it is based
management group. As both findings were based on data from one on data that cannot be easily combined and which has not been
study only, neither are considered robust. properly analysed by the trialists. It would seem unreasonable to
There was no significant difference between ACT and case man- reject these data given the size and consistency of the effects in
agement on imprisonment, mental state, social functioning and favour of ACT (8 of 9 RCTS show a reduction in favour of ACT
self-esteem. However, confidence intervals were wide for these out- of greater than 30%).
comes indicating that there was insufficient evidence to exclude a iii. Clinical and social outcome
clinically significant effect in favour of either ACT or case man-
The review found that ACT was clearly superior to standard care
agement.
on three aspects of clinical and social outcome (accommodation,
employment, and satisfaction), but that it was not superior to The discussion that follows will focus mainly on ACT as an al-
standard care on measures of mental state and social functioning ternative to standard community care and to case management.
(there was insufficient data for definitive statements on other as- Whilst ACT has shown some superiority to hospital-based reha-
pects of clinical and social outcome). ACT was also superior to bilitation, the latter model is no longer widely practised and has
hospital-based rehabilitation on the accommodation aspect of so- limited relevance to modern community care policy.
cial outcome, but otherwise there was insufficient data. There was
insufficient data to make definitive statements about the relative Why is ACT not more widely practised?
effectiveness of ACT and case management.
It has been established that ACT is a clinically effective approach
It is possible that the positive findings described above can be at- to managing the care of severely mentally ill people in the com-
tributed to chance effects (Type II errors) or biased data collection munity. ACT, if correctly targeted on high users of psychiatric in-
techniques. This explanation is unlikely, however, because strict patient care, may bring about substantial reductions in health ser-
criteria were set before a judgement of superiority on a clinical or vice costs, whilst improving outcome and patient satisfaction. This
social outcome could be made (that is the data had to be: i. of being so, why is it that case management, an alternative but inef-
high quality; ii. from more than one trial; and iii. significant at fective approach (Marshall 1998), is much more widely practised
99% confidence intervals). Alternatively, it could be argued that (Ellison 1995)? There are three main reasons. First, ACT, when
the criteria for making a judgement of superiority were too strict, correctly practised, is an expensive treatment with high start-up
and that the review was over-zealous in applying quality criteria costs. It is therefore of limited appeal to the short-sighted policy
to data from outcome scales. Data excluded by the quality crite- maker, who will tend to seek a cheaper alternative. Second, ACT,
ria were, however, proven to be prone to bias, suggesting that the for cost reasons, tends to be restricted to high users of in-patient
quality criteria were valid. This question is discussed further under services, whereas case management can be offered to all comers
implications for researchers. (albeit in an ineffective form). Third, many proponents of case
management seem to believe that the research evidence supports
iv Costs their current practice. This convenient self-deception is possible
because researchers and reviewers have failed to draw a clear dis-
With respect to cost of in-patient care, ACT was consistently su-
tinction between what is ACT and what is case management. This
perior to standard care and case management (there was only lim-
means that it is possible to justify practically any variety of case
ited data for hospital-based rehabilitation). The picture was mixed
management by reference to successful trials of ACT. More sub-
with respect to costs of all health care and total costs. ACT was
tly, it can be argued that almost any variety of case management
usually, but not invariably, superior to standard care and to case
is effective because it incorporates ’elements of the ACT model’.
management. For the latter outcome, the limited data available
The problem with this argument is that the research evidence sup-
favoured case management.
ports only the practice of the whole ACT model, not the selective
There is a consistent pattern in the cost data for in-patient care adoption of any of its particular elements.
and for all health care, despite wide variations in costs across trials
and the fact that most data were incorrectly analysed. The data This review and its sister review on case management (Marshall
shows that when in-patient costs only are considered, ACT seems 1998) are notable for making an attempt to draw a clear distinc-
cheaper than other types of care, but this cost advantage is eroded tion between what is ACT and what is not. The method used to
when the costs of all health care are considered (because costs of all draw this distinction (the label the trialists used to describe the
health care includes the direct treatment costs of providing ACT). intervention) is open to criticism. It is not possible to be certain
This pattern implies that ACT is an expensive treatment whose that the trialists were applying their labels correctly. It may even
cost advantage over other forms of care depends on achieving a be that the interventions in successful trials are more likely to be
substantial relative reduction in the duration of in-patient admis- retrospectively labelled ’ACT’. Thus the classification technique
sions. In other words, ACT is only likely to achieve cost savings used by the reviews requires further refinement (see implications
when applied to populations that are already high users of in-pa- for research). Nonetheless, the application of this rough classifica-
tient care. This observation explains the mixed picture seen in the tion has highlighted a potential disparity in effectiveness between
costs table where the Chandler-California trial (in contrast to the ACT and case management, two superficially similar approaches
other trials) shows ACT becoming up to three times more expen- to the same problem. If this disparity proves to be correct, it has
sive than standard care. It is of interest that Chandler-California major implications for the way we care for severely mentally ill
differed from the other ACT trials by explicitly refusing to focus people in the community.
on high users of hospital care.
It is difficult to understand why total costs should favour case man-

agement, but at present only a limited amount of data is available. AUTHORS’ CONCLUSIONS
Implications for practice might make it possible to: i. identify the ’effective ingredients’ of
the ACT model; ii. identify those patients most likely to benefit
ACT is an effective way of caring for severely mentally ill people
from ACT; and iii. permit a more sophisticated meta-analysis of
in the community. It maintains contact with severely mentally ill
the existing economic data.
people, dramatically reduces the use of in-patient care, and im-
proves some aspects of outcome. ACT is popular with recipients There may be a stronger case for further ACT versus standard
of care and seems to be an attractive way of working for many care comparisons in countries outside the US, particularly those
clinicians. ACT teams could prove particularly useful in environ- with highly developed primary care services, such as the UK. It
ments where psychiatric in-patient care is at a premium. is remarkable that so far only two ’ACT versus standard care’
Policy makers, clinicians and consumers should therefore encour- randomised control trials have taken place outside the US. It is also
age the setting up of ACT teams. ACT, however, is expensive, so interesting that one of these trials (Audini-London) was largely
policy makers must consider how new ACT teams can be financed. unsuccessful. There is a need therefore to establish that the ACT
There are two possible solutions to this problem, neither of which model can be generalised beyond the USA.
necessarily involves considerable increases in expenditure. The first From the point of view of this review, an obvious direction for
solution is to provide ACT only to high users of psychiatric in-pa- future research is to find a more systematic way of classifying
tient care, as the evidence suggests that under these circumstances ACT and case management trials. The way forward is likely to be
ACT teams are self-financing. The second solution is to fund ACT through the application of a validated ACT fidelity scale (such as
teams from resources currently wasted on less effective forms of McGrew 1995). At present, however, there are two problems with
community care, such as case management. this approach. First, existing scales have no obvious cut-off point
for dividing ACT from non-ACT. Second, much of the data re-
Implications for research quired to complete an ACT fidelity scale will have to be obtained
This review has four recommendations to those planning future directly from trialists. Nonetheless, it remains a long-term aim of
research on ACT. First, more care should be taken to describe the reviewers to base our classification of ACT and case manage-
precisely the characteristics of the intervention, preferably by us- ment trials on some form of fidelity scale.
ing a validated fidelity scale such as that developed by McGrew A striking and unexpected finding of this review was the extent to
(McGrew 1995). Second, researchers should use well-validated in- which inadequately validated instruments were used to measure
struments to measure outcome (see below), and should also col- outcome. Of particular interest was the fact that data that failed
lect and report categorical and ’count’ data, such as deaths or days to meet quality criteria was four times more likely to show a sig-
in hospital. Third, researchers should present data in a form that nificant difference between treatment and controls. This finding
can easily be incorporated into a systematic review - this means suggests that there may be as yet some uncharted bias related to
reporting the means and standard deviations (or standard errors) the use of outcome scales in psychiatry. This finding deserves fur-
of all continuous outcome variables. Fourth, researchers should ther investigation and perhaps an attempt at replication on a larger
take care not to apply parametric tests to skewed data. scale.
In terms of the future direction of ACT research, it is unclear
whether there is a need for further comparisons of ACT versus
standard care, at least within the United States. There is a need,
however, to make maximum use of the data which is presently
ACKNOWLEDGEMENTS
available but inaccessible. The best way to achieve this is for ACT
trialists to agree to share individual patient data, perhaps within Jon Deeks for statistical advice. Clive Adams for general support.
the framework of a collaborative review group. Sharing of data Gill Harrison for her help with the reliability study.
REFERENCES
References to studies included in this review Lafave-Ontario {published data only}

∗
Lafave HG, deSouza HR, Gerber GJ. Assertive Community
Aberg-Stockholm {published data only} Treatment of severe mental illness: a Canadian experience.
∗
Aberg-Wistedt A, Cressell T, Lidberg Y, Liljenberg B, Osby U. Psychiatric Services 1996;47:757–759.
Two-year outcome of team-based intensive case management for
Lehman-Baltimore {unpublished data only}
patients with schizophrenia. Psychiatric Services 1995;46:1263–6. ∗
Lehman AF, Dixon LB, Kernan E, Deforge B. Assertive treatment
Audini-London {published data only} for the homeless mentally ill. 148th Annual Meeting of the
∗
Audini B, Marks IM, Lawrence RE, Connolly J, Watts V. Home- American Psychiatric Association. Miami: American Psychiatric
based versus out-patient/in-patient care for people with serious Association, 1995.
mental illness. Phase II of a controlled study. British Journal of
Marx-Madison {published data only}
Psychiatry 1994;165:204–10. ∗
Marx A, Stein L, Test M. Extrohospital management of severe
Bond-Chicago1 {published data only} mental illness. Feasibility and effects of social functioning. Archives
∗
Bond GR, Witheridge TF, Dincin J, Wasmer D, Webb J, De of General Psychiatry 1973;29(4):505–11. [MEDLINE:
Graaf-Kaser R. Assertive community treatment for frequent users of 1974021223]
psychiatric hospitals in a large city: a controlled study. American
Morse-St Louis1 {published data only}
Journal of Community Psychology 1990;18:865–91. ∗
Morse GA, Calsyn RJ, Allen G, Tempelhoff B, Smith R.
Bond-Indiana1 {published data only} Experimental comparison of the effects of three treatment programs
∗
Bond GR, Miller LD, Krumwied RD, Ward RS. Assertive case for homeless mentally ill people. Hospital and Community
management in three CMHCs: a controlled study. Hospital and Psychiatry 1992;3:1005–10.
Community Psychiatry 1988;39:411–8.
Morse-St Louis2 {published data only}
Bush-Atlanta {published data only} ∗
Morse GA, Calsyn RJ, Klinkenberg WD, Trusty ML, Gerber F,
∗
Bush CT, Langford MW, Rosen P, Gott W. Operation outreach: Smith R, et al.An experimental comparison of three types of case
intensive case management for severely psychiatrically disabled management for homeless mentally ill persons. Psychiatric Services
adults [see comments]. Hospital and Community Psychiatry 1990; 1997;48:497–503.
41:647–9. Wolff N, Helminiak TW, Morse GA, Calsyn RJ, Klinkenberg WD,
Chandler-California {published data only} Trusty ML. Cost-efectiveness evaluation of three approaches to case
∗
Chandler D, Meisel J, McGowen M, Mintz J, Madison K. Client management for homeless mentally ill clients. American Journal of
outcomes in two model capitated integrated service agencies. Psychiatry 1997;154:341–8.
Psychiatric Services 1996;47:175–80.
Quinlivan-California {published data only}
De Cangas-Quebec {published data only}
∗
Quinlivan R, Hough R, Crowell A, Beach C, Hofstetter R,
∗
de Cangas JPC. Le “Case management ” affirmatif: une Kenworthy K. Service utilization and costs of care for severely
evaluation complete d’un programme du genre en milieu mentally ill clients in an intensive case management program.
hospitalier. Sante mentale au Quebec 1994;19:75–92. Psychiatric Services 1995;46:365–71.
Essock-Connecticut {published data only} Rosenheck-USA-10site {published data only}

