The rise of assertive community interventions in South Africa: A randomized control trial assessing the impact of a modified assertive intervention on readmission rates; a three year follow-up

BMC Psychiatry (Impact Factor: 2.21). 02/2014; 14(1):56. DOI: 10.1186/1471-244X-14-56
Source: PubMed
ABSTRACT
Background
Many countries have over the last few years incorporated mental health assertive interventions in an attempt to address the repercussions of deinstitutionalization. Recent publications have failed to duplicate the positive outcomes reported initially which has cast doubt on the future of these interventions. We previously reported on 29 patients from a developing country who completed 12 months in an assertive intervention which was a modified version of the international assertive community treatment model. We demonstrated reduction in readmission rates as well as improvements in social functioning compared to patients from the control group. The obvious question was, however, if these outcomes could be sustained for longer periods of time. This study aims to determine if modified assertive interventions in an under-resourced setting can successfully maintain reductions in hospitalizations.

Methods
Patients suffering from schizophrenia who met a modified version of Weidens’ high frequency criteria were randomized into two groups. One group received a modified assertive intervention based on the international assertive community treatment model. The other group received standard care according to the model of service delivery in this region. Data was collected after 36 months, comparing readmissions and days spent in hospital.

Results
The results demonstrated significant differences between the groups. Patients in the intervention group had significantly less readmissions (p = 0.007) and spent less days in hospital compared to the patients in the control group (p = 0.013).

Conclusion
Modified assertive interventions may be successful in reducing readmissions and days spent in hospital in developing countries where standard care services are less comprehensive. These interventions can be tailored in such a way to meet service needs and still remain affordable and feasible within the context of an under-resourced setting.

Full-text

Available from: Niehaus DJ, Jul 07, 2014
RES E AR C H A R T I C L E Open Access
The rise of assertive community interventions in
South Africa: a randomized control trial assessing
the impact of a modified assertive intervention
on readmission rates; a three year follow-up
Ulla A Botha
1*
, Liezl Koen
1
, Ushma Galal
2,3
, Esme Jordaan
2
and Daniel JH Niehaus
1
Abstract
Background: Many countries have over the last few years incorporated mental health assertive interventions in an
attempt to address the repercussions of deinstitutionalization. Recent publications have failed to duplicate the positive
outcomes reported initially which has cast doubt on the future of these interventions. We previously reported on 29
patients from a developing country who completed 12 months in an assertive intervention which was a modified
version of the international assertive community treatment model. We demonstrated reduction in readmission rates as
well as improvements in social functioning compared to patients from the control group. The obvious question was,
however, if these outcomes could be sustained for longer periods of time. This study aims to determine if modified
assertive interventions in an under-resourced setting can successfully maintain reductions in hospitalizations.
Methods: Patients suffering from schizophrenia who met a modified version of Weidens high frequency criteria were
randomized into two groups. One group received a modified assertive intervention based on the international assertive
community treatment model. The other group received standard care according to the model of service delivery in this
region. Data was collected after 36 months, comparing readmissions and days spent in hospital.
Results: The results demonstrated significant differences between the groups. Patients in the intervention group had
significantly less readmissions (p = 0.007) and spent less days in hospital compared to the patients in the control group
(p = 0.013).
Conclusion: Modified assertive interventions may be successful in reducing readmissions and days spent in hospital in
developing countries where standard care services are less comprehensive. These interventions can be tailored in such
a way to meet service needs and still remain affordable and feasible within the context of an under-resourced setting.
Keywords: Assertive interventions, Developing countries, Readmission rates, Days in hospital
Background
Assertive Community Treatment (ACT) is by now a well-
known approach that has been adopted in many countries
[1-3]. Initially, one of the most attractive motivators for
the incorporation of this approach in mental health service
delivery, was its apparent success in reducing readmis -
sion rates in so-called revolving door patients. Though
considered an expensive interve ntion, the costs were
justified under the premise that inpatient cost s were
generally much higher. The approach has been well
researched and tested over the last twenty years , with
many countries reporting on a range of outcomes , such
as readmission rates, patient satisfaction, degrees of
symptomatology and social functioning [4-6,3,2,7]. Initial
studies from particularly the US and Australia, reported
positive outcomes in most of these areas, which prompted
UK decision makers to launch 300 Assertive Outreach
teams nationwide [1-3]. Though some UK studies initially
demonstrated favourable outcomes, few studies reported
reduction in readmission rates. A number of recent UK
* Correspon dence: ullabotha@gma il.com
1
Department of Psychiatry, Faculty of Medicine and Health Sciences,
University of Stellenbosch, PO Box 19063, Tygerberg 7505, South Africa
Full list of author informa tion is available at the end of the article
© 2014 Botha et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly credited.
Botha et al. BMC Psychiatry 2014, 14:56
http://www.biomedcentral.com/1471-244X/14/56
Page 1
studies have demonstrated no benefit from ACT interven-
tions compared with the CMHT control groups and have
concluded that the approach may no longer be justifiable,
considering the cost [1,8]. The discrepancy in findings
by different research groups and countries has created
considerable controversy [1,9,10].
Burns et al. performed a meta-regression with 64 trials
in an attempt to identify factors that may contribute to
outcome s. They foun d that a high baseline days in
hospital (DIH) was often associated with higher reduc-
tion and that teams with high fidelity to the AC T
model, also appeared to have lower readm ission rates.
