Effects of an assertive community program in patients with severe mental disorders and impact on their families


Philippe Huguelet, MD, Département de Santé Mentale et de Psychiatrie, Hôpitaux Universitaires de Genève, Rue du 31-Décembre 8, 1207 Geneva, Switzerland. Email: philippe.huguelet@hcuge.ch


Aims:  Assertive community treatment (ACT) is known to have a positive impact on the number and length of inpatient stays. However, recent studies have shown little or no effect of such programs in European settings. This paper aims to describe the impact of a newly implemented ACT program on patients and their families' burden. Predictive factors have also been examined.

Method:  Fifty-five patients characterized by heavy use of psychiatric care, numerous hospitalizations, or failure to link with outpatient psychiatric care and their relatives were followed. Data were gathered on patients before and after follow up as well as on relatives' burden and costs. The number and domains of clinician interventions have been detailed.

Results:  The ACT program had a positive effect on symptoms, psychosocial adaptation and quality of life. Age was the most significant predictor of changes. Older patients, most of them suffering from delusional disorders, showed no improvement or even some impairment. Finally, the program appeared to have a marked effect on easing families' burden in domains such as assistance in daily life activities and costs.

Conclusions:  ACT appears to be recommended for patients with poor outcome when treated in other settings. Early intervention seems to be justified as highlighted by younger age being the best predictor of positive changes. Families can be helped considerably, particularly those confronted with patients with persistent disturbing symptoms which do not, however, warrant hospitalization. Finally, the fact that patients with delusional disorder do not seem to improve warrants further research.

MANY STUDIES DESCRIBE the clinical improvement brought about by assertive community treatment (ACT) in patients with severe mental disorders.1 ACT appears to have a positive impact on the number and length of inpatient stays. Subjects become more autonomous and are more stable in their home,2 and homelessness is reduced.3 Despite the large number of studies on the effect of ACT, some questions remain.

First, it is necessary to assess the impact of ACT programs in various settings to confirm their usefulness. Indeed, recent studies have shown little or no effect of such programs in England4 and the Netherlands.5 Yet, in Lausanne, Switzerland, a retrospective study6 has shown a positive impact of an ACT program on hospitalization rate and clinical and social outcomes. Also, the establishment of ACT in low-income countries may be beneficial.7 These apparently contradictory findings may be related to the characteristics of patients included, and/or to the comparative ambulatory treatments offered, which may in some places be better than the ‘older care’ which was offered as a control treatment in the older trials.5

Second, little is known about the factors altering the outcome in the patients followed by ACT teams. It is important to understand which patient characteristics influence the effect of ACT, as knowledge of these characteristics may help to determine which patients could be preferentially included in ACT treatments.

Third, the nature and number of mobile interventions needed to influence patients' outcomes remain to be established.8

Fourth, research may help to understand how ACT programs may influence families' burden, e.g. in terms of economic expenditures.9 In research performed in Japan, Sono et al.10 found that more than 70% of families with patients with mental illness reported needs related to ACT. Yet little is known about how specifically ACT may act on families' burden.

This papers aims to supply some answers to these questions by describing the effect of a newly implemented ACT program in a prospective cohort of patients with severe mental disorders, in terms of symptoms, psychosocial functioning, quality of life and hospitalization rate. Predictive factors have been examined, as well as the effect of the program on families' burden and satisfaction.



A mobile team has been implemented in Geneva, Switzerland since October 2007. Its goals are to reach difficult-to-engage patients and to provide them with care as long as needed, to avoid hospital admission and relieve symptoms, psychosocial dysfunction and families' burden. Patients suffering from severe and persistent mental illness characterized by heavy use of psychiatric care, repeated hospitalization or failure to link with outpatient psychiatric care have been followed. Particular emphasis was given to including frequently hospitalized patients (the so-called ‘revolving-door’ patients), patients living with their parents who have symptoms which are persistent but not acute enough to warrant non-voluntary admission and who are not generally reached by care providers and homeless patients. Fidelity to the ACT model was assessed using the Dartmouth index,11 with a score of 3.82 corresponding to a moderate degree of implementation.