∗
Essock SM, Kontos N. Implementing Assertive Community Rosenheck R, Neale M, Gallup P. Community-oriented mental
Treatment Teams. Psychiatric Services 1995;46:679–83. health care: assessing diversity in clinical practice. Psychosocial
Rehabilitation Journal 1993;16:39–50.
Hampton-Chicago {published data only} ∗
Rosenheck R, Neale M, Leaf P, Milstein R, Frisman L. Multisite
∗
Hampton B, Korr W, Bond G, Mayes J, Havis P. Integration experimental cost study of intensive psychiatric community care.
services system approach to avert homelessness; CSP homeless prevention Schizophrenia Bulletin 1995;21:129–40.
project for HMI adults. State of Illinois NIMH Demonstration Rosenheck RA, Neale M. Inter-site variation in the impact of
Grant-Program, Final Report, 1992. intensive psychiatric community care on hospital use. American
Herinckx-Portland {published data only} Journal of Orthopsychiatry 1998;68(2):191–00.
∗
Herinckx HA, Kinney RF, Clarke GN, Paulson RI. Assertive Rosenheck RA, Neale MS. Cost-effectiveness of intensive
Community Treatment versus usual care in engaging and retaining psychiatric community care for high users of in-patient services.
clients with severe mental illness. Psychiatric Services 1997;48: Archives of General Psychiatry 1998;55(5):459–66. [MEDLINE:
1297–306. 1998255488]
Jerrell-SCarolina2 {published data only} Solomon-Philadelphi2 {published data only}

∗
Jerrell JM. Toward managed care for persons with severe mental ∗
Solomon P, Draine J, Meyerson A. Jail recidivism and receipt of
illness: implications from a cost-effectiveness study. Health Affairs community mental health services. Hospital and Community
1995;14:197–207. Psychiatry 1994;45:793–7.
Test-Wisconsin {published and unpublished data} Curtis-New York {published data only}
Cohen LJ, Test MA, Brown RL. Suicide and schizophrenia: data Curtis JL, Millman EJ, Struening E, D’Ercole A. Deaths among
from a prospective community treatment study. American Journal former psychiatric inpatients in an outreach case management
of Psychiatry 1991;147:602–7. program. Psychiatric Services 1996;47:398–402.
Kuhlman TL. Unavoidable tragedies in Madison, Wisconsin: a ∗
Curtis JL, Millman EJ, Struening E, D’Ercole A. Effect of case
third view. Hospital and Community Psychiatry 1992;43:72–3. management on rehospitalisation and utilisation of ambulatory care
Test MA, Knoedler WH, Allness DJ, Burke SS, Brown RL, services. Hospital and Community Psychiatry 1992;43:895–9.
Wallisch LS. Community care of schizophrenia: two year findings. D’Ercole A, Streuning E, Curtis JL, Millman EJ, Morris A. Effects
142nd Annual meeting of the American Psychiatric Association; of diagnosis, demographic characteristics and case management on
1989 May 6-11; San Francisco, California. American Psychiatric rehospitalization. Psychiatric Services 1997;48:682–8.
Association, 1989.
Dean-Birmingham1 {published data only}
∗
Test MA, Knoedler WH, Allness DJ, Burke SS, Brown RL, ∗
Dean C, Gadd EM. Home treatment for acute psychiatric illness.
Wallisch LS. Long term community care through an assertive
BMJ 1990;301:1021–3.
continuous treatment team. In: Tamminga CA, Schulz SC editor
(s). Advances in neuropsychiatry and psychopharmacology. Volume 1: Dean-Birmingham2 {published data only}
Schizophrenia Research. New York: Raven Press, 1991. ∗
Dean C, Phillips J, Gadd EM, Joseph M, England S. Comparison
of community based services with hospital based service for people
References to studies excluded from this review with acute, severe psychiatric illness. BMJ 1993;307:473–6.
Dharwadkar-Victoria {published data only}

Bigelow-Oregon {published data only}
∗
Dharwandkar N. Effectiveness of an assertive outreach
∗
Bigelow DA, Young DJ. Effectiveness of a case management community treatment program. Australian and New Zealand
program. Community Mental Health Journal 1991;27:115–23. Journal of Psychiatry 1994;28:244–9.
Fenton-Montreal {published data only}