One of the common criticisms has been that control
groups have been poorly defined [1]. Clearly, standard
care or treatment a s usual is quite non-specific , a s
standard care services vary considerably within countries
and even more so between countries. Several of the stud-
ies that demonstrated no significant improved outcomes,
reported that the standard care services appeared to have
incorporated many of the s alient fe atures of the ACT
approach, such a s fixed ca seloads (though larger than
those in ACT ), ho me visits and ass ertive follow-up
[11,12]. One study even reported a standard care service
with a higher fidelity score than the AC T group it w as
being compared with [13].
Despite the apparent downfall of the ACT approach in
the UK, it is still being employed with success in other
countries. A recent Germa n study in Hamburg demon-
strated reduced inpatient days in patients followed-up
assertively for 12 months and concluded that the inter-
vention wa s more cost-effective than the standard care
service it was compared with. Though outpatient cost s
had been higher in the inter vention group, the tota l
cost was still lower due to the significantly more expensive
inpatient costs [6]. Similarly, a recent Danish study dem-
onstrated patients rec eiving an assertive intervention
for two years, had less s ubstance use, better adherence
to medication and were more satisfied with their treat-
ment. In addition to this, they also reported significantly
lower usage of inpatient services compared to the control
group [5].
Developing countries face the same challenges of re-
volving door patient s and bed pressures, but have the
additional burden of limited resources and lack of
funding to c ontend with. Hanlon et a l. reported that
only 56.5% of African countries h ave community-ba sed
mental health ser vices and only 50% have existing mental
health policies [14]. One of the important recommenda-
tions from this publication was the need for strengthening
of specialist mental health services and further integration
of mental health service with primary health service. Patel
et al. called for scaling up of cost-effective community
based mental-health services in middle and low-income
countries citing successes in countries such as India, Chile
and China, where interventions had been modified to
meet the resources and needs of the community [15].
Odenwald et al. reported on such an intervention in
Somalia, which offered a 10 month programme to a group
of 35 outpatients with chronic psychotic disorders. The
intervention was a home-based programme which incor-
porated psycho-education, relapse prevention and family
support and was found to be cost-effective and feasible in
a low-income country [16].
In Sou th-Africa, similar attempts have been made to
address the challenges in finding a cost-effective commu-
nity-based initiative. A s part of a provincial initiative,
the Western Cape Province launched three Assertive
Community Treatment (ACT) teams in 2007. The teams
followed a modified version of the ACT model, par-
ticularly in t erms of case loa ds and visit frequency. We
reported that at the one year follow-up the patients
who completed the intervention demonstrated significant
reduction in days spent in hospital and improvements in
socialfunctioningincomparisontopatientsreceiving
the standard care ser vice package. Though the follow-up
period was only 12 months, these were the first indicators
that assertive interventions could be successfully modified
to meet the needs of under-resources areas without
compromising the efficacy of the intervention [17]. This
supports past comments by international authors [1,18]
that assertive interventions may be more effective in
under-resourced areas where standard care services are
less comprehensive.
The important question, however, remains whether
positive outcomes can be sustained over time. It is well-
known that newly established services may initially have
good outcomes due to staff enthusiasm and initial smaller
caseloads, but that these outcomes often tail off over time
as burn-out ensues and pressure rises [11].
Aim
The purpose of this study was to determine if modified
assertive interventions in an under-resourced setting can
successfully maintain reductions in hospitalizations over
a 36 month follow-up period.
Methods
This study was conducted in Stikland Hospital, one of
the three large state mental health hospitals in Cape
Town, South Africa. The hospital, along with two others,
provides inpatient services to the whole of the Western
Cape Province covering a population of approximately
5 million people. The combined b ed capacity for acute
psychotic patient s in the three hospitals is 500. The
Stikland Hospital ACT team initially consisted of a full-
time psychiatrist, a social worker and a chief professional
nurse, but has been expanded over time. Currently the
team consists of a medical officer, a social worker, three
Botha et al. BMC Psychiatry 2014, 14:56 Page 2 of 10
http://www.biomedcentral.com/1471-244X/14/56
Page 2
chief psychiatric nurses and a psychiatrist. The team
has access to an occupational therapist, a dual diagnosis
service and a PSR-based day program.
All s ubjec ts who presented for admission over an
eighteen month period and who had a previously estab-
lished, documented diagnosis of schizophrenia or schizo-
affective disorder (DSM-IV-TR) were eligible for inclusion.
To be included as high frequency users (HFUs), subjects
had to fulfil the inclusion criteria, which was modified
from Weidens HFU-criteria (see List of criteria below) to
accommodate local admission patterns and ensure the
appropriate service users were targeted [19]. S ubjects
were excluded if they had (1) a serious, unstable co-morbid
medical illness that could interfere with their ability to
participate in the intervention; (2) were unable to give
written informed consent or (3) if another co-morbid
Axis I or II diagnosis other than schizophrenia or schizo-
affective disorder was the current focus of treatment.
List of criteria: Modified Weidiens criteria for differ-
entiation high frequency (HFU) and low frequency
(LFU ) schizop hren ia-s pe c trum disord er users of psy -
chiatric services:
General criteria
1) Schizophrenia or Schizo-affective Disorder
2 Age 1859 years (extremes included)
3 Needs current treatment with antipsychotic
Must meet General Criteria PLUS either (A) or (B) or
(C) to be included
(A) 3 admissions in 18 months/ 5 in 36 months
(B) 2 admissions in 12 months AND treated with
clozapine
(C) 2 admissions in 12 months AND 120 days in
hospital
HFUs had to fulfi ll General Criteria PLUS one of A; B or.
