Cost effectiveness of a collaborative stepped care intervention for anxiety disorders in the primary care setting.

*describe the healthcare problem that underlies this proposal: Nowadays the
insight has come that
common mental disorder should best be approached by a collaborative stepped
care approach with
evidence based components.(63-65) Evidence based Guidelines have been developed
for several
mental disorders, including anxiety disorders.(1-5) Although many Dutch GPs
have followed courses into
the diagnosis and treatment of anxiety disorders, the amount in which these
guidelines are followed is
insufficient(6) in terms of recognition(7) and treatment of mental disorders in
primary care(8) with
consequent overburdening of the health care system.(9,10,59) This problem could
be solved with more
efficient treatment according to collaborative stepped care
principles.(11-13,52-59) The aim of this study
is to implement and evaluate an organization of care in the primary care
setting in terms of collaborative stepped care for anxiety disorders and to
evaluate its cost effectiveness. *what is the disease subject of
this proposal: In this proposal Generalized Anxiety Disorder (GAD) and panic
disorder (PD) are targeted
because they are the most prevalent and most costly mental disorder in terms of
societal impact(60,61)
that can be treated effectively(56,52-53,58)in the primary care
setting,(14,15)if recognition and treatment
is enhanced (1-3,5-7). *what is the group of patients, targeted in this
proposal: The group of patients at
which this study targets is primary care patients diagnosed with GAD and/or PD,
aged older than 18, not
under current psychiatric teatment, and having sufficient knowledge of the
Dutch language to fill in the
questionnaires. To enhance the external validity of the RCT no other exclusion
criteria will be applied. *if
applicable in relation to the target group, indicate how variations in sex, age
or cultural background are
taken into account: Although women suffer from anxiety disorders more often
than men,(14-16) both
sexes will be sufficiently represented in the sample to consider gender as a
variable in the analysis. The
same holds for age and cultural background. *describe the usual care in the
Netherlands for the
(sub)group of patients involved: If the anxiety disorder is recognized, the GP
should follow the guideline
of the Dutch College of General Practitioners (NHG standard)(3) for anxiety
disorders. Many GPs feel
incapable to deliver the cognitive behavioural interven-tions described in
these guidelines.(19) Almost
half of the patients receive antidepressant medication instead or are referred
to psychiatric
care.(6,8,9,19)In summary, Care As Usual often implies suboptimal care,
including referral to medical
specialists to exclude somatic disease. *who are involved in the usual care,
and what is their
participation in this proposal: GPs give the usual care and participate in this
proposal via the LUMC (The
Hague region)and VUMC (Amsterdam region) departments of primary care. In The
Hague the local GPs
collaborate with the regional mental health care institution Parnassia. The
same holds for GPs in the
south/west region of Amsterdam, who collaborate with the regional mental health
care institution GGZ
Buiten Amstel. *describe your motivation for the intervention, subject of this
proposal and motivate the
effectiveness of the intervention. The intervention is a collaborative stepped
care intervention for GAD
and PD in the primary care setting. The algorithm is built up from 3
interventions that have separately
been proven effective and feasible in the primary care setting.(17,19):
manual-guided self help protocol,
three-session cognitive behavioural approach in which the patient scores his
symptoms and evaluates
them with the GP, and an algo-rithm of antidepressant medication for anxiety
disorders prescribed by
the GP.(17) Globally, minor interventions precede more invasive interventions.
All subsequent steps
have been proven effective in randomized clinical trials performed by this
research group.(17,19)
Evidence exists that such a stepped collaborative care model for anxiety
disorder is cost-effective in the
GP setting.(54,55,57)However, this model is not yet used in the Netherlands.
This approach could
improve effectiveness and costs of care for GAD and PD. *if applicable in
relation to the intervention, pay
attention to compliance: In the collaborative care framework, compliance and
adherence enhancing
techniques are essential(see further). This method has been proven effective in
an RCT by the research
group before.(17,20)

Aim of the study
The aim of this study is to evaluate effects and costs of a collaborative
stepped care approach in the
primary care setting for patients with prevalent anxiety disorders PD and GAD
compared with treatment
as usual.
Question of the study is:
Is a 12-24 weeks collaborative stepped care approach applied by the GP in the
primary care setting
more (cost-)effective on improvement of GAD and PD compared with care as usual
in the primary care
setting in the short term (6 months) and long-term (12 months)?