Bond-Chicago2 {published data only} ∗
Fenton PR, Tessier L, Streuning EL. A comparative trial of home
∗
Bond GR, Witheridge TF, Wasmer D, Dincin J, McRae SA,
and hospital psychiatric care. One year follow-up. Archives of
Mayes J, Ward RS. A comparison of two crisis housing alternatives
General Psychiatry 1979;36:1073–9.
to psychiatric hospitalization. Hospital and Community Psychiatry
1989;40:177–83. Ford-London {unpublished data only}
∗
Ford R, Beadsmoore A, Ryan P, Repper J, Craig T, Muijen M.
Bond-Indiana2 {published data only} Providing the safety net: case management for people with a serious
∗
Bond GR, McDonel EC, Miller LD, Pensec M. Assertive mental illness. Journal of Mental Health 1995;4:91–7.
community treatment and reference groups: an evaluation of their Ford R, Raferty J, Ryan P, Beadsmoore A, Craig T, Muijen M.
effectiveness for young adults with serious mental illness and Intensive case management for people with serious mental illness -
substance abuse problems. Special issue: serving persons with dual site 2: cost effectiveness. Journal of Mental Health 1997;6:191–9.
disorders of mental illness and substance use. Psychosocial Ford R, Ryan P, Beadsmoore A, Craig T, Muijen M. Intensive case
Rehabilitation Journal 1991;15:31–43. management for people with serious mental illness- site 2: clinical
and social outcome. Journal of Mental Health 1997;6:181–90.
Borland-Spokane {published data only}
Lear G. Managing care at home. Nursing Times 1993;89(5):26–7.
∗
Borland A, McRae J, Lycan C. Outcomes of five years of
continuous intensive case management. Hospital and Community Franklin-Houston {published data only}
Psychiatry 1989;40:369–76. ∗
Franklin J, Solovitz B, Mason M, Clemons J, Miller G. An
evaluation of case management. American Journal of Public Health
Burns-London {published data only} 1987;77:674–8.
∗
Burns T, Beadsmoore A, Ashok VB, Oliver A, Mathers C. A
controlled trial of home-based acute psychiatric services. I: Clinical Glick-New York {published data only}
and social outcome. British Journal of Psychiatry 1993;163:49–54.
∗
Glick ID, Fleming L, DeChillo N, Meyerkopf N, Jackson C,
Burns T, Raftery J. Cost of schizophrenia in a randomized trial of Muscara D, et al.A controlled study of transitional day care for non-
home-based treatment. Schizophrenia Bulletin 1991;17:407–10. chronically ill patients. American Journal of Psychiatry 1986;143:
Burns T, Raftery J, Beadsmore A, McGuigan S, Dickson M. A 1551–6.
controlled trial of home-based acute psychiatric services. II. Goering-Toronto {published data only}
Treatment patterns and costs. British Journal of Psychiatry 1993; ∗
Goering PN, Wasylenki DA, Farkas M, Lancee WJ, Ballantyne R.
163:55–61. What difference does case management make?. Hospital and
Champney-Ohio {published data only}
∗
Champney TF, Dzurec LC. Involvement in productive activities Herz-New York {published data only}
and satisfaction with living situation among severely mentally ∗
Herz MI, Endicott J, Spitzer RL. Brief hospitalization: a two year
disabled adults. Hospital and Community Psychiatry 1992;43: follow-up. American Journal of Psychiatry 1977;134:502–7.
899–903. [MEDLINE: 1977154956]
Holloway-London {published data only} Marshall-Oxford {published data only}
∗
Holloway F, Carson J. Intensive case management for the severely Conway M. Care-management for mental illness. Lancet 1995;
mentally ill. Controlled trial. British Journal of Psychiatry 1998; 345:926–7. [MEDLINE: 1995223073]
172:19–22. [MEDLINE: 1998196325] ∗
Marshall M, Lockwood A, Gath D. Social services case-
management for long-term mental disorders: a randomised
Hornstra-Kansas {published data only} controlled trial. Lancet 1995;345:409–12.
∗
Hornstra RK, Bruce-Wolfe V, Sagduyu K, Riffle DW. The effect
of intensive case management on hospitalization of patients with Martin-Delaware {published data only}
schizophrenia. Hospital and Community Psychiatry 1993;44:844–7.
∗
Martin SM, Scarpitti FR. An intensive case management approach
for paroled IV drug users. Journal of Drug Issues 1993;23:43–59.
Hoult-Sydney {published data only}
Hoult J, Reynolds I. Schizophrenia: a comparative trial of McFarlane-New York {published data only}
community oriented and hospital oriented psychiatric care. Acta
∗
McFarlane WR, Stastny P, Deakins S. Family-aided assertive
psychiatrica scandinavia 1984;69:359–72. community treatment: a comprehensive rehabilitation and
∗
Hoult J, Reynolds I, Charbonneau-Powis M, Weekes P, Briggs J. intensive case management approach for persons with
Psychiatric hospital versus community treatment: the results of a schizophrenic disorders. New Directions for Mental Health Services
randomized trial. Australian and New Zealand Journal of Psychiatry 1992;53:43–54. [MEDLINE: 1992252807]
1983;101:160–7. McGowan-California {unpublished data only}
Reynolds I, Hoult JE. The relatives of the mentally ill: a ∗
McGowan M, Madison K, Meisel J, Chandler D. AB3777 Final
comparative trial of community-oriented and hospital oriented Report: The Integrated Service Agencies. Report to California
psychiatric care. Journal of Nervous and Mental Disease 1984;172: Department of Mental Health. Sacramento: Lewin-VHI, Inc, 1995:
480–9. Lewin-VHI, Inc. 1995.
Jerrell-SCarolina {published data only} McGrew-Indiana {published data only}
Jerrell JM. Toward cost-effective care for persons with dual ∗
McGrew JH, Bond GR, Dietzen L, Salyers M. Measuring the
diagnosis. Cost-effective Care 1996;23:329–37. fidelity of implementation of a mental health program model.
Jerrell JM, Hu T, Ridgely MS. Cost-effectiveness of substance Journal of Consulting and Clinical Psychology 1994;62:670–678.
disorder interventions for people with severe mental illness. Journal
of Mental Health Administration 1994;21:283–97. Merson-London {published data only}
∗
Jerrell JM, Ridgely MS. Comparative effectiveness of three
∗
Merson S, Tyrer P, Oynett S, Lack S, Birkett P, Lynch S, Johnson
approaches to serving people with severe mental illness and T. Early intervention in psychiatric emergencies: a controlled
substance abuse disorders. Journal of Nervous and Mental Diseases clinical trial. Lancet 1992;339:1311–3.
1995;183:566–76.
Modcrin-Kansas {published data only}
Knight-Los Angeles {published data only}
∗
Modcrin M, Rapp C, Poertner J. The evaluation of case
∗
Knight RG, Carter PM. Reduction of psychiatric inpatient stay management services with the chronically mentally ill. Evaluation
for older adults by intensive case management. The Gerontologist and Program Planning 1988;11:307–14.
1990;30:510–5. Mosher-San Francisco {published data only}
Kuldau-California {published data only}
∗
Mosher LR, Menn A, Matthews SM. Soteria: Evaluation of a
∗
Kuldau JM, Dirks SJ. Controlled evaluation of a hospital- home-based treatment for schizophrenia. American Journal of
originated community transitional system. Archives of General Orthopsychiatry 1975;45(3):455–67.
Psychiatry 1977;34:1331–40. Muijen-London1 {published data only}
Knapp M, Beecham J, Koutsgeorgopoulu V, Hallam A, Fenyo A,
Langsley-Denver {published data only}
Marks IM, et al.Service use and costs of home-based versus
∗
Langsley DG, Machotka P, Flomenshaft K. Avoiding mental
hospital-based care for people with serious mental illness. British
hospital admissions: a follow up study. American Journal of
Journal of Psychiatry 1994;165:195–203.
Psychiatry 1971;127:1391–4.
Marks IM, Connolly J, Muijen M, Audini B, McNamee G,
Lehman-Maryland {published data only} Lawrence RE. Home-based versus hospital-based care for people
∗
Lehman AF, Herron JD, Schwartz RP, Myers CP. Rehabilitation with serious mental illness. British Journal of Psychiatry 1994;165:
for adults with severe mental illness and substance use disorders. A 179–94.
clinical trial. Journal of Nervous and Mental Disease 1993;181:
∗
Muijen M, Marks I, Connolly J, Audini B. Home based care and
86–90. standard hospital care for patients with severe mental illness: a
randomised controlled trial. BMJ 1992;304:749–54.
Macias-Utah {published data only} Muijen M, Marks I, Connolly J, Audini B, McNamee G. The Daily
∗
Macias C, Kinney R, Farley WO, Jackson R, Vos B. The role of Living Programme. Preliminary comparison of community versus
case management within a community support system: partnership hospital-based treatment for the seriously mentally ill facing
with psychosocial rehabilitation. Community Mental Health Journal emergency admission. British Journal of Psychiatry 1992;160:
1994;30:323–39. 379–84.
Muijen-London2 {published data only} Susser-New York {published data only}
McCrone P, Beecham J, Knapp M. Community psychiatric nurse ∗
Susser E, Valencia E, Conover S, Felix A, Tsai WY, Wyatt RJ.
teams: cost-effectiveness of intensive support versus generic care. Preventing recurrent homelessness among mentally ill men: a
British Journal of Psychiatry 1994;165:218–21. “critical time”intervention after discharge from a shelter. American
∗
Muijen M, Cooney M, Strathdee G, Bell R, Hudson A. Journal of Public Health 1997;87(2):256–62. [MEDLINE:
Community Psychiatric Nurse Teams: Intensive support versus 1997226125]
generic care. British Journal of Psychiatry 1994;165:211–7. Teague-New Hamps {published data only}
Pai-Bangalore {published data only}
∗
Teague GB, Drake RE, Ackerson TH. Evaluating the use of
∗
Pai S, Kapur RL. Evaluation of home care treatment for continuous treatment teams for persons with mental illness and
schizophrenic patients. Acta Psychiatrica Scandinavica 1983;67: substance abuse. Psychiatric Services 1995;46:689–95.
80–8. Thornicroft-Baltim {published data only}
Pai S, Kapur RL. Impact of treatment intervention on the ∗
Thornicroft G, Breakey WR. The COSTAR programme 1:
relationship between dimensions of clinical psychopathology, social Improving social networks of the long-term mentally ill. British
dysfunction and burden on the family of psychiatric patients. Journal of Psychiatry 1991;159:245–9.
Psychological Medicine 1982;12:651–8.
Toro-New York {published data only}
Pai S, Roberts EJ. Follow-up study of schizophrenic patients ∗
Toro PA, Bellavia CW, Wall DD, Passero-Rabideau JM,
initially treated with home care. British Journal of Psychiatry 1983;
Daeschler CV, Thomas DM. Evaluating an intervention for
143:447–50.
homeless persons: results of a field experiment. Journal of
Polak-Denver {published data only} Consulting and Clinical Psychology 1997;65:476–84.
∗
Polak PR, Kirkby MW. A model to replace psychiatric hospitals.
Tyrer-London {published data only}
Journal of Nervous and Mental Disease 1976;162:13–22. ∗
Tyrer P, Morgan J, Van Horn E, Jayakody M, Evans K, Brummell
Reibel-Manhattan {published data only} R, et al.A randomised controlled study of close monitoring of
∗
Reibel S, Herz MI. Limitations of brief hospital treatment. vulnerable psychiatric patients. Lancet 1995;345:756–9.
American Journal of Psychiatry 1976;133:518–21.
Vincent-Cleveland {published data only}
Rossler-Mannheim1 {published data only} ∗
Vincent P, Price JR. Evaluation of a VNA Mental Health Project.
∗
Rossler W, Loffler W, Fatkenheuer B, Reicher-Rossler A. Does Nursing Research 1977;26:361–7.
case management reduce the rehospitalization rate?. Acta
Wood-New Zealand {published data only}
Psychiatrica Scandinavica 1992;86(6):445–9. ∗
Wood K, Anderson J. The effect on hospital admissions of
Rossler-Mannheim2 {unpublished data only} psychiatric case management involving general practitioners:
∗
Rossler W, Loffler B, Fatkenheuer A, Riecher-Rossler A. Case preliminary results. Australian and New Zealand Journal of
management for schizophrenic patients at risk of rehospitalization - Psychiatry 1994;28:223–9.
a case control study. European Archives of Psychiatry and Clinical
Neuroscience 1995;246(1):29–36. [MEDLINE: 1996369210] References to studies awaiting assessment
Santiago-Arizona {published data only}
Fekete {unpublished data only}
∗
Santiago JM, McCall-Perez F, Bachrach LJ. Integrated services for
Fekete DM, Bond GR, McDonel EC, Salyers MP, Chen A, Miller
chronic mental patients: theoretical perspective and experimental
L. Rural assertive community treatment: a field experiment.
results. General Hospital Psychiatry 1985;7:309–315.
Psychiatric Rehabilitation Journal 1998;21(4):371.
Solomon-Philadelphi1 {published data only}
Godley-Illinois {published data only}
Solomin P, Draine J. Family perceptions of consumers as case
Godley SH. The Illinois MI/SA Project: a treatment system united for
managers. Community Mental Health Journal 1994;30:165–76.
persons with mental illness and substance abuse. Bloomington,
∗
Solomon P, Draine J. Satisfaction with mental health treatment in
Illinois: Chestnut Health Systems, Inc., 1995.
a randomized trial of consumer case management. Journal of ∗
Godley SH, Hoewing-Roberson R, Godley MD. Final MISA
Nervous and Mental Disease 1994;182:179–84.
Report: Technical Report. Bloomington, Illinois: Chestnut Health
Stein-Madison {published data only} Systems, Inc., 1994.
∗
Stein LI, Test MA. Alternative to mental hospital treatment. I:
Jerrell-San Jose {published data only}
Conceptual model, treatment programme, and clinical evaluation. ∗
Jerrell J, Hu T. Cost-effectiveness of intensive clinical and case
Archives of General Psychiatry 1980;37:392–7.
management compared with an existing system of care. Inquiry
Stein LI, Test MA, Marx AJ. Alternative to the hospital: a
1989;26:224–34.
controlled study. American Journal of Psychiatry 1975;132:517–22.
Test MA, Stein LI. Alternative to hospital treatment. III. Social Mulder-Missouri {published data only}
Cost. Archives of General Psychiatry 1980;37:409–12. [MEDLINE:
∗
Mulder R. Evaluation of the Harbringer Program. Manuscript
1980152643] 1982. Evaluation of the Harbringer Program 1985.
Weisbrod BA, Test MA, Stein LI. Alternative to mental hospital Shern-New York {unpublished data only}
treatment. II. Economic benefit-cost analysis. Archives of General ∗
Shern DL, Tsemberis S, Anthony W, Lovell AM, Richmond L
Psychiatry 1980;37(4):400–5. [MEDLINE: 1980152642] Felton HC, Winarski J, Cohen M. Serving street dwelling
individuals with psychiatric disabilities: Outcomes of a psychiatric Holloway 1995
rehabilitation trial.. Manuscript 1996:outcomes of a psychiatric Holloway F. Home treatment as an alternative to hospital
rehabilitation trial. Manuscript (1996). admission. In: Tyrer P, Creed F editor(s). Community Psychiatry in
Action. Cambridge: Cambridge University Press, 1995.
References to ongoing studies
Holloway et al 1995
Holloway F, Oliver N, Colllins E, Carson J. Case management: a
Creed-UK-3centre {unpublished data only}
critical review of the outcome literature. European Psychiatry 1995;
∗
Creed F. A multi-centre trial of high versus low intensity case
10:113–28.
management. Currently underway at a number of sites in the UK..
Kanter 1989
Personal communication 1998.
Kanter JS. Clinical case management: definition, principles,
Hu-California {published data only} components. Hospital and Community Psychiatry 1989;40:361–8.
∗
Hu T, Jerrell J. Cost-effectiveness of alternative approaches in
Kluiter 1997
treating severely mentally ill in California. Schizophrenia Bulletin
Kluiter H. Inpatient treatment and care arrangements to replace or
1991;17:461–7.
avoid it - searching for an evidence-based balance. Current Opinion
Additional references in Psychiatry 1997;10:160–7.
Larsen 1979
Altman 1996 Larsen DH, Attkison CC, Hargreaves WA. Assessment of client/
Altman DG, Bland JM. Detecting skewness from summary patient satisfaction: development of a general scale. Evaluation and
information. BMJ 1996;313:1200. [: ACT020600] Program Planning 1979;2:197–207.
Anthony 1988 Lehman 1983
Anthony WA, Cohen M, Farkas MD, Cohen BF. Case Lehman A. The well being of chronic mental patients: assessing
management - more than a response to a dysfunctional system. their quality of life. Archives of General Psychiatry 1983;40:369–73.
Community Mental Health Journal 1988;24:219–228.
Marshall 1998
Audit Commission1986 Marshall M, Gray A, Lockwood A, Green R. Case management for
Audit Commission. Making a reality of community care. London: people with severe mental disorders. Cochrane Database of
HMSO, 1986. Systematic Reviews 1998, Issue 2. [DOI: 10.1002/
Burns 1995 14651858.CD000050]
Burns BJ, Santos AB. Assertive Community Treatment: an update McGrew 1994
of randomized trials. Psychiatric Services 1995;46:669–75. McGrew JH, Bond GR, Dietzn L, Salyers M. Measuring the fidelity
Chamberlain 1991 of implementation of a mental health program model. Journal of
Chamberlain R, Rapp CA. A decade of case management: a Consulting and Clinical Psychology 1994;62:670–8.
methodological review of outcome research. Community Mental McGrew 1995
Health Journal 1991;27:171–88. McGrew JH, Bond GR. Critical ingredients of Assertive
Clark 1968 Community Treatment: judgements of the experts. Journal of
Clark AW. The personality and social network adjustment scale. Mental Health Administration 1995;22:113–25.
Human Relations 1968;21:85–96. Melzer 1991
Derogatis 1983 Melzer D, Hale S, Malik SJ, Hogman GA, Wood S. Community
care for patients with schizophrenia one year after hospital
Derogatis LR, Melisaratos N. The Brief Symptom Inventory: An
introductory report. Psychological Medicine 1983;13:595–605. discharge. BMJ 1991;303:1023–6.
Modcrin 1985
Ellison 1995
Modcrin M, Rapp CA, Chamberlain R. Case management with
Ellison ML, Rogers ES, Sciarappa K, Cohen M, Forbess R.
psychiatrically disabled individuals: curriculum and training program.
Characteristics of mental health case management: results of a
Lawrence, Kansas: University of Kansas, School of Social Welfare,
national survey. Journal of Mental Health Administration 1995;22:
1985.
101–12.
Mulrow 1997
Endicott 1976
Mulrow CD, Oxman AD, editors. Cochrane Reviewers’ Handbook
Endicott J, Spitzer RL, Fleiss JL, Cohen J. The Global Assessment
[updated September 1997]. Cochrane Database of Systematic
Scale. Archives of General Psychiatry 1976;33:766–71.
Reviews 1997, Issue 4.
Harris 1987 Olfson 1990
Harris M, Bergman HC. Case management with the chronically Olfson M. Assertive Community Treatment: an evaluation of the
mentally ill: a clinical perspective. American Journal of experimental evidence. Hospital and Community Psychiatry 1990;
Orthopsychiatry 1987;57:296–302. 41:634–41.
Holloway 1991 Overall 1962
Holloway F. Case management for the mentally ill: looking at the Overall JE, Gorham DR. The Brief Psychiatric Rating Scale.
evidence. International Journal of Social Psychiatry 1991;37:2–13. Psychological Reports 1962;10:799–812.
Parker 1997 Taube 1990
Parker. Case management: an evidence-based review fails to make Taube CA, Morlock L, Burns BJ, Santos AB. New directions in
its case. Current Opinion in Psychiatry 1997;10:261–3. research on Assertive Community Treatment. Hospital and
Rosenberg 1979
Rosenberg M. Concieving the self. New York: Basic Books, 1979. Teague 1995
Teague GB, Drake RE, Ackerson TH. Evaluating the use of
Scott 1995 continuous treatment teams for persons with mental illness and
Scott JE, Dixon LB. Assertive Community Treatment and Case substance abuse. Psychiatric Services 1995;46:689–95.
management for schizophrenia. Schizophrenia Bulletin 1995;21:
657–67. Test 1992
Test MA. The Training in Community Living Model: delivering
Shern 1994 treatment and rehabilitation services through a continuous
Shern DL, Wilson NZ, Coen AS, Patrick DC, Foster M, Bartsch treatment team. In: Liberman RP editor(s). Handbook of
DA. Client outcomes II: longitudinal client data from the Colorado Psychiatric Rehabilitation. New York: MacMillan, 1992.
Treatment Outcome Study. Milbank Quarterly 1994;72:123–48.
Thompson 1990
Solomon 1992 Thompson KS, Griffity EEH, Leaf PJ. A historical review of the
Solomon P. The efficacy of case management services for severely Madison Model of community care. Hospital and Community
mentally disabled clients. Community Mental Health Journal 1992; Psychiatry 1990;41:625–34.
28:163–80.
Thornicroft 1991
Solomon 1995 Thornicroft G. The concept of case management for long term
Solomon PS, Draine J, Delaney MA. The working alliance and mental illness. International Review of Psychiatry 1991;3:125–32.
consumer case management. Journal of Mental Health Weissman 1971
Administration 1995;22:126–34. Weissman MM, Paykel ES, Siegel R, Klerman GL. The social role
Stein 1992 performance of depressed women: comparisons with a normal
Stein LI. On the abolishment of the case manager. Health Affairs group. American Journal of Orthopsychiatry 1971;41:390–405.
1992;11:172–7. ∗
Indicates the major publication for the study
CHARACTERISTICS OF STUDIES
Characteristics of included studies [ordered by study ID]
Aberg-Stockholm
Methods Allocation: randomised - no further details.