The study was approved by the research ethics commit-
tees of both the Universities of Stellenbosch and Cape
Town. The research component constituted approxi-
mately half of the caseload of the ACT t eam w herea s a
non-research component provided the same intervention
to high frequency users (HFUs) with other diagnoses. Re-
search numbers therefore do not reflect overall caseloads.
The trial took the form of a randomized, non-blinded
parallel group study. Subjects (n = 65) identified as HFUs
who provided informed, written consent, were considered
for inclusion. Randomization was done using standardized
tables, patients were allocated to one of two treatment
groups (See Figure 1).
Subjects from both groups, a) treatment as usual (n = 26)
and b) Intervention group (n = 34), received visit s at
inclusion, pr ior to discharge and after 12 months. At
36 month follow-up, data wa s colle cted from patient
folders or from patients directly where no readmissions
had been documented. The same method was used for
both groups to obtain readmission information. Admis-
sions d ata is ea sily accessible on the provincial data
systems and also indicates admissions to other hospitals in
the province. Where no readmissions were documented,
patients or families were contacted to rule out any out
of area admissions or adverse events. Where patients or
families could not be reached, the community mental
health practitioner was contacted to obtain information.
Information was collected about number of readmissions,
number of days spent in hospital at each admission,
months in service (control or intervention respectively),
number of days until first admission, number of admissions
to intermediate care facility, adverse events and demo-
graphics were confirmed. Data was collected over a 72
month period; 36 months prior to date of inclusion (pre-
DOI) and 36 months post -date of inclusion (post-DOI).
Subjects from the treatment as usual group were dis-
charged into the existing standard care system. At 36
month follow-up data was collected from subject files
and directly from subjects where files did not provide
sufficient information.
Subjects from the intervention group were each
assigned a key worker in the form of a senior social
worker or a chief professional nurse. Key workers started
engaging subject s and carers prior to discharge with the
primary focus on building a therapeutic relatio nship.
The nature of the intervention was tailored as close as
possible to the international model of assertive commu-
nity treatment, with the two main exceptions in the size
of caseloads and frequency of visits. It was agreed at the
onset that the caseloads carried by international teams
would not be realistic in the context of a pressured system
in an under-resourced developing country. A con sens us
of 80 patients per team was reached, with individual
caseloads not exceeding 3 5. Fidelity to the international
model was assessed with the Dartmouth Assertive Com-
munity Treatment Scale (DACTS) with a total score of 3.1
[20]. Key workers acted as main care coordinators but
caseloads were often shared between members of the
team. A major focus in the team was on engageme nt
and maintaining compliance on medication. The team
also attempted to make use of existing resources in the
community in addition to the service provided by the
team. This may be considered a minor deviation from
the ACT model where care is coordinated solely by the
ACT team, but may be a practical option where teams
need to spread themselves thinly. F requency of patient
contacts wa s individualized according to patient need
with minimum of fortnightly contact s by any member
of the team. Patients had access to occupational therapy
and psychology services although no full-time staff was
available from these disciplines. The majority of contacts
(>50%) were in the community, mainly home visits.
Botha et al. BMC Psychiatry 2014, 14:56 Page 3 of 10
http://www.biomedcentral.com/1471-244X/14/56
Page 3
The team was based at Stikland Hospital, one of the
three state mental hospitals in the Western Cape. This
had both advantages and disadvantages since the team
was able to draw from the various resources in the hospital
setting to strengthen the service it provided, such as access
to a day programme offering psycho-social rehabilitation.
The team also acted as bridge between hospital-based care
and community mental health services, offering valuable
liaison and streamlining communication between services.
Readmissions were treated like adverse events; the team
would liaise with inpatient staff and commence follow-up
of patient upon discharge without any change to the
follow-up period. Incarcerations were not considered in
the same manner as admissions, since incarceration does
not necessarily imply that appropriate psychiatric care is
given. One intervention patient was incarcerated during
the 36 month period but the team continued to perform
visits in prison. One patient from the control group had
been incarcerated during the 36 month period.
Patients in this catchment area have access to an inter-
mediate rehabilitation facility. This unit functions as a
step-up/step-down fac ility and offers a psycho-social
rehabilitation (PSR) program. Data was collected separately
on the number of patients who attended this program
since this may impact on readmission rates. Patients from
both groups had access to the facility but these admissions
were not considered i n the same way as admissions to the
acute wards.
At 12 month follow-up, additional information was
collected about readmissions and any changes in medi-
cation. Patients in the intervention group remained in
the service and those in the control group were given
the option to be included in the intervention group.
Subsequently, two patients from the control group were
Figure 1 Flow diagram describing allocation of patients.
Botha et al. BMC Psychiatry 2014, 14:56 Page 4 of 10
http://www.biomedcentral.com/1471-244X/14/56
Page 4
included in the intervention group and completed a 36
month follow-up period. There was no official drop-out
policy and none of the intervention patients dropped out
during the first 12 months of the study. Two patients from
the intervention group were discharged after 14 and 16
months of follow-up respectively. In both instances
patients had been following up in the community for
longer than six months without concerns about com-
pliance or indications of relapse. Both patients had
been well integrated with their respe ctive community
ser vices and appeared no longer to require a ssertive
input. One patient died after two months in the study.
Time to readmission for this person was censored at time
of death. Four patients who signed informed consent
initially were referred to long-term wards (See Figure 1).
Patients who were referred to long-term wards prior to
their index visit were excluded. Patients who were re-
admitted and referred to long-term wards after a period
of follow-up, were not excluded and DIH were included
for analysis. Both groups we re considered in the same
way. Service contacts were not measured for the control
group. The spectrum of standard care is very diverse and
contacts are o ften infrequent, depending on which
particular community service patients made use of.