Clinical study
*preliminary studies by applicants on the subject of this proposal: By Van der
Feltz e.a., a multi center
randomized trial has been performed in primary care in which 3 session
cognitive behavioral techniques
and an antidepressant algorithm, both performed by the GP, in a collaborative
stepped care framework,
were effective in patients with PD (panic disorder) and/or GAD (generalised
anxiety disorder) presenting
with Medically Unexplained Medical Symptoms (MUMS). Effect sizes were between
0.80 and 2.60
Cohen*s delta.(17,20) In another RCT, Van Boeijen e.a. established
effectiveness of a manual guided
self help method for patients with PD and/or GAD.(19) In this RCT, the pre-post
effect size of the guided
self-help manual was 0.8 on the main outcome measure the STAI. Van der Feltz
e.a. set up a
nationwide program in which collaborative care techniques such as contracting
and compliance and
adherence improving techniques are implemented in primary care settings.(6) Now
is the time to
establish cost effectiveness of the combined collabo-rative stepped care
strategy for GAD and PD.
*Design: Two armed cluster-randomized pragmatic clinical trial, with
randomization between primary
care practices in order to prevent contamination and thus dilution of the
effect.(26)A flow chart is
included in the appendix.
*sample size calculation (motivate assumptions) and feasibility of recruitment.
In this study, cluster randomisation between GP practices, and a MLA where GPs
are the first hierarchical level and patients
are the second level, is used. The BAI is the primary outcome measure. Power
calculation: The power
calculation is for a MLA analysis. sample size calculation/data analysis: In
studies examining the effect
size of monotherapy with antidepressants for PD, short term effect sizes of
0.50 and long term effect
sizes of 0.70 were found.(75,76) In the study evaluating effectiveness of a two
step protocol, effect sizes
between 0.80 and 2.60 were found depending on the outcome measure.(17) Because
in this study the BAI was not used as main outcome measure, an estimate of the
clinically relevant difference should be
made. In this study, we would like to be able to detect a clinically relevant
difference in terms of a
duplication of the effect of monotherapy, that is an effect of 1.2, as
difference with the CAU on the
continuous measure of the BAI. In unpublished NESDA data, 10 points decrease on
the BAI, which is a
difference of * SD, may be considered a clinically relevant difference.(48) A
variance of 1.15 as
intraclass correlation would be acceptable as presumption if the contrast
between the experimental
conditions would be rather high, that is if the practiced CAU would be
different than the collaborative
care intervention as performed in the different practices. Consequently, the
sample size should be 1.12 x
60,(31) that is 2 x 67 completers. We estimate a drop out rate of 30% and must
include 2 x 87 patients in
the study. (alpha = .05; power 0.90).
*data-analysis and presentation / synthesis: Intention-to-treat analysis will
be performed by MLA with
GPs in the primary hierarchical level. If at this level differences in outcome
will be found, age, gender
and experience, and practice characteristics will be explored as variables
influencing the outcome.
Subsequently, at the secondary hierarchical level, patient characteristics
influencing the outcome will be
explored. To control for possible skewness in the randomised groups as far as
distribution of
confounders is concerned, propensity scores will be calculated. Possible
confounders such as age,
gender, and immigrant status, will be taken as variables in the analysis. In
case of missings, two
methods of analysis will be used. The methods that are available to deal with
the missing data of
patients who withdraw can be distinguished into so-called naïve and principled
methods.(72) Naïve
methods aim to provide an estimate of the mean costs by omitting patients, e.g.
complete case analyses
or LOCF. Another method is to use MLA and to cluster the outcome measures per
measurement: all T1
measurements are taken together, al T2 measurements, etc, and they are compared
to each other as
clustered outcome measures. This is an elegant method to deal with incidental
missings that did not lead
to complete loss to follow up.In this trial, this method will be used to
account for missings in the
effectiveness analysis. An advantage of multilevel analysis of longitudinal
data is the ability to handle
missing data. Because multilevel regression models (in contrast to traditional
analysis of variance of
repeated measures) do not assume equal number of observations or even fixed
time points all cases
remain in the analysis increasing the precision of the estimates and the power
of the statistical tests.
Attrition in longitudinal designs normally is not random. In longitudinal
designs information about the
dropouts from earlier measurements in normally available. It may be assumed
that, conditional on these
variables, the missingness is random. It is not assumed that the missingness is
completely independent
of all other variables. Multilevel modeling of repeated measures with missing
data assumes that the data
are missing at random, provided that Maximum likelihood estimation is used.
This is in contrast to the
complete cases method in traditional analysis of variance of repeated measures,
which assumes that
data are missing completely at random. Principled methods aim to provide an
unbiased estimate of the
variance by taking account of the missing observations. Currently, principled
methods e.g. multiple
imputation are the standard to deal with missing data in economic evaluations.
The principled methods
take into account the special characteristics of cost data that affects their
analysis.(69) Hence, we will
apply a principled method for dealing with missing data in our economic
evaluation.
.time schedule: Preparation: 6 months. Subsequent to the approval of the
Medical Ethical Board PCP
practises are recruited and GPs are trained. Inclusion and intervention phase:
24 months. Analysis and
writing: 6 months. Expected inclusion: every two months, in both regions 5-6
completers are expected to
be included, leading to 134 completers during the 24 month inclusion phase. The
study will last 3 years.
Baseline measurements take place before inclusion (T0), follow up at 3
months,(T1) 6 months (T2), 9
months(T3) and 12 months(T4).