Follow up: 24 months.
Lost to follow up: 5/40.
Participants Setting: Kungsholmen sectorised service, Stockholm, Sweden.

Inclusion criteria: i. DSM-III-R schizophrenic disorder; ii. age 25-55; iii. recently admitted to ward /
currently in OPD.
N = 40.
Age: mean 38 years.
Sex: 35% F.
Interventions 1. “Intensive team-based case management based on...principles [of ]...Stein and Test” (1 hosptial, 1 OPD
multi-disciplinary team (size = 4); shared case load, team held primary responsibility, unlimited follow up,
24 hour care available (not from team), contact > 4.5 hrs / week / client, approach emphasised medication
compliance & offered life skills training & support*, staff:client ratio 1:2.5. N=20.
2. “Standard psychiatric services” from multi-disciplinary specialist OPD, each assigned 1 contact person,
co-ordinated via team meetings, staff:client ratio ~ 1:10. N=20.
Outcomes Lost to follow up.

Unable to use -
Days in hospital (no mean, no SD).
Number of emergency visits (no mean, no SD).
Quality of life (no mean, no SD).
Social network size (measure validated on children only, no mean, no SD).
Burden of care (no mean, no SD).
Notes * sharing of case load, frequency of team meetings, role of co-ordinator, frequency & location of contacts
- not clear.
Risk of bias
Item Authors’ judgement Description
Allocation concealment? Unclear B - Unclear
Audini-London

Follow up: 4, 15 months.
Lost to follow up: 8/66 (12%).
Participants Setting: London, UK.

Inclusion criteria: “seriously mentally ill” people (30% schizophrenia) who had completed > 20/12s in
Audini-London (Continued)
treatment arm of a trial of ACT (see Muijen-London).

N= 66.
Age: median 37.
Sex: 54% F.
Race 26% black.
History: mean 0.17 admissions in last year.
Interventions 1. ACT: “modelled on the ACT care services developed in Madison by Stein and Test”. N=33.
2. Routine care from the psychiatric services. N=33.
Outcomes Hospital admission.

Numbers lost to follow-up.
Deaths.
Days in hospital.
Mental state (BPRS).
Social functioning (SAS).
Patient satisfaction (CSQ).
Unable to use -
Days in hospital (no SD).
Mental state (GAS, PSE, same variable (mental state) assessed by multiple measures)
Relative’s satisfaction (not peer-reviewed scale).
Notes People in this study recruited after 20-30 months of ACT within Muijen-London.
Authors report that the ACT team became “depleted and demoralized” in the course of this trial.
Risk of bias
Bond-Chicago1
Methods Allocation: randomised, sealed envelope, independent secretary performed assignment.

Lost to follow up: 30/88 34(%).
Participants Setting: Chicago, Illinois, USA.

Inclusion criteria: i. >18 years; ii. schizophrenia, schizo-affective disorder, affective disorder, or incapaci-
tating personality disorder; iii. living within 3 miles of treatment program; iv. 3 hospitalisations in last 2
years; v. total of 5 hospitalisations in life-time; vi. not already receiving ACT.
N = 88.
Interventions 1. ACT. N=45.

2. Standard care from a drop-in centre. N=43.
Bond-Chicago1 (Continued)

Days in hospital.
Lost to follow-up.
Deaths.
Client satisfaction (number reporting treatment as “very helpful”).
Imprisoned / arrested.
Unable to use -
Social functioning (not peer-reviewed scale, no SD)
Subjective Quality of Life ((not peer-reviewed scale, no SD).
Satisfaction with Care (CSQ, no SD).
Global functioning (GAS, no SD).
Notes
Risk of bias
Allocation concealment? Yes A - Adequate
Bond-Indiana1
Methods Allocation: random allocation - no further details.

Lost to follow up: 43/167 (25.7%).
Participants Setting: 3 CMHCs, Indiana, USA.

Inclusion criteria: i. >17years; ii. diagnosis of psychotic disorder; and either iii. discharged from a state
hospital in past year; or iv. hospitalised > 3 times in past 2 years; or v. believed by staff to be at risk of
readmission; or vi. awaiting committment to an Indiana State Hospital; or vii. presenting for admission
to the CMHC inpatient unit and having had four hospitalisations in last 2 years.
N = 167.
Age: mean 34.5.
Sex: 38% F.
Race: 34% black.
Diagnosis: 61% schizophrenia.
History: mean number previous admissions 8.8; mean number admissions in previous year: 1.5.
Interventions 1. “PACT developed by Stein and Test”, staff:client ratio ~ 1:7. N=84.
2. Public mental health services (included unspecified amount of brokerage-style case management). N=
83.

Lost to follow-up.
Deaths (data from centre A only).
Arrests.
Days in hospital.
Costs - total.
Bond-Indiana1 (Continued)
Unable to use -
Quality of life (not peer-reviewed scale, no data).
Notes Results reported seperately for the 3 centres & data reported inconsistantly across centres. ’N’ therefore
varies depending on quality of data reporting.
Risk of bias
Bush-Atlanta
Methods Allocation: random assignment - no further details.

Lost to follow up: 0/28.
Participants Setting: Atlanta, USA.

Inclusion criteria: i. “severe mental illness”*; ii. “high rates of recidivism”*; iii. “difficulties in community
living”*.
N = 28.
Interventions 1. ACT (Madison model), staff:client ratio ~ 1:2. N=14.

2. Low intensity case management & rehabilitation from standard services. N=14.

Deaths.
Days in hospital.
Unable to use -
Compliance (no data).
Accommodation (no data).
Notes * Not defined further.
Risk of bias
Chandler-California

Follow up: 12 months
Participants Setting: 1 urban, 1 rural but integrated service agencies in Califiornia, USA.
Inclusion criteria: i. “serious & persistent mental disorder” - not substance abuse; ii. functional impairment
due to mental disorder; iii. eligible for public assistance due to mental disorder.
N = 516.
Age: ~ a third > 45 years.
Sex: 41% F.
Race: 37% black.
History: 26% admitted in last year.
Interventions 1. “TCL model described by Test” - multi-disciplinary teams (psychiatrist involved); team takes primary
responsibility; 24 hour cover; unlimited intervention; separate site from hosptial; shared case load; emphasis
on assertive outreach and in vivo treatment. N=252.
2. “Usual services” - included OPD, day treatment, case management and minimal rehabilitative services.
N=264.

Hospital admission.
Arrest.
Employment.
Days in hospital.
Costs of hospital care & of all mental health care.
Unable to use -
Quality of life (QOLS, no data).
Self esteem (RSES, no data).
Mental state (CSI, no data).
Social Activity (not peer-reviewed scale).
Notes
Risk of bias
De Cangas-Quebec

Follow up: 6 months after discharge from hospital.
Participants Setting: Quebec, Canada.

Inclusion criteria: i. Just admitted to psychiatric hospital; ii. no primary diagnosis of organic brain disorder,
personality disorder, substance abuse or mental retardation.
De Cangas-Quebec (Continued)
N = 120.
Age, sex: data reported on completers, unable to give data of those randomised.
Interventions 1. “Le case management affirmatif ” (ACT) from team of 3 nurses, staff:client ratio 1:20. N=60.
2. Routine inpatient care, then routine community care from hospital services. N=60.

Lost to follow-up.
Imprisonment, arrests.
Employment.
Deviant behaviour (REHAB).
Costs - total.
Unable to use -
Days in hospital (error in data table).
Social functioning (REHAB, full data not reported).
Family Burden: (no SD).
Quality of life (“Oregon Quality of Life Inventory” - not peer-reviewed scale, reporting by items).
Expressed emotion (Five Minute Speech Sample - trivial outcome in an ACT trial).
Notes
Risk of bias
Essock-Connecticut
Methods Allocation: “randomly assigned” - no further details.

Lost to follow up: ~ 13/262.
Participants Setting: 3 centre study, Connecticut, USA.

Inclusion criteria: i. severe mental disorder; ii. high service use (>2 admissions last 2 years / 1 admission
>180 days last 2 years / >2 contacts with crisis services last 2 years); and iii. significant difficulty meeting
demands of everyday life (homeless at some time in past year / required extensive supervision or assistance
> weekly to meet personal needs).
N = 262.
Age: mean 41 years.
Sex: 36% F.
History: mean number admissions > 2.
Interventions 1. ACT - 3 teams (2 F/T nurses, P/T psychiatrists, 10 members), no individual case loads, 24 hr cover;
9.1hrs face-to-face/month; 66% contacts in non-office setting, staff:client ratio ~1:6. N=131.
2. “High quality” case management - generalist model but case managers mobile, seeing patients in own
homes, and “assertive” on their behalf, graduate (usually) social workers carried discrete caseloads; no 24
Essock-Connecticut (Continued)
hr cover; 1.3hrs face-to-face/month; 78% contacts in non-office setting, staff:client ratio ~ 1:25/30. N=
131.
Outcomes Days in hospital.

Days homeless.
Notes Results not fully published as yet.
Risk of bias
Hampton-Chicago

Follow up: 6 & 12 months.
Lost to follow up: ~ 26/165 (15.7%).
Participants Setting: 2 centres, Chicago, USA.

Inclusion criteria: i. admitted to inpatient units a state hospital; ii. homeless on admission / at risk of
homelessness on discharge.
N=165.
Age: mean 37.3 years.
Sex: 23% F.
Race: 55.7% black.
Diagnosis: 42.1% schizophrenia.
Histroy: mean number previous admissions ~ 13; mean number admissions in last year ~ 2.8.
Interventions 1. “Assertive case management” & “ACT”, staff:client ratio ~1:10. N=82.
2. Routine follow up care from psychiatric services. N=83.
Outcomes Lost to follow-up (site 1).

Deaths.
Housing status at end of study.
Days in hospital.
Unable to use -
Lost to follow up (site 2 - figures in project report do not add up - > than total N in study).
Admissions (no SD).
Notes In one centre there were possible problems with program implementation - this centre had less positive
results.
It is not clear how far the stated ’Ns’ include those leaving the study early.
Authors are being contacted for further information.
Risk of bias
Hampton-Chicago (Continued)
Herinckx-Portland

Follow up: 2.4 years
Lost to follow up: 70/174 (40%) at 2.4 years.
Participants Setting: Portland, Oregon, USA.

Inclusion criteria: i. severe mental disorder (schizophrenia, bipolar disorder, major depression); ii. persistent
history of psychotic symptoms not due to substance abuse; iii. impaired functioning in > 2 of:- social role,
daily living, social acceptability; iv. not mentally retarded.
N = 174.
Sex: 39% F.
Race: 18% black.
History: 61% 2+ admissions in last 6 months.
Interventions 1. “ACT” (non consumer), 4 F/T & 1 P/T case managers (includes team leader), input from nurse &
psychiatrist, trained in ACT-style treatment, caseload assigned to entire team, services delivered in vivo,
small case loads; 24 hour cover, continuity of care, team had ultimate responsibility. N = 58.
2. “ACT” (consumer), as above but consumer case managers (60% had suffered from bipolar disorder).
N=58.
3. Care from 1 of 4 CMHCs & a number of smaller agencies (none providing assertive outreach). N=58.
Outcomes Lost to follow-up (at 15 months).
Notes Participants were being transfered from inpatient care or moving from other services in the community.
Study design suggests that other outcomes will become available.
Risk of bias
Jerrell-SCarolina2

Lost to follow up: none.
Jerrell-SCarolina2 (Continued)
Participants Setting: “Large urban mental health system”, USA.