The multidisciplinary team comprises three full-time-equivalent (FTE) nurses, one FTE social worker, one FTE psychologist and 0.5 FTE psychiatrists. The program's main characteristics are home-based interventions, a 1:10 caseload ratio, 5 working days per week, individual caseloads with team backup, assertive outreach and focus on engagement, and systematic contact with relatives and other caregivers.


The patients included came from the catchment area of Geneva, which has 450 000 inhabitants. Data were gathered for all patients from the beginning of the program. The study was approved by the ethics committee of the University Hospitals of Geneva. Subjects (both patients and caregivers) who had signed a detailed written consent form participated in the study. Recruitment began on 1 October 2007 and ended on 20 August 2009.


Data-gathering took place during the first appointments and sometimes required some weeks because of the reluctance of some patients. A second assessment took place between 6 and 8 months after the initial contact. Families were contacted after patients' consent, and assessed both at inclusion and after 6 months' follow up. The number of days of hospitalization was calculated for the preceding year and for the entire follow-up period.



Patients were rated using the Brief Psychiatric Rating Scale (BPRS).12 Social functioning was assessed with the Health of the Nation Outcome Scales (HONOS),13 and Multnomah Community Ability Scale (MCAS)14,15 and quality of life with the World Health Organization Quality of Life (WHOQOL).16 The working alliance was assessed by the Working Alliance Inventory (WAI)17 and satisfaction with care with the Verona Service Satisfaction Scale (VSSS-54) client version.18 Drug and alcohol consumption was assessed with the Alcohol Use Scale (AUS), Drug Use Scale (DUS) and Substance Abuse Treatment Scale (SATS).19,20 Patients' compliance with pharmacological treatment was assessed with the Medication Adherence Rating Scale (MARS)21 and recovery from mental illness with the Recovery Assessment Scale (RAS).22

The activities of clinicians with patients and their families were measured with a daily contact log, the Relevé Quotidien des Contacts (RQC),8 which covers the location, time and actors involved in the episode of care and the nature of the intervention (crisis, representing, accompanying, discussing) in 10 areas (i.e. medication, daily living activities, housing, relationships, substance abuse, legal, etc.).


Patients' families were evaluated with the Family Experience Interview Schedule (FEIS)23 and the Psychiatric Symptoms Index (PSI),24 a screening tool that assesses a range of symptoms, such as depression, cognitive disturbance, anxiety, and anger. The direct and indirect costs for the families arising from the care of their relative were also measured,9 as was the time spent by relatives for both daily living and acute management. The adequacy and availability of the network was assessed on the basis of Guay's five modalities: ‘adequate and available,’‘exhausted and overwhelmed,’‘inactive and unstable,’‘inadequate and incompetent,’ and ‘absent and nonexistent’.25

Statistical analyses

Linear mixed models26,27 with random individual effects were used to evaluate the effects of ACT on each of the continuous measures. These models were refitted with additional fixed effects of linear and quadratic functions of time, age, sex, and baseline severity of each individual scale to analyze predictors of response to ACT. The fixed effect of treatment length was used to estimate changes in the effect sizes over the study period. The results of regression models are presented as standardized regression coefficients (β) with 95% confidence intervals which can be interpreted as effect size. Predictors of dropout for the family members during follow up were assessed using Cox proportional hazard regression with adjustment for age and sex. All analyses were performed using stata release 10 (StataCorp LP, College Station, TX, USA).



Of the 91 patients consecutively treated by the ACT program for at least 3 months, 55 consented to inclusion in the research. Twenty-one allowed us to contact their families (out of 37 who had relatives in the area). Table 1 presents the baseline clinical and demographic characteristics of the 55 subjects.

Table 1.  Baseline demographic and clinical characteristics (n = 55)
Age at onset of disorder24.011.6
Number of hospitalizations prior to admission6.99.0
Length of hospitalization prior to admission (days)211.1297.6
Number of hospitalizations in the year preceding the admission1.61.7
Length of hospitalization in the year preceding the admission (days)55.471.7
Average years of education11.79.8
Way of living  
 With family1120.0
 In couple59.1
Civil status  
 Has children1832.7
 Has a legal representative2240.0
 Has a disability pension3869.1
 Has no occupation5192.7
DSM-IV diagnoses  
 Schizophrenia, paranoid type1730.9
 Schizophrenia, undifferentiated type712.7
 Other psychotic disorders1425.5
 Schizoaffective disorder47.3
 Bipolar and major depressive disorders1323.6
Axis I and II comorbidities  
 2 or more1120.0
 Current substance abuse/dependence1527.3