(See Table 1 for comparisons of care received between
the two groups).
Statistical analysis
Data was summarised through counts (n) and frequencies
(%), medians and interquartile ranges (IQR) or means and
standard deviations (sd). Wilcoxon MannWhitney Rank
Sum tests (non-pa rametric) were used to test median
differences while the T-test was used to test differences
in means for normally distributed data. Fisher tests of
association were used for count data.
Time to readmission was considered as the number of
days to first admission after discharge, and was used as
the outcome variable in the Survival analyses methods
employed. Patients who completed a 36 month follow-up
without readmissions were censored since their readmis-
sion history is not known beyond the end of the study.
For Kaplan-Meier methods, data was assumed to be
right-censored. To statistically test whether or not there is
adifferenceintimetoreadmissionbetweenthecases
and controls, and to test for cova riates in the model, a
Cox proportional hazards regression wa s carried out.
The Cox regression model is a non-parametric model
which assumes that the hazard rate is proportional. This
assumption wa s tested graphically and using goodness-
of-fit tests and found to be valid. Hazard ratios were
calculated from the results of the regression. Since the
effect of group membership on readmission was of
interest, separate cur ves were produced so that they
could be compared graphically. All statistical analyses
were carried out using the package R: A Language for
Data Analysis and Graphics [21].
Results
Data were analyzed for 32 patients in the (cases) inter-
vention group and 24 patients in the con trol group. The
baseline demographics from both groups confirmed the
homogeneity of the group for all demographic variables,
except place of residence (Fishers test p-value = 0.01), at
a 5% level of significance (Table 2). There was no missing
data; all patients were successfully located at 36 month
follow-up. The median age of the control group was 27.5
Years (IQR = (23.8, 3 6.8) ye ars) and that for the cases
was 32.0 yea rs (IQR = (26.8, 42.8) years). To test this
difference, the nonparametric Wilcoxon MannWhitney
Rank Sum test was used and showed the difference was
Table 1 Work style of modified ACT team compared to standard care
Modified ACT team Community mental health team
Overall patient load 80-100 patients ± 600 patients excluding assessments of new patients
Individual caseload Maximum 35 250
Workstyle Key workers act as care coordinator but caseloads are shared Individual caseloads
Site of most visits >50% contacts are home visits Office based
Engagement Assertive; focus on engagement, immediate response to
non-compliance
Non-assertive, no follow-up of missed appointments/
reports of non-compliance
Working hours Office hours Office hours
24 hour cover Patients referred to hospital-based after-hours service
coordinated by ACT when in crisis.
After-hours service of catchment area.
Frequency of contacts Individualized according to patient need at least fortnightly Depends on caseloads, varies between monthly to
three monthly.
Disciplines available Full-time psychiatrist, social worker, Psychiatric nurse, access to
psychologist, occupational therapist, dual diagnosis service.
Full-time psychiatric nurse, access to social worker
and psychiatrist, varied access to occupational therapist
and psychologist.
Botha et al. BMC Psychiatry 2014, 14:56 Page 5 of 10
http://www.biomedcentral.com/1471-244X/14/56
Page 5
not statistically significant (p-value = 0.253). See Table 3
for comparisons between the two groups with regards to
days spent in hospital (DIH) and number of admissions
pre-and-post inc lusion. When comparing days s pent in
hospital in the 36 months prior to inclusion (pre-DOI)
in this study ( Table 3), no significant difference w as
Table 2 Demographic differences between cases and controls
Cases Controls Total
Gender Male 22 (69) 19 (79) 41 (73)
Female 10 (31) 5 (21) 15 (27)
Total: 32 (100) 24 (100) 56 (100)
Ethnicity Caucasian 1 (3) 0 (0) 1 (2)
Coloured 30 (94) 21 (88) 51 (91)
Xhosa 1 (3) 3 (13) 4 (7)
Total: 32 (100) 24 (100) 56 (100)
Marital status Single 27 (84) 18 (75) 45 (80)
Married 3 (9) 2 (8) 5 (9)
Divorced 2 (6) 4 (17) 6 (11)
Total: 32 (100) 24 (25) 56 (100)
Language Afrikaans 29 (94) 21 (88) 50 (91)
English 1 (3) 0 (0) 1 (2)
Xhosa 1 (3) 3 (13) 4 (7)
Total: 31 (100) 24 (100) 55 (100)
Employment Unemployed 31 (97) 24 (100) 55 (98)
Casual 1 (3) 0 (0) 1 (2)
Total: 32 (100) 24 (100) 56 (100)
Residence** Metro 32 (100) 19 (79) 51 (91)
Rural 0 (0) 5 (21) 5 (9)
Total: 32 (100) 24 (100) 56 (100)
Accommodation Family 32 (100) 24 (100) 56 (100)
Total: 32 (100) 24 (100) 56 (100)
Highest Level of Education Elementary 14 (44) 5 (21) 19 (34)
Secondary 14 (44) 14 (58) 28 (50)
Matric 4 (13) 4 (17) 8 (14)
None 0 (0) 1 (4) 1 (2)
Total: 32 (100) 24 (100) 56 (100)
Adverse events None 30 (94) 24 (100) 54 (96)
Pregnancy 1 (3) 0 (0) 1 (2)
Death 1 (3) 0 (0) 1 (2)
Total: 32 (100) 24 (100) 56 (100)
Status changes No change 27 (84) 20 (83) 47 (84)
Discharge from Intervention 2 (6) 0 (0) 2 (4)
Included in Intervention 2 (6) 4 (17) 6 (11)
Death 1 (3) 0 (0) 1 (2)
Total: 32 (100) 24 (100) 56 (100)
Disability grant Yes 28 (88) 22 (92) 50 (89)
No 4 (13) 2 (8) 6 (11)
Total: 32 (100) 24 (100) 56 (100)