Inclusion criteria: i. DSM-III-R psychotic or major affective disorder; ii. 2+ in-patient admissions in last
year or lengthy residential treatment & repeated emergency visits; and iii. 2+ of:- poor work history,
eligible for public assistance, poor living skills, poor social support, history of inappropriate behaviour.
N = 122.
Interventions 1. “PACT adaption model”, multi-disciplinary team including psychiatrist, daily meetings, mainly home-
based treatment, staff:client ratio ~ 1:15-20. N=40.
2. “Intensive broker case management”, intensive support from case managers (“paraprofessionals” work-
ing independently & soley in the field), weekely meetings, service “relationship oriented”, focussing on
“empowering clients”, staff:client ratio ~ 1:15-18. N=42.
3. Clinical team with some supplemental case management for 25% most unstable clients, staff:client
ratio 1:35 or more. N=40.
Outcomes Costs - total of psychiatric care.

Social behaviour (SAS).
Unable to use -
Mental state (data skewed).
Role functioning (not peer-reviewed scale).
Notes
Risk of bias
Lafave-Ontario

Outcome assessment: by independent consultants.
Lost to follow up: 13/65 (20%).
Participants Setting: Brockville, Ontario, Canada.

Inclusion criteria: i. age 16-65; ii. referred for “psychiatric rehabilitation” to long-term hospital.
N = 65.
Age: mean 36 years.
History: mean number previous admissions 3.8.
Interventions 1. “ACT”, provided by multidisciplinary team (psychiatrist, nurses, social workers, vocational counsellor,
’consumer’ support-worker), staff:client ratio > 1:4. N=24.
2. Standard psychosocial rehabilitation for high service users, +/- rehabilitaton assessment (hospital-based)
& psycho-social rehabilitation (community-based) from hospital treatment team. N=41.
Lafave-Ontario (Continued)
Outcomes Lost to follow-up.

Admitted to hospital.
Days in hospital.
Accommodation outcome.
Unable to use -
Quality of life (QOLS, no SD).
Satisfaction with care (CSQ, no SD).
Quality of environment (Environmental Index, unclear if this is an appropriate measure - designed for
use in institutions, no SD).
Notes Unclear if those randomised to group 2 were admitted to hospital as a matter of course. If this were the
case it would be preferable to compare post-discharge readmission rates.
Risk of bias
Lehman-Baltimore
Methods Allocation: stratified random assignment - no further details.

Follow up: 2,6,12 months.
Lost to follow up: 26/152 (17%) (12 months).
Participants Setting: Baltimore, Maryland, USA.

Inclusion criteria: i. homeless*; ii. severe mental disorder**.
N = 152.
Age: mean 37 years.
Sex: 33% F.
Race: 72% black.
Interventions 1. “PACT model of Stein and Test”, continuity of care, 24 hour availability, 12 F/T staff (including social
worker director, psychiatrist, 6 clinical case managers (nurses, social workers), 2 consumer advocates,
receptionist, family outreach worker, P/T nurse practitioner - each assigned to mini-team), whole team
knew clients & took part in decision-making, daily meetings, site visit confirmed fidelity to ACT model,
staff:client ratio ~ 1:10. N=77.
2. Routine psychiatric care, mainly CMHCs, emergency facilities, some generic case management in
comparison programme (degree unclear). N=75.

Lost to follow-up.
Living independently.
Mental state (CSI).
Quality of life (QOLS).
Days in hospital.
Days in stable accommodation.
Lehman-Baltimore (Continued)
General health (SF36).

Unable to use -
Specific items reported from QOLS (not global assessments).
Social functioning (objectiive QOLS, no data).
Days in jail (so infrequent as to be a trivial outcome).
Days homeless (split reporting of different types of homelessness, no SD).
Notes *Homeless - on street or shelter for >4days last 45 or >14 last 180; or in temporary accommodation with
> 2 residential moves in last 6 months.
** Severe mental disorder - diagnosis of schizophrenia or schizophrenia-like illness or recieving benefit
because of mental disorder or had another axis I disorder & either > 2 hosptialisations of > 21 days in
past 3 years or a total of > 42 days prior to current hospitalisation or > 90 days in psychiatric hospital or
nursing home in past 3 years or mental disability lasting > 1 year during which not able to spend > 75%
of time in some gainful activity.
Note complex inclusion criteria.
Risk of bias
Marx-Madison

Participants Setting: Madison, Wisconsin, USA.

Inclusion criteria: i. Inpatient at Mendota State Hospital; ii. considered by staff to be incapable of sustained
community living; iii. duration of admission 3-18 months; iv. < 50% of past 4 years in hospital; v. 20-45;
vi. no primary diagnosis of organic brain disease or substance abuse.
N = 41.
Age: mean 29 years.
Sex: 36.5% F.
Histroy: mean number past admissions ~ 3.5 in past 4 years.
Interventions 1. Early version of TCL model (undertaken by inventors of ACT approach), inpatients judged to require
hospital rehabilitation were instead diischarged to “total in-community treatment”, staff:client ratio at
worst 1:10. N=21.
2. Prepareded for discharge on rehabilitation unit by the same staff who provided ACT in group 2. N=
20.
Outcomes Lost to follow-up.

Employed (paid work).
Unable to use -
Marx-Madison (Continued)
Days in hospital (no SD).

Numbers in hospital at end of study (not independent of outcome ).
Global functioning (SCRS, IMPS, no data).
Social functioning (KATZ self report, ACL, MACC II Behavioral Adjustment scale, no data).
Self esteem (RSES, no data).
Notes Those in hospital randomised to immediate ACT (community-based) or active rehabilitation (hospital-
based) from the same ACT team members.
There was further comparison group receiving standard hospital care.
Risk of bias
Morse-St Louis1
Methods Allocation: random assignment*.

Participants Setting: Homeless mentally ill services, St Louis, USA.

Inclusion criteria: i. homeless; ii. no indication of serious violent behaviour; iii. serious psychiatric disorder
(previous psychiatric hospitalisation, score > 90th centile on GSI, & on psychoticism, paranoid ideation,
or depression subscales of BSI).
N = 116.
Sex: 42% F.
Race: 52.5% black.
History: 75% >1 admissions.
Interventions 1. “Guided by principles from...ACT programs associated with the TCL program”, staff:client ratio 1:10.
N = 52.
2. Routine care from outpatient psychiatric services operated by Missouri Department of Mental Health,
psychotherapy, psychiatric medication, assistance in obtaining social security. N=64.
Outcomes Days homeless.

Lost to follow-up.
Satisfaction with care.
Mental state (GSI of BSI).
Social functioning (PSNAS).
Self esteem (RSES).
Unable to use -
Housing state (no data).
Monthly income (data skewed).
Alienation (not peer-reviewed scale).
Morse-St Louis1 (Continued)
Alcohol consumption (data skewed).
Notes *Initially 50 assigned to each group - those who subsequently refused treatment, failed to be linked to
treatment, or were lost within one month of screening (n=28) were replaced by people also randomly
assigned to the 3 groups. Data in this review is based on sample sizes after replacement of early drop-outs.
Data from a drop-in centre control group is not included in this analysis.
Risk of bias
Morse-St Louis2

Participants Setting: ACT teams - city centre, case management team - St Louis Mental Health Centre, USA.
Inclusion criteria: i. severe mental illness (not defined); ii. recently homeless, or history of frequent
homelessness, or in acute crisis as indicated by current treatment in an emergency room or hospital unit.
N = 165.
Sex: 42% F.
Race: 55% black.
Interventions 1.“ACT”, indefinite duration, 5-7 F/T staff; 2 hours from psychiatrist / week, no nurses, team took full
responsibility for clients, staff:client ratio ~ 1:10. N=55.
2. “ACT with community workers” - as for 1. but a paraprofessional community worker also assigned to
each client. N=55.
3. “Broker case management” - case manager (mainly office-based) assigned to develop individual service
plan, arrange & purchase mental health & social services, monitor the quality of purchased care & adjust
services accordingly, staff:client ratio 1:85. N=55.
Outcomes Mental state (BPRS).

Self-esteem (RSES).
Satisfaction with care.
Unable to use -
Severity of substance misuse (ASI, no global score provided).
Income (highly skewed data).
Notes
Risk of bias
Morse-St Louis2 (Continued)
Quinlivan-California
Methods Allocation: “random allocation” - no further details.

Follow up: 2 years.
Lost to follow up: apparently none - not clear.
Participants Setting: San Diego County, USA.

Inclusion criteria: i. age >18; ii. DSM-III-R axis I disorder; iii. 3+ hospitalizations in last 2.5 years.
N = 90.
Age: mean 37 years.
Sex: 56% F.
Race 18% black.
Interventions 1. ACT style case management, included assertive outreach, team working, control of patient finances,
staff:client ratio 1:15 (or less). N=30.
2. Low-intensity case management, staff:client ratio 1:40. N=30.
3. Treatment from public mental health system. N=30.
Outcomes Days in hospital (reported but not suitable for meta-analysis).

Costs of psychiatric in patient care (reported but not suitable for meta-analysis).
Unable to use -
Costs of other psychiatric care (not one of the three cost outcomes examined by this study).
Notes
Risk of bias
Rosenheck-USA-10site
Methods Allocation: random assignment by independent researcher.

Lost to follow up: not clear - outcome based on service utilisation statistics.
Participants Setting: 4 neuropsychiatric hospital & 6 general medical hospital intensive psychiatric community care
programs, USA.
Inclusion criteria: i. Current inpatient in VA psychiatric unit; ii. no primary diagnosis of substance abuse
or organic brain disease; iii. recent high user of psychiatric care (definition varied by site).
N = 873.
Sex: 100% M.
Rosenheck-USA-10site (Continued)
Race: 20.2% non-white.

Interventions 1. “ACT-like”.. “..based on consultation with..expert in the Wisconsin ACT model”, fidelity of interven-
tion assessed by “a second expert in the PACT approach”, program inadequately implemented, staff:client
ratio 1:7-15. N=454.
2. Routine care from the psychiatric services. N=419.
Outcomes Days in hospital.

Cost - total (data highly skewed but log transformed before analysis - findings reported in review text).
Notes
Risk of bias
Allocation concealment? Yes A - Adequate
Solomon-Philadelphi2

Lost to follow up: not reported.
Participants Setting: Jail system of large urban center, USA.

Inclusion criteria: i. about to be released from prison; ii. homeless; and iii. seriously mentally ill.
N=140.
Age: mean 35.4.
Sex: 100% M.
Race: 64% black.
Interventions 1. Intensive case management from forensic case manager working individually with CMHC, staff:client
ratio ~ 1:4. N=43.
2. “Intensive case management using the ACT team approach” - team included 4 case managers & 1.5
psychiatrist equivalents, staff:client ratio ~ 1:10. N=42.
3. Referral to local CMHC. N=55.
Outcomes Imprisonment.
Notes
Risk of bias
Test-Wisconsin
Methods Allocation: random assignment - no further details.

Follow up: 6,12,18,24 months.
Participants Setting: Madison, Wisconsin, USA.

Inclusion criteria: i. 18-30; ii. resident in Dane County, Wisconsin; iii. schizophrenia, schizoaffective
disorder, or schizotypal personality; iv. < 12 months spent in psychiatric and penal institutions.
N = 112.
Age: median 23.1 years.
Sex: 32.8% F.
Race: 4.1% black.
History: mean number previous admissions ~ 3.4.
Interventions 1. “ACT”, staff:client ratio ~ 1:9. N=75.

2. Routine care from Dane County psychiatric services - included an unspecified degree of case manage-
ment. N=47.
Outcomes Hospital admisson.

Days in hospital.
Lost to follow-up.
Deaths.
Imprisoned.
Unable to use -
Mental State (BPRS, BSI, no data).
Quality of life (SLS, not peer-reviewed scale, no data).
Notes Preliminary report from 12 year ongoing study.