Follow up and predictors of response

The mean number and time of intervention per individual was 80.9 interventions (SD = 60.1) and 3700 min (SD = 2714.7), respectively, for the whole follow-up period. Most of the interventions were conducted with the clients themselves (n = 57.7; SD = 46.7), followed by interventions with professionals from the ambulatory and psychiatric ward facilities (n = 18.3; SD = 15.8), interventions with the legal representative/social workers (n = 13.7; SD = 13.0), with the family (n = 10; SD = 12.1) and finally with the citizens (n = 2.8; SD = 4.1). Most of the interventions were performed to treat symptoms (n = 76.4; SD = 56.2), for medication (n = 57.2; SD = 50.5), for physical health (n = 38.6; SD = 37.8) and finally for financial problems (n = 30.6; SD = 32.4) (sum of mean scores is more than the total score, because more than one task was sometimes performed during one intervention).

After adjustment for the time spent during follow up most of the studied variables evolved favorably. The highest effects (effect sizes greater than 1) were observed for BPRS total score (from 74.3 [SD = 21.3] to 49.4 [SD = 17.6]) and its subscale BPRS agitation-mania from 3.5 (SD = 1.3) to 2.0 (SD = 0.8) (b = −1.08, P < 0.0001 and b = −1.10, P < 0.0001, respectively), for HONOS scores (from 20.5 [SD = 6.0] to 11.0 [6.0], b = −1.09, P < 0.0001), for the Analog scale (b = 1.11, P < 0.0001), for three of the VERONA scales and for all the MCAS subscales and total score (Table 2).

Table 2.  Changes in clinical and social variables over the treatment period
  • *

    Significant results are bold-faced.

  • AUS, Alcohol Use Scale; BPRS, Brief Psychiatric Rating Scale; DUS, Drug Use Scale; HONOS, Health of the Nations Outcome Scales; MARS, Medication Adherence Rating Scale; MCAS, Multnomah Community Ability Scale; RAS, Recovery Assessment Scale; SATS, Substance Abuse Treatment Scale; VSSS-54F, Verona Service Satisfaction Scale – 54 items; WAI, Working Alliance Inventory; WHOQOL, World Health Organization Quality of Life.

BPRS global score74.321.349.417.6−1.08<0.0001
BPRS positive symptoms2.91.31.91−0.78<0.0001
BPRS negative symptoms31.42.21.1−0.63<0.0001
BPRS agitation-mania3.51.320.8−1.1<0.0001
BPRS depression-anxiety3.−0.82<0.0001
WHOQOL physical20.76.924.55.30.590.0004
WHOQOL psychological17.55.920.
WHOQOL social relations7.
WHOQOL environmental24.36.3295.10.74<0.0001
WAI patient60.115.565.29.50.420.032
WAI therapist53.112.859.112.90.450.001
SATS alcohol0.−0.10.512
SATS drugs1.−0.220.178
MCAS interference with functioning92.719.53.51.72<0.0001
MCAS adjustment to living7.62.8112.61.09<0.0001
MCAS social competence6.62.114.841.58<0.0001
MCAS behavior problems9.83.5163.51.32<0.0001
MCAS global score44.99.261.310.91.27<0.0001
Network Assessment3.−0.99<0.0001
Psychiatrists' skills and behavior
Psychologists' skills and behavior17.86.921.37.60.430.117
Nursing staff skills and behavior21.27.828.84.20.93<0.0001
Social workers' skills and behavior19.67.727.94.41.03<0.0001
Information and access to services8.12.811.11.71.06<0.0001
Efficacy of services144.619.32.11.11<0.0001
Relatives' involvement7.
Relatives' involvement

When the BPRS total score was considered after adjustment for baseline scores, age, sex and length of follow up, the best predictor of response was younger age (b = 0.13, P = 0.006). Few or no interventions with the family as assessed by the RQC during follow up were a significant correlate of poor response (b = 0.22, P = 0.039). Indeed most subjects whose psychotic signs and symptoms worsened did not have any intervention with the family (n = 5/9, 55.6%) and the remaining one had only one contact during the entire follow up. Interestingly, all of these subjects except one were suffering from a disorder other than paranoid schizophrenia, mainly delusional disorder. No other baseline characteristics predicted outcome. It is important to note that the patients in question were older, which may explain why there were no family members to contact in some cases or in others consent to our meeting relatives was refused.