**Significant difference detected (Fishers test: p-value = 0.01).
Botha et al. BMC Psychiatry 2014, 14:56 Page 6 of 10
http://www.biomedcentral.com/1471-244X/14/56
Page 6
found between the two groups (p-value = 0.376). For
the post date of inclusion data (post-DOI), there was a
significant difference between the ca ses and controls
(p-value = 0.002). To compare the pre and post-DOI
data of the two groups, the difference was c alculated as
pre-DOI minus post-DOI number of days in hospital
for each group separately. A Wilcoxon Ma nnWhitney
rank sum test on th is data yielded a p-value of 0.013,
indicating a significant difference between the two groups.
However, the large confidence interval (CI = (24,177))
indicates lack of precision in the estimation, which is
likely due to a patient in the control group with a long
length of stay post-DOI. The Wilcoxon test was re-
peated without this patient and y ielded a p-value of
0.023 (CI = ( 15, 163)). Thus , one can still conclude that
there is a statistically significant difference between the
pre-minus post-DOI days in hospital between the two
groups.
Table 3 also summarizes the number of readmissions
for each group. The difference between the groups is again
demonstrated using pre-DOI number of admissions
minus post-DOI number of admissions. This data was
normally distributed thus a t-test was used to demonstrate
a difference in the means. It gave a p-value of 0.007, which
indicates a significant difference in the mean pre-minus
post-DOI admissions between cases and controls.
There was no significant difference between the two
groups in ter ms of adm issio ns to the intermedia te
Table 3 Summary - days in hospital and number of admissions, for each group
Cases n = 32; controls n = 24 Mean (SD) % of n with admissions
Days in hospital
Pre-date of inclusion Cases 264.8 (108.0) 100%
Controls 261.5 169.8) 100%
Post-date of inclusion Cases 35.2 (64.4) 40.60%
Controls 51.5 (219.2) 75%
Pre-Post date of inclusion Cases 229.7 (130.2)
Controls 110 (187.6)
Median (IQR) Wilcoxon test
Estimate (Cl) p-value
Pre-date of inclusion Cases 256.0 (174.2, 319.2) 27 (38, 92) 0.376
Controls 202.0 (152.8, 311.5)
Post-date of inclusion Cases 0.0 (0.0, 52.0) 53 (96, 6) 0.002
Controls 88.0 (6.8, 161.2)
Pre-Post date of inclusion Cases 230.0 (147.8, 314.8) 93 (24, 177) 0.013
Controls 130.0 (57.8, 235.0)
Cases n = 32; controls n = 24 Mean (SD)
Number of admissions
Pre-date of inclusion Cases 4 (1.8)
Controls 4 (1.5)
Post-date of inclusion Cases 1.5 (0.8)
Controls 2 (1.3)
Pre-Post date of inclusion Cases 4 (2)
Controls 2 (1.8)
Median (IQR) Wilcoxon test
Estimate (Cl) p-value
Pre-date of inclusion Cases 4.0 (3.0, 5.0) 0 (0, 1) 0.515
Controls 3.0 (3.0, 4.0)
Post-date of inclusion Cases 0.0 (0.0, 1.0) 1(2, 0) 0.001
Controls 2.0 (0.8, 2.3)
*Pre-Post date of inclusion Cases 3.56 (2.0) 1.4 (0.4, 2.5) 0.007
Controls 2.13 (1.8)
*Summarised using mean (sd).
Botha et al. BMC Psychiatry 2014, 14:56 Page 7 of 10
http://www.biomedcentral.com/1471-244X/14/56
Page 7
rehabilitation facility (Fisher test p-value = 0.543). Of
note was that though the number of admissions was
the similar (n = 5 for controls , n = 6 for ca ses) in both
groups, t he inter vention group had two patients with
more than one admission to the facility, whereas all pa-
tients admitted from the control group had single admis-
sions to the facility (See Table 4).
To compare the readmission experience of cases and
controls, Kaplan-Meier (K-M) Survivor Curves were plot-
ted separately for each group (not shown). Two patients
were censored; one patient died two months into the
study (case) and another was included much later than
other patients (control). The K-M plot clearly demon-
strated that in the first 200 days there was little difference
between the two group but after 200 days, the controls
were more likely to be readmitted. The curves also moved
further apart over time, indicating that the intervention
became more beneficial over time. The K-M curves do not
provide statistical evidence of a significant difference. To
demonstrate this , the log-rank test was appli ed and
gave a p-value of 0.027, which demonstrated a statistically
significant difference between the readmission rates of the
two groups. However, we cannot adjust for covariates
using a log-rank test so a Cox Proportional Hazards model
was carried out where age, gender, number of admissions
pre-DOI and days in hospital pre-DOI were all adjusted
for (Table 5). After adjusting for these factors, the model
gave a hazard ratio of 3.0 (CI = (1.4, 6.7)), indicating a
significant difference between admission experience of the
cases and controls. A hazard ratio of 3.0 indicates that
being in the control increases your hazard of readmission
three-fold, on average. Figure 2 shows a plot of this model
with separate curves for the cases and controls. Censored
patients are represented by the crosses on the curves.
Discussion
Reductions in readmission rates are widely accepted to
be an effective way to assess outcome of assertive inter-
ventions and are often used as primary outcome [3].