Data reported seperately for first 6 months (regarded as stabiliisation period) - data reported above thus
refers to months 7-24.
Risk of bias
General abreviations
~ - about.
ACT - Assertive Community Treatment.
CMHC - Community mental health centre.
F - female.
F/T - full time.
M - male.
N = number.
OPD - out-patient department.
Schizophrenia - includes “schizophrenia-like” disorders.
SD - standard deviation.
PACT - Program of Assertive Community Treatment .
P/T - part time.
TCL - Treatment in Community Living.
VA - Veterans Administration.
Scales / operational checklists
ACL - Adjective Check List.
ASI - Addiction Severity Index.
BPRS - Brief Psychiatric Rating Scale.
BSI - Brief Symptom Inventory.
CSI - Colorado Symptom Index.
CSQ - Client Satisfaction Questionnaire.
DSM-III-R - Diagnostic Statistical Manual, 3rd Edition, revised.
GAS - Global Assessment Scale.
GSI - Global Severity Index.
IMPS - Inpatient Multidimensional Psychiatric Scale.
PSE - Present State Examination.
PSNAS - Personality and Social Network Adjustment Scale.
QOLS - Quality of Life Scale.
REHAB - a scale of social functioning.
RSES - Rosenberg Self Esteem Scale.
SAS - Social Adjustment Scale.
SCRS - Short Clinical Rating Scale.
SLS - Satisfaction with Life Scale.
Characteristics of excluded studies [ordered by study ID]
Bigelow-Oregon Allocation: not randomised, quasi-experimental design.
Bond-Chicago2 Allocation: not randomised, matched groups design.

Interventions: two types of crisis housing.
Bond-Indiana2 Allocation: not randomised, allocation to ACT and reference group was not random in one of the three
participating centres. The study could be included if separate data can be obtained from the two centres
where randomisation took place.
Borland-Spokane Allocation: not randomised.
(Continued)
Burns-London Allocation: randomised.

Intervention: multi-disciplinary team home treatment, not ACT.
Champney-Ohio Allocation: randomised.

Intervention: all 4 comparison groups received some form of case management, no ACT.
Curtis-New York Allocation: randomised.

Participants: those with serious mental illnesses.
Interventions: Intensive case management versus routine care, not ACT.
Dean-Birmingham1 Allocation: not randomised.
Dean-Birmingham2 Allocation: not randomised, quasi-experimental study.
Dharwadkar-Victoria Allocation: not randomised, before and after design.
Fenton-Montreal Allocation: randomised.

Intervention: not clearly case management or ACT.
Ford-London Allocation: randomised.

Participants: those with severe mental illnesses.
Interventions: intensive community support versus routine care, not ACT.
Franklin-Houston Allocation: randomised.

Interventions: case management versus routine care, not ACT.
Glick-New York Allocation: randomised.

Intervention: day hospital care vs out-patient group therapy, not ACT.
Goering-Toronto Allocation: not randomised, historical controls.
Herz-New York Allocation: randomised.

Interventions: brief hospitalization vs standard hospital care, not ACT.
Holloway-London Allocation: randomised.

Interventions: case management, not ACT.
Hornstra-Kansas Allocation: not randomised, historical controls.
Hoult-Sydney Allocation: randomised.

Interventions: ACT vs acute admission to a psychiatric hospital, this hospital diversion ACT will be a focus
of another review.
Jerrell-SCarolina Allocation: randomised.

Interventions: ACT vs 12 step recovery program and behavioural skills training, no standard care control
group.
(Continued)
Knight-Los Angeles Allocation: not randomised, quasi-experimental design.
Kuldau-California Allocation: randomised.

Interventions: rapid discharge vs hospital care, not ACT.
Langsley-Denver Allocation: randomised.

Interventions: out patient family crisis management vs hospital admission.
Lehman-Maryland Allocation: randomised.

Interventions: intensive case management (non-ACT, staff/patient ratio 1:15) vs lower intensity case man-
agement (staff/patient ratio 1:25).
Macias-Utah Allocation: randomised.

Interventions: case management versus psychosocial rehabilitation programme at CMHC, not ACT.
Marshall-Oxford Allocation: randomised.

Martin-Delaware Allocation: unclear if randomised.

Interventions: ACT.
McFarlane-New York Allocation: unclear if randomised.

Interventions: ACT vs ACT plus family support (FACT), no standard care group.
McGowan-California Allocation: unclear if randomised, control and treatment groups were “randomly selected” from a population
already receiving ACT or standard care.
McGrew-Indiana Allocation: not randomised, before and after design.
Merson-London Allocation: randomised.

Interventions: multi disciplinary team home treatment vs emergency assessment at hospital, no standard
care group.
Modcrin-Kansas Allocation: randomised.

Interventions: strengths model of case management vs standard case management, not ACT.
Mosher-San Francisco Allocation: not randomised, alternative assignment.
Muijen-London1 Allocation: randomised.

Interventions: ACT vs acute admission to a psychiatric hospital, this hospital diversion ACT will be a focus
of another review.
Muijen-London2 Allocation: randomised.

Interventions: case management versus care from CPNs in primary care, not ACT.
(Continued)
Pai-Bangalore Allocation: not randomised, alternative assignment.
Polak-Denver Allocation: randomised.

Interventions: admission to small “community-based therapeutic environments” vs standard hospital care.
Reibel-Manhattan Allocation: randomised.

Interventions: brief hospital admission, not ACT.
Rossler-Mannheim1 Allocation: not randomised, case control study.
Rossler-Mannheim2 Allocation: not randomised, case control study.
Santiago-Arizona Allocation: randomised.

Participants: those with serious mental illnesses.
Interventions: case management versus standard care, not ACT.
Solomon-Philadelphi1 Allocation: randomised.

Interventions: one type of case management vs another, not ACT.
Stein-Madison Allocation: randomised.

Interventions: ACT versus acute admission to a psychiatric hospital, this hospital diversion ACT will be a
focus of another review.
Susser-New York Allocation: randomised.

Interventions: “Critical time intervention”, neither case managment nor ACT - it is a time-limited approach
aimed at stabilising the patient’s social support network.
Teague-New Hamps Allocation: not randomised.
Thornicroft-Baltim Allocation: not randomised.
Toro-New York Allocation: randomised.

Participants: a minority of the participants suffered from severe mental illness, around 80% were simply
homeless.
Interventions: “intensive case management”.
Outcomes: follow-up rate below acceptable levels.
Tyrer-London Allocation: randomised.

Vincent-Cleveland Allocation: not randomised, alternative assignment.
Wood-New Zealand Allocation: not randomised, case control study.
ACT - Assertive Community Treatment
CMHC - Community Mental Health Centre
CPN - Community psychiatric nurse
DATA AND ANALYSES
Comparison 1. ACT vs STANDARD CARE
No. of No. of
Outcome or subgroup title studies participants Statistical method Effect size
1 Number lost to follow up 10 1597 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.51 [0.40, 0.65]
2 Death (all causes) 5 691 Peto Odds Ratio (Peto, Fixed, 95% CI) 1.13 [0.35, 3.68]
3 Admitted to hospital during 6 1047 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.59 [0.45, 0.78]
study
4 Admissions - duration Other data No numeric data
5 Trouble with the police 6 Peto Odds Ratio (Peto, Fixed, 95% CI) Subtotals only
5.1 arrest during study 2 604 Peto Odds Ratio (Peto, Fixed, 95% CI) 1.17 [0.60, 2.29]
5.2 imprisonment during 4 471 Peto Odds Ratio (Peto, Fixed, 95% CI) 1.19 [0.70, 2.01]
study
5.3 police contact (includes 2 149 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.76 [0.32, 1.79]
arrests)
6 Not living independently at end 3 362 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.46 [0.29, 0.74]
of study
7 Homeless during or at end of 3 374 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.24 [0.11, 0.51]
study
8 Mean days per month in stable 1 152 Mean Difference (IV, Fixed, 95% CI) -4.20 [-7.06, -1.34]
accommodation
9 Unemployed at end of study 2 604 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.31 [0.19, 0.50]
10 Mental state at about 12 3 255 Mean Difference (IV, Fixed, 95% CI) -0.14 [-0.36, 0.08]
months
10.1 Brief Psychiatric Rating 1 58 Mean Difference (IV, Fixed, 95% CI) -0.90 [-7.70, 5.90]
Scale (high score = poor)
10.2 Brief Symptom 1 72 Mean Difference (IV, Fixed, 95% CI) 0.06 [-0.27, 0.39]
Inventory (high score = poor)
10.3 Colorado Symptom 1 125 Mean Difference (IV, Fixed, 95% CI) Not estimable
Index (low score = poor)
11 Social functioning at about 12 3 206 Mean Difference (IV, Fixed, 95% CI) 0.03 [-0.28, 0.34]
months
11.1 Social Adjustment Scale 1 58 Mean Difference (IV, Fixed, 95% CI) 0.10 [-0.58, 0.78]
(low score = poor)
11.2 Personality and Social 1 66 Mean Difference (IV, Fixed, 95% CI) 0.03 [-0.32, 0.38]
Network Adjustment Scale
(low score = poor)
11.3 Social Adjustment Scale 1 82 Mean Difference (IV, Fixed, 95% CI) -3.30 [-7.78, 1.18]
(adapted version, low score =
poor)
12 Satisfaction with care (Client 2 120 Mean Difference (IV, Fixed, 95% CI) -0.56 [-0.77, -0.36]
Satisfaction Questionnnaire,
low score = poor)
13 Self esteem (Rosenberg Scale, 1 66 Mean Difference (IV, Fixed, 95% CI) 0.17 [-0.05, 0.39]
low score = poor)
14 Quality of life (general 1 125 Mean Difference (IV, Fixed, 95% CI) -0.52 [-0.99, -0.05]
well-being in Quality of Life
Scale, low scores = poor)
15 Costs Other data No numeric data
15.1 inpatient care Other data No numeric data
15.2 all health care Other data No numeric data
Comparison 2. ACT vs HOSPITAL-BASED REHABILITATION
No. of No. of
1 Numbers lost to follow up 3 226 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.82 [0.42, 1.57]
2 Admitted to hospital during 2 185 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.20 [0.11, 0.38]
study
4 Trouble with the police 1 Peto Odds Ratio (Peto, Fixed, 95% CI) Subtotals only
4.1 arrest during study 1 120 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.24 [0.11, 0.56]
4.2 imprisonment during 1 120 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.25 [0.05, 1.28]
study
5 Not living independently at end 2 106 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.19 [0.08, 0.42]
of study
6 Unemployed at end of study 2 161 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.30 [0.16, 0.57]
7 Costs (total) Other data No numeric data
Comparison 3. ACT vs CASE MANAGEMENT
No. of No. of
1 Numbers lost to follow up 1 28 Peto Odds Ratio (Peto, Fixed, 95% CI) Not estimable
3 Trouble with the police 1 85 Peto Odds Ratio (Peto, Fixed, 95% CI) 2.66 [1.14, 6.20]
(imprisonment during study)
4 Mean days per month in stable 1 110 Mean Difference (IV, Fixed, 95% CI) -7.7 [-12.64, -2.76]
accommodation
5 Mental state at about 12 months 1 90 Mean Difference (IV, Fixed, 95% CI) -1.31 [-2.98, 0.36]
(Brief Psychiatric Rating Scale,
high score = poor)
6 Social functioning at about 12 1 82 Mean Difference (IV, Fixed, 95% CI) -1.25 [-5.50, 3.00]
months (Social Adjustment
Scale, low score = poor)
7 Satisfaction with care (Client 1 90 Mean Difference (IV, Fixed, 95% CI) -0.48 [-0.71, -0.25]
Satisfaction Questionnnaire,
low score = poor)
8 Self esteem (Rosenberg Scale, 1 90 Mean Difference (IV, Fixed, 95% CI) -0.05 [-0.25, 0.15]
low score = poor)
9 Costs Other data No numeric data
9.1 inpatient care Other data No numeric data
9.2 all health care Other data No numeric data
9.3 total costs Other data No numeric data
Analysis 1.1. Comparison 1 ACT vs STANDARD CARE, Outcome 1 Number lost to follow up.
Review: Assertive community treatment for people with severe mental disorders
Comparison: 1 ACT vs STANDARD CARE
Outcome: 1 Number lost to follow up
Study or subgroup Treatment Control Peto Odds Ratio Weight Peto Odds Ratio
n/N n/N Peto,Fixed,95% CI Peto,Fixed,95% CI
Aberg-Stockholm 2/20 3/20 1.7 % 0.64 [ 0.10, 4.07 ]
Audini-London 3/33 4/32 2.4 % 0.71 [ 0.15, 3.34 ]
Bond-Chicago1 11/44 19/43 7.6 % 0.43 [ 0.18, 1.04 ]
Bond-Indiana1 18/84 25/83 12.2 % 0.64 [ 0.32, 1.27 ]
Chandler-California 44/252 72/264 34.2 % 0.57 [ 0.38, 0.86 ]
Hampton-Chicago 12/80 15/82 8.6 % 0.79 [ 0.35, 1.80 ]
Herinckx-Portland 26/116 29/58 12.8 % 0.28 [ 0.14, 0.55 ]
Lehman-Baltimore 10/77 17/75 8.5 % 0.52 [ 0.23, 1.19 ]
Morse-St Louis1 15/52 29/64 10.4 % 0.50 [ 0.24, 1.06 ]
Test-Wisconsin 1/73 4/45 1.7 % 0.16 [ 0.03, 1.00 ]
Total (95% CI) 831 766 100.0 % 0.51 [ 0.40, 0.65 ]