With regard to HONOS scores, no baseline characteristics significantly predicted outcome. We observed a tendency for a low number of hospitalizations in the year prior to follow up to be a poor predictor of response (b = −0.11; P = 0.079).

No other baseline characteristics predicted outcome for the other scales with an effect size greater than 1. With the exception of intervention with the families, the nature and number of interventions as assessed by the RQC did not correlate with outcome measures.

There was a non-significant decrease in the percentage of days spent in hospital during follow up compared to the year prior to follow up (from 17.0% [SD = 21.6] to 12.1% [SD = 19.3], b = −0.23, P = 0.204). However, when looking at subjects with more than 10% (50th percentile) days spent in hospital in the year prior to follow up (n = 24), there was a highly significant decrease in the percentage during follow up (from 34.3% [SD = 19.7] to 13.4% [SD = 22.1], b = −1.02, P = 0.0002).


Most of the family members were women (77.2%) or the patient's mother/father (63.6%). Mean age was 51 years (SD = 15.3). Table 3 shows the baseline and end-scores for variables describing family involvement in the patient's care.

Table 3.  Baseline and end-scores for variables describing family involvement in the patient's care
Financial burden**−0.710.026
Number of financial costs2.−0.210.515
Financial costs (CHF)967.91581.1380.0726.4−0.560.031
Frequency of assistance in daily life15.−0.690.013
Inconvenience of this assistance8.−0.660.003
Frequency of supervision4.−0.390.04
Inconvenience of this supervision4.−0.360.125
Number of crises observed6.612.
Frequency of observing crises2.
  • *

    Significant results are bold-faced.

  • **

    Score ranging from 0 = none to 3 = severe.

  • CHF, Swiss franc; PSI, Psychiatric Symptoms Index.

Has observed a crisis1482.3872.2−0.160.31

Eight family members dropped out during the study. The only predictor of dropout was female sex. All the dropouts were female.

Several variables evolved favorably during follow up. The best changes were observed for the inconvenience relating to assistance in daily life and the PSI. We did not find any predictors of these changes.

It is worth noting that good response in terms of BPRS total score and HONOS scores was significantly correlated with the family's reporting that it was less exposed to the patient's crises (b = −0.26; P = 0.0005 and b = −0.29; P < 0.0001).


The following results can be emphasized on the basis of the present report: first, this study shows the extensive impact of the ACT program; second, this effect manifested in many parameters involving both administered scales and self reports; third, age was a significant predictor of response to ACT as older patients, most of whom were suffering from delusional disorders, showed no improvement or even deterioration in outcome in some cases; fourth, the nature and number of interventions provided to the patients by their clinicians did not appear to influence outcome and finally, the program had a great effect on families' burden in various domains.

The effects of ACT would appear to be evident in this study, even if it was conducted in an area characterized by a high level of ambulatory care, in terms of numbers of both medical and social facilities. These results contradict those of the REACT study4 and the NDL study5 showing no significant effect of ACT in terms of clinical and social outcome for the former and no effect on hospitalization rate for the latter. This could be due to the fact that in our program, only patients characterized by a large number of hospitalizations and/or frequent former treatment refusal were included. Hence, in contrast to most of our patients who were likely to improve because of the severity of their awful conditions, patients in the English and Dutch studies may not have been in need of the specific potentials of ACT. This hypothesis is confirmed by the fact that BPRS total score was 74 at inclusion in our sample compared to about 40 in both other studies and is supported by the results of Bak et al.28 who showed that patients with higher BPRS scores were more likely to reach symptomatic remission when treated by ACT. The case of hospitalizations as an outcome measure is a particularly good example of possible misinterpretation of results: globally, the number of days spent in hospital does not seem to be altered in our sample, as it is in the REACT study.4 When the data are examined more closely, however, it appears logical that only the subgroup frequently hospitalized improved. Concomitantly, a substantial fraction of patients were in too poor a condition to be treated by outpatient facilities; hence they stayed at home or on the streets, thus avoiding hospitalizations. The same pattern has been shown with homeless severely ill patients, whose higher hospitalization when using ACT could be considered a success.3 Overall our findings and the above-mentioned literature suggest that ACT is effective through clinicians' assertiveness. It allows clinicians to reach and treat patients who are not willing to go to outpatient facilities, furthermore, to help these patients solve social issues. Conversely patients with more favorable profiles (i.e. aware of their need of care and without comorbidities, such as substance abuse) can be followed in regular outpatient settings.