*Intervention(s): Evidence exists for effectiveness of stepped care in PD and
GAD,(68) more specific for
two of the three proposed steps. The effect size of the combined step 2 and
three ranged between 0.80
and 2.60 in a former study of the research group.(17)Evidence exists as well
that the combination of self
help with CGT is effective in treatment of PD and/or GAD in primary care.(74)A
three-step algorithm is
favoured because of good results with three steps in a large algorithm study on
treatment of depressive
disorder (STAR*D trial), in which relatively high remission rates were found
for the first three steps, but
not for later steps: 36.8%, 30.6%, and 13.7% for the respective
steps.(71)Recent publications make a
case for a similar stepped care approach for PD and GAD in the primary care
setting. The systematic
review in the appendix gives an overview. The international consensus group on
Depression and
Anxiety suggests this approach as well.(50) In this study, the most feasible
stepped care intervention
that is in line with these recommendations, and has been proven partially
effective is chosen. The
probability that this intervention, if proven cost effective, would become part
of the GP guidelines and
would be implemented is deemed high. The approach is expected to yield higher
remission rates in
primary care, less referral to mental health care, less costs and more health
gain. The intervention
follows a collaborative stepped care model lasting for at least 12 weeks and
maximum 24 weeks. In
collaborative care, a casemanager (SPV) and a consultant psychiatrist support
the GP. The patient is
actively involved in the treatment plan by contracting by the GP. Motivation
and monitoring is perfor-med
by the casemanager by means of written instructions, regular visits and self
report of the patient with the
BAI. Adherence to the stepped care protocol is supported by feedback on these
by a psychiatrist, as
shown effective in prior research.(17)The stepped care approach comprises the
following steps: a)
manual guided self help by casemanager, a method proven effective in a former
study (19); b) three
cognitive behavioral sessions by the GP. This consists of 3 sessions in which
the GP instructs the
patient to score the disturbing symptoms, such as panic attacks or MUMS as
symptoms of PD and/or
GAD, on a daily basis, and to report this every two weeks to the GP. In 2
sessions, the GP analyses and
discusses this with the patient and gives instructions to cope with the
symptoms. This approach has
been proven effective in a former study(17) with effect sizes 0.80-2.60
depending on the outcome
measure; c) antidepressants according to a step-up algorithm, prescribed by GP.
He chooses one of two
antidepressants, registered for GAD or PD. In two periods of two months, the
antidepressant is
evaluated and if needed, another antidepressant is prescribed. This algorithm
was used in a former
study with good results.(17)Goal of the intervention is improvement of the
patients, as found in the Beck
Anxiety Inventory(BAI) during the monitoring process of the stepped care
algorithm. During the
intervention period, the patients are monitored every 2 months. Clinically
meaningful improvement is
defined as a decrease of 10 points on the BAI, based on unpublished NESDA data.
If patients relapse,
defined as an increase of 10 points on the BAI during the 12-month period, the
next intervention step is
offered. In case of non-response after all steps have been taken, patients are
referred to the regional
mental health care institution.
Treatment of control group. Half of the PCP practices function as control
group, and within these
practices patients receive CAU. The actual content of the CAU treatment (e.g.
medication, number of
contacts in primary care, and referral policy) will be assessed with the Scale
for Medical Utilisation of
Health Services.

Not applicable