One criticism against this method has been that this
outcome may have large appeal to managers due to cost-
implications and may not necessarily reflect a positive
outcome for patients. However, combining this outcome
with length of stay (and measurements in degrees of psy-
chopathology) prov ide the necessary reassurances that
the outcomes are not simply being produced by denying
patients access to necessary care, but that the intervention
actually reduces the need for admission [1,3,17].
The results indicate that the reduction in inpatient
days that has been previously reported on, can indeed be
sustained in the long run. This was considered one of
the limitations of the original 12 month follow-up, since
there are several factors that may contribute to short-
term outcomes [17]. Both Sytema and Killaspy reported
on the fact the n ewly establish ed teams may initially
produce positive outcomes that could tail off over time
[8,11]. Newly established teams often start off with high
staff enthusiasm and pressure to succeed and depending
on the recruitment style, initial caseloads may be smaller
than anticipated. Once caseloads increase and the novelty
of the approach wears off, staff may have less time for as
frequent and comprehensive input and could possibly be
less likely to go the extra mile, so to speak. Also, since
most patients were included in the service directly after an
Table 5 Estimated hazard ratios from a Cox
regression model
Hazard ratio (95% CI)
Group 2.43 (1.06, 5.57)
Gender 1.15 (0.46, 2.84)
Age 0.98 (0.94, 1.03)
Days in hospital pre-DOI 1.00 (1.00, 1.00)
Number of admissions pre-DOI 1.02 (0.77, 1.35)
Time to first readmission (days)
Proportion not readmitted
Intervention
Control
Hazard ratio=3.0
0
0.0 0.2 0.4 0.6 0.8 1.0
200 400 600 800 1000
Figure 2 Survival curves for cases and controls from a Cox
proportional hazards model.
Table 4 Summary of admissions to intermediate
rehabilitation facility
Cases Controls Total
Number of admissions
to rehab facility
n (%) n (%) n (%)
0 27 (84) 19 (79) 46 (82)
1 3 (9) 5 (21) 8 (14)
2 1 (3) 0 (0) 1 (2)
3 1 (3) 0 (0) 1 (2)
Total 32 (100) 24 (100) 56 (100)
Botha et al. BMC Psychiatry 2014, 14:56 Page 8 of 10
http://www.biomedcentral.com/1471-244X/14/56
Page 8
admission, one may speculate that this would reduce the
number of readmissions in 12 months in itself. In addition
to this, global pressures on inpa tient beds mean that
patients generally have to be quite ill in order to warrant
admission, which in turn means that patients who are
not behaviourally disturbed and pose less of a risk, may
be managed in the community. These factors necessitate
the use of a control group to adequately assess the effect-
iveness of such an intervention.
Demonstrating sustained reduction in readmissions over
a 36 month period, means that the new team factor is
less likely to contribute. By 18 months, all key workers
were managing case loads of approximately 30 patients or
more. In addition to this, one may speculate that having
successfully demonstrated significantly reduced days in
hospital, the team may have felt less pressured to sustain
the le vel of input that had been given thus far. Surely, if
new team enthusiasm and pressure to succeed may
contribute to a teams ability to produce positive outcomes,
thelackthereofmaybeexpectedtohavetheopposite
effect. This wa s not t he case howe ver, despite the team
experiencing other typical phenomenon described by
established teams, such a s staff-burn-out.
Days to first admission reflects the number of days
from initial inclusion to first readmission. This does not
necessarily reflect how stable the patients were during this
time. Previous publications by our group have demon-
strated higher degrees of symptomatology in the control
group [17]. Clinica l experience has shown that patient s
followed-up in a standa rd care setting, are seen less
frequently by mental health practitioners and a re often
left to cope with significantly higher degrees of symptom-
atology prior to admission. Also, their access to inpatient
services may be less streamlined than that of patients
followed-up in an assertive intervention, where staff often
facilitates admissions directly. This would contribute to
the mean dura tion of stay which ha s been shown to be
significantly longer in controls than in the inter vention
group. Another contributing factor in this regard, may
be the fact that the particular inpatient unit s these pa-
tients had access to are continuously under significant
pressure, with a crisis discharge policy often prompting
early discharges to create beds for patient who are more
ill. However, it is likely that patients who are followed-up
by an intervention service may be less likely to be dis-
charged early, since their key workers may intervene on
their behalf and request that they be optimally stabilized.
It is importa nt to note that the intervention group still
had 18 readmissions. Keeping in mind that all patients
met a HFU criteria at inclusion, it is inevitable that some
patients will require readmission no matter how effective
and comprehensive the intervention is and that the aim
is not to avoid readmission at all cost. In fact, in some
cases, a readmission may provide a necessary time-out
for both patient and staff to revisit the treatment plan
and refine the therapeutic relationship.
Though a number of recent international publications
have raised serious doubts about the future of ACT teams,
the approach has provided a large body of research evidence
and has been vital in the development of new services [1].
While the evidence clearly indicates that benefits may
be limited and cost unjustifiable in settings where
standard care services are well-resourced and able to
provide comprehensive care, the contrary may be true
in under-resourced area s. In fact , our result s indic ate
that even modified versions of the original approach,
with significantly larger caseloads and less frequent
visits , can successfully reduce inpatient u sage in high
frequency patients. Once again, this may reflect more on
the nature of standard care than the efficacy of the inter-
vention. Certainly there is the hope that even in under-
resourced settings an approach such as this will influence
the way standard care is delivered and that over time some
of the salient features of the intervention will be incorpo-
rated into standard care practice, as has been the case in
other settings.