Total events: 142 (Treatment), 217 (Control)
Heterogeneity: Chi2 = 6.65, df = 9 (P = 0.67); I2 =0.0%
Test for overall effect: Z = 5.42 (P < 0.00001)
0.1 0.2 0.5 1 2 5 10
Analysis 1.2. Comparison 1 ACT vs STANDARD CARE, Outcome 2 Death (all causes).
Outcome: 2 Death (all causes)
Audini-London 0/33 1/33 9.1 % 0.14 [ 0.00, 6.82 ]
Bond-Chicago1 1/45 0/43 9.1 % 7.07 [ 0.14, 356.56 ]
Bond-Indiana1 0/29 1/32 9.1 % 0.15 [ 0.00, 7.53 ]
Hampton-Chicago 2/82 1/82 27.0 % 1.96 [ 0.20, 19.15 ]
Test-Wisconsin 2/75 5/237 45.7 % 1.29 [ 0.22, 7.42 ]
Total (95% CI) 264 427 100.0 % 1.13 [ 0.35, 3.68 ]

Test for overall effect: Z = 0.20 (P = 0.84)
0.1 0.2 0.5 1 2 5 10
Analysis 1.3. Comparison 1 ACT vs STANDARD CARE, Outcome 3 Admitted to hospital during study.
Outcome: 3 Admitted to hospital during study
Audini-London 9/33 9/33 6.5 % 1.00 [ 0.34, 2.93 ]
Bond-Chicago1 32/45 34/43 8.2 % 0.66 [ 0.25, 1.72 ]
Bond-Indiana1 12/50 33/53 12.6 % 0.21 [ 0.10, 0.47 ]
Test-Wisconsin 15/75 26/47 12.8 % 0.21 [ 0.10, 0.45 ]
Total (95% CI) 532 515 100.0 % 0.59 [ 0.45, 0.78 ]

Heterogeneity: Chi2 = 18.78, df = 5 (P = 0.002); I2 =73%
0.1 0.2 0.5 1 2 5 10
Analysis 1.4. Comparison 1 ACT vs STANDARD CARE, Outcome 4 Admissions - duration.
Admissions - duration
Audini-London 1. ACT: mean 3.47 days / month.

2. Standard care: mean 3.4 days / month.
Difference: +2.0% (negative favours ACT).
Bond-Chicago1 1. ACT: mean 3.22 days / month.

Difference: -39.2% (negative favours ACT).
Bond-Indiana1 1. ACT: mean 2.23 days / month.

Difference: -66.8% (negative favours ACT) - statistically significant difference but not based on correct
test.
Chandler-California 1. ACT: mean 0.24 days / month.

Hampton-Chicago 1. ACT: mean 2.38 days / month.

Lehman-Baltimore 1. ACT: mean 2.95 days / month.

Difference: -47% (negative favours ACT) - statistically significant difference but not based on correct test.
Quinlivan-California 1. ACT: mean 1.09 days / month.

Difference: -80.2% (negative favours ACT) - statistically significant difference but not based on correct
test.
Rosenheck-USA-10site 1. ACT: mean 7.66 days / month.

Difference: -32.9% (negative favours ACT) - statistically significant difference based on correct test.
Analysis 1.5. Comparison 1 ACT vs STANDARD CARE, Outcome 5 Trouble with the police.
Outcome: 5 Trouble with the police
1 arrest during study

Bond-Chicago1 3/45 1/43 11.4 % 2.69 [ 0.37, 19.77 ]
Subtotal (95% CI) 297 307 100.0 % 1.17 [ 0.60, 2.29 ]

2 imprisonment during study
Bond-Chicago1 1/45 2/43 5.3 % 0.48 [ 0.05, 4.76 ]
Hampton-Chicago 2/82 3/82 8.9 % 0.66 [ 0.11, 3.92 ]
Solomon-Philadelphi2 25/42 22/55 43.8 % 2.17 [ 0.97, 4.82 ]
Test-Wisconsin 19/75 14/47 41.9 % 0.80 [ 0.35, 1.81 ]
Subtotal (95% CI) 244 227 100.0 % 1.19 [ 0.70, 2.01 ]

3 police contact (includes arrests)
Bond-Chicago1 4/45 11/43 59.7 % 0.31 [ 0.10, 0.94 ]
Bond-Indiana1 7/29 3/32 40.3 % 2.88 [ 0.75, 11.09 ]
Subtotal (95% CI) 74 75 100.0 % 0.76 [ 0.32, 1.79 ]

Test for subgroup differences: Chi2 = 0.81, df = 2 (P = 0.67), I2 =0.0%
0.1 0.2 0.5 1 2 5 10
Analysis 1.6. Comparison 1 ACT vs STANDARD CARE, Outcome 6 Not living independently at end of
study.
Outcome: 6 Not living independently at end of study
Bond-Chicago1 33/45 37/43 21.1 % 0.46 [ 0.16, 1.29 ]
Test-Wisconsin 20/75 25/47 39.5 % 0.32 [ 0.15, 0.69 ]
Total (95% CI) 197 165 100.0 % 0.46 [ 0.29, 0.74 ]

0.1 0.2 0.5 1 2 5 10
Analysis 1.7. Comparison 1 ACT vs STANDARD CARE, Outcome 7 Homeless during or at end of study.
Outcome: 7 Homeless during or at end of study
Bond-Chicago1 1/45 2/43 11.5 % 0.48 [ 0.05, 4.76 ]
Hampton-Chicago 0/82 15/82 54.0 % 0.11 [ 0.04, 0.32 ]
Test-Wisconsin 5/75 5/47 34.5 % 0.59 [ 0.16, 2.23 ]
Total (95% CI) 202 172 100.0 % 0.24 [ 0.11, 0.51 ]

0.1 0.2 0.5 1 2 5 10
Analysis 1.8. Comparison 1 ACT vs STANDARD CARE, Outcome 8 Mean days per month in stable
accommodation.
Outcome: 8 Mean days per month in stable accommodation
Study or subgroup Treatment Control Mean Difference Weight Mean Difference

N Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI
Lehman-Baltimore 77 -17.5 (9) 75 -13.3 (9) 100.0 % -4.20 [ -7.06, -1.34 ]
Total (95% CI) 77 75 100.0 % -4.20 [ -7.06, -1.34 ]
Heterogeneity: not applicable

-10 -5 0 5 10
Analysis 1.9. Comparison 1 ACT vs STANDARD CARE, Outcome 9 Unemployed at end of study.
Outcome: 9 Unemployed at end of study
Bond-Chicago1 43/45 41/43 5.6 % 1.05 [ 0.14, 7.71 ]
Total (95% CI) 297 307 100.0 % 0.31 [ 0.19, 0.50 ]

0.1 0.2 0.5 1 2 5 10
Analysis 1.10. Comparison 1 ACT vs STANDARD CARE, Outcome 10 Mental state at about 12 months.
Outcome: 10 Mental state at about 12 months

1 Brief Psychiatric Rating Scale (high score = poor)

Audini-London 30 41.4 (14) 28 42.3 (12.4) 0.1 % -0.90 [ -7.70, 5.90 ]
Subtotal (95% CI) 30 28 0.1 % -0.90 [ -7.70, 5.90 ]

2 Brief Symptom Inventory (high score = poor)
Morse-St Louis1 37 0.95 (0.76) 35 0.89 (0.65) 45.7 % 0.06 [ -0.27, 0.39 ]
Subtotal (95% CI) 37 35 45.7 % 0.06 [ -0.27, 0.39 ]

3 Colorado Symptom Index (low score = poor)
Lehman-Baltimore 67 -4.1 (0.83) 58 -3.8 (0.87) 54.2 % -0.30 [ -0.60, 0.00 ]
Subtotal (95% CI) 67 58 54.2 % -0.30 [ -0.60, 0.00 ]

Total (95% CI) 134 121 100.0 % -0.14 [ -0.36, 0.08 ]
Test for subgroup differences: Chi2 = 2.59, df = 2 (P = 0.27), I2 =23%
-10 -5 0 5 10
Analysis 1.11. Comparison 1 ACT vs STANDARD CARE, Outcome 11 Social functioning at about 12
months.
Outcome: 11 Social functioning at about 12 months

1 Social Adjustment Scale (low score = poor)

Audini-London 30 3 (1.6) 28 2.9 (1) 21.1 % 0.10 [ -0.58, 0.78 ]
Subtotal (95% CI) 30 28 21.1 % 0.10 [ -0.58, 0.78 ]

2 Personality and Social Network Adjustment Scale (low score = poor)
Morse-St Louis1 37 -2.88 (0.75) 29 -2.91 (0.71) 78.4 % 0.03 [ -0.32, 0.38 ]
Subtotal (95% CI) 37 29 78.4 % 0.03 [ -0.32, 0.38 ]

3 Social Adjustment Scale (adapted version, low score = poor)
Jerrell-SCarolina2 40 -80.8 (10.66) 42 -77.5 (10.01) 0.5 % -3.30 [ -7.78, 1.18 ]
Subtotal (95% CI) 40 42 0.5 % -3.30 [ -7.78, 1.18 ]

Total (95% CI) 107 99 100.0 % 0.03 [ -0.28, 0.34 ]
Test for subgroup differences: Chi2 = 2.16, df = 2 (P = 0.34), I2 =7%
-10 -5 0 5 10
Analysis 1.12. Comparison 1 ACT vs STANDARD CARE, Outcome 12 Satisfaction with care (Client
Satisfaction Questionnnaire, low score = poor).
Outcome: 12 Satisfaction with care (Client Satisfaction Questionnnaire, low score = poor)

Audini-London 28 -3.33 (0.56) 26 -2.86 (0.4) 61.3 % -0.47 [ -0.73, -0.21 ]
Morse-St Louis1 37 -3.37 (0.49) 29 -2.66 (0.78) 38.7 % -0.71 [ -1.03, -0.39 ]
Total (95% CI) 65 55 100.0 % -0.56 [ -0.77, -0.36 ]

-10 -5 0 5 10
Analysis 1.13. Comparison 1 ACT vs STANDARD CARE, Outcome 13 Self esteem (Rosenberg Scale, low
score = poor).
Outcome: 13 Self esteem (Rosenberg Scale, low score = poor)

Morse-St Louis1 37 -1.88 (0.49) 29 -2.05 (0.43) 100.0 % 0.17 [ -0.05, 0.39 ]
Total (95% CI) 37 29 100.0 % 0.17 [ -0.05, 0.39 ]

-10 -5 0 5 10
Analysis 1.14. Comparison 1 ACT vs STANDARD CARE, Outcome 14 Quality of life (general well-being in
Quality of Life Scale, low scores = poor).
Outcome: 14 Quality of life (general well-being in Quality of Life Scale, low scores = poor)

Lehman-Baltimore 67 -4.69 (1.35) 58 -4.17 (1.33) 100.0 % -0.52 [ -0.99, -0.05 ]
Total (95% CI) 67 58 100.0 % -0.52 [ -0.99, -0.05 ]

-10 -5 0 5 10
Analysis 1.15. Comparison 1 ACT vs STANDARD CARE, Outcome 15 Costs.

Costs
inpatient care
Bond-Indiana1 1. ACT: mean $46.2.

2. Standard care: mean $278.
Difference: -83% (negative figure favours ACT).
Chandler-California Site 1 only (Long Beach). Site 2 only (Stanislaus).