Age was the only predictor found in our study. Other studies found substance abuse, higher age and lower educational level predictive of worse outcome,29 as well as a diagnosis of paranoid schizophrenia.30 The fact that older patients had a poor outcome may have been due to a longer duration of illness, as commented by Kortrijk et al.29 It could also relate to an a priori bias in the selection process: older patients referred to ACT were obviously resistant to conventional care. In contrast, younger patients show a more open perspective, despite being at that time in a similar clinical context. Another important point is that all these patients except one suffered from delusional disorder, that is, a difficult-to-treat condition.31 This appears to be an important finding, as many of these difficult patients had been proposed for ACT care in this program. The lack of improvement is likely to be due to the fact that none of these patients accepted medication, despite repeated attempts on the part of their clinicians. Overall this finding emphasizes the need to find evidence-based ways of treatment, with both medications and psychosocial interventions.

Analyses of the role of interventions in patients' outcome, as assessed by the RQC,8 did not bring notable results, beyond descriptive data. This could relate to the small sample size included in this kind of analysis, but also to a chicken-and-egg causality dilemma: interventions are likely to improve patients' conditions, yet at the same time patients with multiple and resistant problems are more likely to get a higher amount of sustained attention.

Families' burden was considerably improved with the ACT treatment. Families' relief was associated with patients' improvement in terms of symptoms, and there was no correlation between the number of interventions with families and their level of relief. In addition, families' relief may have been related to the fact that the ACT team helped patients with issues that had to date been dealt with by families, such as assisting in getting food or in developing social relationships.32 The literature underscores the burden associated with having a sibling suffering from a severe condition requiring ACT care.9,33 Hanzawa et al.33 confirmed the familiar notion that the families' burden was correlated with patients' status, in terms of social function and care needs. Sono et al.10 associated burden with the number of supportive behaviors involved for families. Yet we know of no studies which assessed the effect of ACT on the family's burden, even if most families underscore the need for this kind of program.10


Generalization of these results may be hampered by the small sample size, the open study design and the high refusal rate for participation. Also, the follow-up period was quite short. In relation to the latter issue, Kortrijk et al.29 showed that improvement occurred mainly in the first months of treatment with ACT. The lack of a control group makes it impossible to state whether the observed improvements are exclusively explained by participation in the program or reflect a natural evolution in the psychiatric disorder. However these subjects suffered from severe and chronic disease and improvement was observed not only by clinicians but also by families and the subjects themselves.


Currently, long-term follow up shows that at least 80% of patients with severe mental disorders would need mobile interventions, such as ACT at one time or another.34 The ACT programs devoted to patients with difficult profiles (i.e. with current substance abuse and/or dissocial personality traits and/or lack of insight) consistently showed their efficacy, as clinical intuition would suggest. Hence ACT should be recommended for patients with the worst outcomes, that is, characterized by multiple treatment interruptions, substance abuse and poor ability to represent their suffering from a neuropsychiatric disorder. This even applies in areas where ambulatory care is fairly well developed.

The population likely to be helped is of younger age and has schizophrenia rather than delusional disorder. Families can also be substantially helped, particularly those coping with patients whose symptoms do not remit but whose clinical state is not impaired enough to warrant hospitalization. In this context, ACT can be a substitute for the care provided by families, and progressively reintroduce treatment and help with social issues. Our results support the integration of assessment of the family in trials on ACT effectiveness.

The particular case of older patients with delusional disorder warrants attention. These are patients who often refuse everything although they need the most, both in terms of treatment and measures to prevent aggression and/or self-damaging behaviors. In this study, ACT did not appear to offer substantial help, at least in statistical terms. More research is needed to obtain knowledge as to how to intervene on this population.


This study was partially supported by a private grant (Mr A. Engelhorn).