Conclusion
Assertive inter ventions ca n successfully be modified in
under-resourced settings and sustain reductions in in-
patient usage over time, while still remaining afford-
able and fea sible within the context of a developing
country. Such inter ventions need n ot be exclusive and
limited to a small number of patient s but can be suc-
cessfully incorporated into existing services and tai-
lored according to the needs of the c ommunity and
resources available.
Limitations
Single-site studies on the effectiveness of ACT tend to
have small sample sizes (range 41 to 64) [6,7,11]. The
reason for this may vary from country to country, but in
a developing country such as RSA, limited human and
financial resources are the main drivers behind this. Our
sample size was limited by the small ACT team size and
the limitation on caseloads (n = 80 per team member).
Despite the small sample size, we were still able to dem-
onstrate a clear advantage for ACT in terms of time to
first admission and total number of re-admiss ions over
the observation period. These findings are based on a
per-protocol statistical analysis and thus only include pa-
tients who completed the treatment originally alloc ated.
One could argue that the sample size would have been
increased by including a more diverse diagnostic group.
Thedisadvantageofsuchanapproachisthatthelikeli-
hood of unbalanced groups (in terms of diagnostic cat-
egories) will increase significantly and thus require
significantly larger samples and resources. It is thus
Botha et al. BMC Psychiatry 2014, 14:56 Page 9 of 10
http://www.biomedcentral.com/1471-244X/14/56
Page 9
important to interpret the findings in l ight of limited
diagnostic generalizability.
Abbreviations
ACT: Assertive community treatment; DIH: Days in hospital;
CMHT: Community mental health teams; PSR: Psycho-social rehabilitation;
DSM-IV-TR: Diagnostic and statistical manual of mental disorders, Text
revision; HFUs: High frequency users; DACTS: Dartmouth Assertive
Community Treatment Scale; K-M: Kaplan Meier.
Competing interests
The authors declare that they have no competing interests.
Authors contributions
All authors conceived of and designed the study. UB acquired the data. UG,
EJ and DN performed the statistical analysis. UB prepared the first draft of
the manuscript and both LK and DN made significant contributions to the
final draft. All authors read and approved the final manuscript.
Acknowledgements
We thank the Department of Health in the Western Cape and the
management of Stikland Hospital for allowing us to conduct this study.
Author details
1
Department of Psychiatry, Faculty of Medicine and Health Sciences,
University of Stellenbosch, PO Box 19063, Tygerberg 7505, South Africa.
2
Medical Research Council, Bellville, South Africa.
3
Department of Statistical
Sciences, University of Cape Town, Rondebosch, South Africa.
Received: 25 June 2013 Accepted: 18 February 2014
Published: 27 February 2014
References
1. Burns T: The rise and fall of assertive community treatment? Int Rev
Psychiatry 2010, 22(2):130137.
2. Dixon L: Assertive community treatment: twenty-five years of gold.
Psychiatr Serv 2000, 51:759 765.
3. Smith L, Newton R: Systematic review of case management. Aust N Z J
Psychiatry 2007, 41:29.
4. Marshall M, Lockwood A: Assertive community treatment for people
with severe mental disorders. Cochrane Database Syst Rev
2000, (2):CD0 01089.
5. Petersen L, Jeppesen P, Thorup A, Abel MB, Øhlenschlaeger J, Christensen TØ,
Krarup G, Jørgensen P, Nordentoft M: A randomised multicentre trial of
integrated versus standard treatment for patients with a first episode of
psychotic illness. BMJ 2005, 331(7517):602.
6. Karow A, Reimer J, König HH, Heider D, Bock T, Huber C, Schöttle D, Meister K,
Rietschel L, Ohm G, Schulz H, Naber D, Schimmelmann BG, Lambert M:
Cost-effectiveness of 12-month therapeutic assertive community treatment
as part of integrated care versus standard care in patients with schizophrenia
treated with quetiapine immediate release (ACCESS trial). J Clin Psychiatry
2012, 73(3):e402e408.
7. Nishio M, Ito J, Oshima I, Suzuki Y, Horiuchi K, Sono T, Fukaya H, Hisanaga F,
Tsukada K: Preliminary outcome study on assertive community treatment
in Japan. Psychiatry Clin Neurosci 2012, 66(5):383389.
8. Killaspy H, Kingett S, Bebbington P, Blizard R, Johnson S, Nolan F, Pilling S,
King M: Randomised evaluation of assertive community treatment:
3-year outcomes. Br J Psychiatry 2009, 195:8182.
9. Fiander M, Burns T, McHugo GJ, Drake RE: Assertive Community Treatment
across the Atlantic: comparison of model fidelity in the UK and USA. Br J
Psychiatry 2003, 182:248254.
10. Bodén R, Sundström J, Lindström E, Wieselgren I, Lindström L: Five-year
outcome of first-episode psychosis before and after the implementation
of a modified assertive community treatment programme. Soc Psychiat
Epidemiol 2010, 45:665674.
11. Sytema S, Wunderink L, Bloemers W, Roorda L, Wiersma D: Assertive
community treatment in the Netherlands: a randomized controlled trial.
Acta Psychiatr Scand 2007, 116:105112.
12. Burns T: End of the road for treatment-as-usual studies?
Br J Psychiatry
2009, 195:56.
13. McHugo GJ, Drake RE, Teague GB, Xie H: Fidelity to assertive community
treatment and client outcomes in the New Hampshire dual disorders
study. Psychiatr Serv 1999, 50:818824.
14. Hanlon C, Wondimagegn D, Alem A: Lessons learned in developing
community mental health care in Africa. World Psychiatry 2010,
9:185189.