1. ACT: mean $20.8. 1. ACT: mean $44.0.
2. Standard care: mean $59.7. 2. Standard care: mean $69.9.
Jerrell-SCarolina2 1. ACT: mean $243.

Quinlivan-California 1. ACT: mean $69.5.

Rosenheck-USA-10site 1. ACT: mean $977.

all health care
Bond-Indiana1 1. ACT: mean $230.

Costs (Continued)
Chandler-California Site 1 only (Long Beach). Site 2 only (Stanislaus).

1. ACT: mean $367. 1. ACT: mean $422.
2. Standard care: mean $113. 2. Standard care: mean $152.
Jerrell-SCarolina2 1. ACT: mean $448*.

2. Standard care: mean $471*.
* only mental health costs, not medical costs.
Quinlivan-California 1. ACT: mean $182.

Rosenheck-USA-10site 1. ACT: mean $1229. The statistical significance of all above differences
2. Standard care: mean $1489. either not reported or based on incorrect statistical
Difference: -18% (negative figure favours ACT). analysis (usually the application of a parametric test
to skewed untransformed data).
Analysis 2.1. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 1 Numbers lost to

follow up.
Comparison: 2 ACT vs HOSPITAL-BASED REHABILITATION
Outcome: 1 Numbers lost to follow up
De Cangas-Quebec 17/60 18/60 69.8 % 0.92 [ 0.42, 2.02 ]
Lafave-Ontario 4/24 9/41 27.5 % 0.72 [ 0.21, 2.52 ]
Marx-Madison 0/21 1/20 2.8 % 0.13 [ 0.00, 6.50 ]
Total (95% CI) 105 121 100.0 % 0.82 [ 0.42, 1.57 ]

0.1 0.2 0.5 1 2 5 10
Analysis 2.2. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 2 Admitted to
hospital during study.
Outcome: 2 Admitted to hospital during study
De Cangas-Quebec 11/60 31/60 71.6 % 0.23 [ 0.11, 0.49 ]
Lafave-Ontario 13/24 37/41 28.4 % 0.14 [ 0.04, 0.44 ]
Total (95% CI) 84 101 100.0 % 0.20 [ 0.11, 0.38 ]

0.1 0.2 0.5 1 2 5 10
Analysis 2.3. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 3 Admissions -

duration.
Lafave-Ontario 1. ACT: mean 1.62 days / month.

2. Hospital-based rehabilitation: mean 10.66 days / month.
Difference: -84.8% (negative favours ACT), statistically significant on corrct test.
Marx-Madison 1. ACT: mean 1.27 days / month.

2. Hospital-based rehabilitation: mean 20.61 days / month.
Analysis 2.4. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 4 Trouble with the
police.
Outcome: 4 Trouble with the police
1 arrest during study

De Cangas-Quebec 7/60 23/60 100.0 % 0.24 [ 0.11, 0.56 ]
Subtotal (95% CI) 60 60 100.0 % 0.24 [ 0.11, 0.56 ]

2 imprisonment during study
De Cangas-Quebec 1/60 5/60 100.0 % 0.25 [ 0.05, 1.28 ]
Subtotal (95% CI) 60 60 100.0 % 0.25 [ 0.05, 1.28 ]

Test for subgroup differences: Chi2 = 0.00, df = 1 (P = 0.98), I2 =0.0%
0.1 0.2 0.5 1 2 5 10
Analysis 2.5. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 5 Not living

independently at end of study.
Outcome: 5 Not living independently at end of study
Lafave-Ontario 12/24 33/41 55.9 % 0.24 [ 0.08, 0.72 ]
Marx-Madison 4/21 14/20 44.1 % 0.13 [ 0.04, 0.45 ]
Total (95% CI) 45 61 100.0 % 0.19 [ 0.08, 0.42 ]

0.1 0.2 0.5 1 2 5 10
Analysis 2.6. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 6 Unemployed at
end of study.
Outcome: 6 Unemployed at end of study
De Cangas-Quebec 16/60 33/60 77.1 % 0.31 [ 0.15, 0.65 ]
Marx-Madison 12/21 17/20 22.9 % 0.27 [ 0.07, 1.02 ]
Total (95% CI) 81 80 100.0 % 0.30 [ 0.16, 0.57 ]

0.1 0.2 0.5 1 2 5 10
Analysis 2.7. Comparison 2 ACT vs HOSPITAL-BASED REHABILITATION, Outcome 7 Costs (total).

Costs (total)
De Cangas-Quebec 1. ACT: mean $281*, SD 139.

2. Hospital-based rehabilitation: mean $488*, SD 299.
No Log transformation necessary, p<0.05, ACT 42% less expensive.
*?Canadian dollars.
Analysis 3.1. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 1 Numbers lost to follow up.
Comparison: 3 ACT vs CASE MANAGEMENT
Outcome: 1 Numbers lost to follow up
Study or subgroup Treatment Control Peto Odds Ratio Peto Odds Ratio
Bush-Atlanta 0/14 0/14 0.0 [ 0.0, 0.0 ]
Total (95% CI) 14 14 0.0 [ 0.0, 0.0 ]

0.1 0.2 0.5 1 2 5 10
Analysis 3.2. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 2 Admissions - duration.
Bush-Atlanta 1. ACT: mean 1.57 days / month.

2. Case management: mean 2.39 days / month.
Difference: -34.3% (negative favours ACT), statistically significant diffence but not based on correct test.
Essock-Connecticut 1. ACT: mean 2.31 days / month.

Difference: -46.9% (negative favours ACT), statistically significant diffence but not based on correct test.
Quinlivan-California 1. ACT: mean 1.09 days / month.

Analysis 3.3. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 3 Trouble with the police
(imprisonment during study).
Outcome: 3 Trouble with the police (imprisonment during study)
Solomon-Philadelphi2 25/42 15/43 100.0 % 2.66 [ 1.14, 6.20 ]
Total (95% CI) 42 43 100.0 % 2.66 [ 1.14, 6.20 ]

0.1 0.2 0.5 1 2 5 10
Analysis 3.4. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 4 Mean days per month in stable
accommodation.
Outcome: 4 Mean days per month in stable accommodation

Morse-St Louis2 55 -23.7 (11.4) 55 -16 (14.8) 100.0 % -7.70 [ -12.64, -2.76 ]
Total (95% CI) 55 55 100.0 % -7.70 [ -12.64, -2.76 ]

-10 -5 0 5 10
Analysis 3.5. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 5 Mental state at about 12 months
(Brief Psychiatric Rating Scale, high score = poor).
Outcome: 5 Mental state at about 12 months (Brief Psychiatric Rating Scale, high score = poor)

Morse-St Louis2 45 8.2 (3.55) 45 9.51 (4.5) 100.0 % -1.31 [ -2.98, 0.36 ]
Total (95% CI) 45 45 100.0 % -1.31 [ -2.98, 0.36 ]

-10 -5 0 5 10
Analysis 3.6. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 6 Social functioning at about 12
months (Social Adjustment Scale, low score = poor).
Outcome: 6 Social functioning at about 12 months (Social Adjustment Scale, low score = poor)

Jerrell-SCarolina2 40 -80.8 (10.66) 42 -79.55 (8.82) 100.0 % -1.25 [ -5.50, 3.00 ]
Total (95% CI) 40 42 100.0 % -1.25 [ -5.50, 3.00 ]

-10 -5 0 5 10
Analysis 3.7. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 7 Satisfaction with care (Client
Satisfaction Questionnnaire, low score = poor).
Outcome: 7 Satisfaction with care (Client Satisfaction Questionnnaire, low score = poor)

Morse-St Louis2 45 -3.4 (0.59) 45 -2.92 (0.51) 100.0 % -0.48 [ -0.71, -0.25 ]
Total (95% CI) 45 45 100.0 % -0.48 [ -0.71, -0.25 ]

-10 -5 0 5 10
Analysis 3.8. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 8 Self esteem (Rosenberg Scale, low
score = poor).
Outcome: 8 Self esteem (Rosenberg Scale, low score = poor)

Morse-St Louis2 45 -1.89 (0.48) 45 -1.84 (0.51) 100.0 % -0.05 [ -0.25, 0.15 ]
Total (95% CI) 45 45 100.0 % -0.05 [ -0.25, 0.15 ]

-10 -5 0 5 10
Analysis 3.9. Comparison 3 ACT vs CASE MANAGEMENT, Outcome 9 Costs.

Costs
inpatient care

2. Case management: mean $340.
Morse-St Louis2 1. ACT: mean $263.

Quinlivan-California 1. ACT: mean $69.5.

all health care

Difference: -11.8% (negative figure favours ACT).

Difference: 0%.
Quinlivan-California 1. ACT: mean $182.

Difference: -27.3% (negative figure favours ACT).
total costs
Costs (Continued)

Difference: +9.8% (negative figure favours ACT).
WHAT’S NEW
Last assessed as up-to-date: 24 February 1998.
15 February 2010 Amended Contact details updated.
HISTORY
Protocol first published: Issue 2, 1998
Review first published: Issue 2, 1998
21 October 2008 Amended Converted to new review format.
25 February 1998 New citation required and conclusions have changed Substantive amendment
DECLARATIONS OF INTEREST
None known.
SOURCES OF SUPPORT
Internal sources
• Manchester University Department of Psychiatry, UK.
External sources
• Nuffield Trust, UK.
NOTES
Cochrane Schizophrenia Group internal peer review complete (see Module).
External peer review scheduled.
INDEX TERMS
Medical Subject Headings (MeSH)

∗ Community Mental Health Centers; ∗ Community Mental Health Services; Mental Disorders [∗ therapy]
MeSH check words

Humans

Tac Chocrane

Uploaded by

Copyright:

Available Formats

You might also like

Tac Chocrane

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tac Chocrane

Uploaded by

Copyright:

Available Formats

Assertive community treatment for people with severe

mental disorders (Review)

Assertive community treatment for people with severe

Max Marshall1 , Austin Lockwood2

Editorial group: Cochrane Schizophrenia Group.

ACT versus standard community care

ACT versus hospital-based rehabilitation services

ACT versus case management

PLAIN LANGUAGE SUMMARY

Assertive community treatment for people with severe mental disorders

ACT is often confused with case management, a superficially sim- OBJECTIVES

Secondary outcomes Electronic searches

It is difficult to understand why total costs should favour case man-

References to studies included in this review Lafave-Ontario {published data only}

Essock-Connecticut {published data only} Rosenheck-USA-10site {published data only}

Jerrell-SCarolina2 {published data only} Solomon-Philadelphi2 {published data only}

Dharwadkar-Victoria {published data only}

Fenton-Montreal {published data only}

Characteristics of included studies [ordered by study ID]

Methods Allocation: randomised - no further details.

Participants Setting: Kungsholmen sectorised service, Stockholm, Sweden.

Outcomes Lost to follow up.

Item Authors’ judgement Description

Allocation concealment? Unclear B - Unclear

Methods Allocation: randomised - no further details.

Participants Setting: London, UK.

treatment arm of a trial of ACT (see Muijen-London).

Outcomes Hospital admission.

Item Authors’ judgement Description

Allocation concealment? Unclear B - Unclear

Methods Allocation: randomised, sealed envelope, independent secretary performed assignment.

Participants Setting: Chicago, Illinois, USA.

Interventions 1. ACT. N=45.

Outcomes Hospital admission.

Item Authors’ judgement Description

Allocation concealment? Yes A - Adequate

Methods Allocation: random allocation - no further details.

Participants Setting: 3 CMHCs, Indiana, USA.

Outcomes Hospital admission.

Item Authors’ judgement Description

Allocation concealment? Unclear B - Unclear

Methods Allocation: random assignment - no further details.

Participants Setting: Atlanta, USA.

Interventions 1. ACT (Madison model), staff:client ratio ~ 1:2. N=14.

Outcomes Lost to follow up.

Notes * Not defined further.

Item Authors’ judgement Description

Allocation concealment? Unclear B - Unclear

Methods Allocation: randomised - no further details.

Outcomes Lost to follow up.

Item Authors’ judgement Description

Allocation concealment? Unclear B - Unclear

Methods Allocation: random allocation - no further details.

Participants Setting: Quebec, Canada.

Outcomes Hospital admission.

Item Authors’ judgement Description

Allocation concealment? Unclear B - Unclear

Methods Allocation: “randomly assigned” - no further details.