15. Patel V, Araya R, Chatterjee S, Chisholm D, Cohen A, De Silva M, Hosman C,
McGuire H, Rojas G, van Ommeren M: Treatment and prevention of
mental disorders in low-income and middle-income countries. Lancet
2007, 370:9911005.
16. Odenwald M, Lingenfelder B, Peschel W, Haibe FA, Warsame AM, Omer A,
Stöckel J, Maedl A, Elbert T: A pilot study on community-based outpatient
treatment for patients with chronic psychotic disorders in Somalia:
Change in symptoms, functioning and co-morbid khat use. Int J Ment
Health Syst 2012, 6(1):8.
17. Botha UA, Koen L, Joska JA, Hering LM, Oosthuizen PP: Assessing the
efficacy of a modified assertive community-based treatment programme
in a developing country. BMC Psychiatry 2010, 10:73.
18. Ty rer P: The future of specialist community teams in the care of
those with severe mental illness. Epidemiol Psichiatr Soc 2007,
16:225230.
19. Weiden P, Glazer W: Assessment and treatment selection for revolving
door inpatients with schizophrenia. Psychiatr Q 1997, 68:377392.
20. Teague GB, Bond GR, Drake RE: Program fidelity in assertive community
treatment: development and use of a measure. Am J Orthopsychiatry
1998, 68:216232.
21. R Core Team: R: A language and environment for statistical computing.
R Foundation for statistical Computing. Vienna, Austria; 2012. URL http://
www.R-project.org/. ISBN 3-900051-07-0.
doi:10.1186/1471-244X-14-56
Cite this article as: Botha et al.: The rise of assertive community
interventions in South Africa: a randomized control trial assessing the
impact of a modified assertive intervention on readmission rates; a
three year follow-up. BMC Psychiatry 2014 14:56.
Submit your next manuscript to BioMed Central
and take full advantage of:
Convenient online submission
Thorough peer review
No space constraints or color figure charges
Immediate publication on acceptance
Inclusion in PubMed, CAS, Scopus and Google Scholar
Research which is freely available for redistribution
Submit your manuscript at
www.biomedcentral.com/submit
Botha et al. BMC Psychiatry 2014, 14:56 Page 10 of 10
http://www.biomedcentral.com/1471-244X/14/56
Page 10
  • Source
    • "Furthermore, this model does not include systematic links to primary health care, other elements of the broader health and social system, or to families and community stakeholders. While the ACT model is still evolving in the United States (Donahue et al., 2012), it is also now starting to be adapted and tested in South Africa (Botha, Koen, Galal, Jordaan, & Niehaus, 2014). There are several studies outside South Africa currently underway [at the time of this report], evaluating potentially sustainable treatment models for supporting individuals with severe mental illness within the LMIC context. "
    [Show abstract] [Hide abstract] ABSTRACT: This study examined the quality of family relationships and its associations with the severity of unmet needs of individuals admitted to a tertiary psychiatric hospital in South Africa. The quality of family relations and perceived unmet needs were assessed using the Lehman Quality of Life Interview and Camberwell Assessment of Needs, respectively. The results show that higher total unmet needs were associated with lower quality of family relations. The main areas of serious unmet needs included accessing government benefits and information, and establishing social relations. The results have implications for hospital-based social workers beyond managing psychiatric symptoms in South Africa.
    Full-text · Article · Jan 2016 · Social Work in Health Care
  • Source
    [Show abstract] [Hide abstract] ABSTRACT: These international clinical practice guidelines were developed with detailed input from 29 invited international consultants, who provided content as well as detailed feedback on draft versions. The final draft of the guidelines was ratified by the Executive of the International Early Psychosis Association and presented and formally endorsed at the Third International Conference on Early Psychosis held in Copenhagen, September 2002. They have been revised slightly to include medications that were not available in 2002, although a fully comprehensive process of update has not yet been conducted. The final version is published in this Supplement with the aim of encouraging further discussion as well as providing practical guidance to clinicians and researchers. A second edition is planned for publication in 2008.
    Full-text · Article · Aug 2005
  • Source
    [Show abstract] [Hide abstract] ABSTRACT: Background: Frequent users of health care services are a relatively small group of patients who account for a disproportionately large amount of health care utilization. We conducted a meta-analysis of the effectiveness of interventions to improve the coordination of care to reduce health care utilization in this patient group. Methods: We searched MEDLINE, Embase and the Cochrane Library from inception until May 2014 for randomized clinical trials (RCTs) assessing quality improvement strategies for the coordination of care of frequent users of the health care system. Articles were screened, and data abstracted and appraised for quality by 2 reviewers, independently. Random effects meta-analyses were conducted. Results: We identified 36 RCTs and 14 companion reports (total 7494 patients). Significantly fewer patients in the intervention group than in the control group were admitted to hospital (relative risk [RR] 0.81, 95% confidence interval [CI] 0.72-0.91). In subgroup analyses, a similar effect was observed among patients with chronic medical conditions other than mental illness, but not among patients with mental illness. In addition, significantly fewer patients 65 years and older in the intervention group than in the control group visited emergency departments (RR 0.69, 95% CI 0.54-0.89). Interpretation: We found that quality improvement strategies for coordination of care reduced hospital admissions among patients with chronic conditions other than mental illness and reduced emergency department visits among older patients. Our results may help clinicians and policy-makers reduce utilization through the use of strategies that target the system (team changes, case management) and the patient (promotion of self-management).
    Full-text · Article · Sep 2014 · Canadian Medical Association journal
Show more