Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
A COST OF ILLNESS STUDY OF
GENERALIZED ANXIETY DISORDER IN CANADA
AUTHOR
BASIL GREGORY BEREZA
A THESIS SUBMITTED IN CONFORMITY WITH THE REQUIREMENTS FOR THE DEGREE
OF MASTER OF SCIENCE. GRADUATE DEPARTMENT OF PHARMACEUTICAL SCIENCES,
UNIVERSITY OF TORONTO.
© 2010
ii
A cost of illness study of generalized anxiety disorder in Canada
Basil Gregory Bereza
Master of Science
2010
Department of Pharmaceutical Sciences
University of Toronto.
Abstract
Background: Economic evaluations of generalized anxiety disorder (GAD) have been
limited to ≤18 months. A decision model was developed; quantifying the lifetime cost-of-
illness (COI) of GAD.
Methods: An incidence-based Markov-model was developed using TreeAge® software,
reflecting 9 health-states (HS): physician-assessed patients (3HS), maintenance
therapies(4HS), discontinuation(1HS) and death(1HS). Onset probability (ages 18-80)
determined model entry. Canadian Psychiatric Association (CPA) guidelines determined
pharmaco-therapy, with revisions/validation by an expert panel. Response, remission
based on pooled-analysis of CPA-cited evidence. Remaining clinical rates, absenteeism
and hospitalization retrieved from literature. Direct (clinician, pharmacotherapy,
hospitalization) and indirect costs (wage rate) retrieved from government publications.
Results discounted at 5%.
Results: The mean COI was 2008 Canadian $31,213(SD=$9,100)/patient; 96%
attributed to absenteeism. Mean age=31years, discontinued treatment=85% by 2nd
year, treatment discontinuation duration, 14(SD=9) years.
CONCLUSION: GAD is a costly disease with a lifetime COI<$32k/patient; absenteeism
exerts a significant impact. Limited prospective data contributes to uncertainty of
estimate.
iii
Acknowledgements
This thesis would not have been possible without the generous and continuous support of the faculty members at the departments of Pharmaceutical Sciences and Health Policy, Management and Evaluation at the University of Toronto, as well as of fellow students, colleagues and family members. In particular, I am deeply indebted to the following individuals: Thomas Einarson PhD, for his technical guidance, history lessons, puns, observations and perspective on life, without whom the voyage to the defense would have been that much less interesting. Márcio Machado Dias Ferreira, PharmD PhD, for his academic guidance, discussions on spirituality and philosophy, the creation of new patient preference tools, all things Brazilian and providing traffic directions in Toronto. Most of all I am grateful for the laughter, and for his friendship. Emmanuel Papadimitropoulos PhD, for taking the time and undertaking the task, of providing my first lessons in pharmacoeconomics. I am deeply indebted for his constant selflessness on my behalf in relentlessly creating practical and academic opportunities. Beth Sproule PharmD and Arun Ravindran MD, I am indebted to Drs Sproule and Ravindran and for providing their support and clinical expertise. Adrian R. Levy PhD, for his timely and efficient, review and advice related to the thesis defense (as well as to the employees of Oxford Outcomes for their enthusiastic support). Audrey Laporte PhD, for formally introducing me to the topic of health economics. This thesis would not have been possible without Dr Laporte‘s continued support and conviction that I attain a graduate degree. Christopher Longo PhD, for his keen observations in academia and humanity and for ‘showing me the ropes’ when we were both in industry. I am honored to have worked with him. Roman Bereza MD MSc, and Eugene Bereza MD CM, CCFP, my brothers, who provided clinical and epidemiologic guidance early on in my pursuit of the Master‘s degree. Furthermore, each of them, in different ways, provided the life lessons I would eventually make use of, during this voyage. Elaine Kielo, patience personified, for her editing skills, unwavering moral support, and love, without whom, there would be no thesis. Rawley Kielo, who just listened (I think).
I am deeply honored to have met each of you. For all that you have given me, and all that will transpire from this gift, all I can say is a very humble, “thank you”.
iv
TABLE OF CONTENTS
Abstract ............................................................................................................... ii
List of Figures ................................................................................................... vii
List of Tables .................................................................................................... viii
List of abbreviations and acronyms ................................................................. ix
Section 1: Introduction ....................................................................................... 1
Overview ......................................................................................................................................... 1
Generalized anxiety disorder ....................................................................................................... 1 Epidemiology, Nature and Course ................................................................................................ 3 Management of GAD .................................................................................................................... 4 Relapse ......................................................................................................................................... 8
Economic Evaluations .................................................................................................................. 8
Decision Analytic Modeling .......................................................................................................... 9
Review of Literature .................................................................................................................... 11 Methods ...................................................................................................................................... 12
Acceptability criteria ............................................................................................................... 12 Presentation of literature review ............................................................................................ 13 Quality Assessment ............................................................................................................... 14
Results ........................................................................................................................................ 15 Literature search .................................................................................................................... 15 Description of Included Studies ............................................................................................. 15 Full Economic Evaluations .................................................................................................... 20 Quality Assessment of Full Economic Evaluations ............................................................... 22 Partial Economic Evaluations ................................................................................................ 24
Resource Utilization ........................................................................................................... 24 Impairment ......................................................................................................................... 32 Quality of life ...................................................................................................................... 33
Literature review discussion ....................................................................................................... 41 Limitations .............................................................................................................................. 45
Statement of problem .................................................................................................................. 46
Purpose of the study and objectives ......................................................................................... 46
Research question ...................................................................................................................... 46
Section 2: Methods ........................................................................................... 47
Overview ....................................................................................................................................... 47
v
Pharmacoeconomic model ......................................................................................................... 48
Structure ....................................................................................................................................... 48 Analytic perspective .................................................................................................................... 48 Audience ..................................................................................................................................... 49 Target population ........................................................................................................................ 49 Type of model ............................................................................................................................. 49 Health states overview ............................................................................................................... 50 Time horizon ............................................................................................................................... 50 Attrition rate ................................................................................................................................ 51
Process ......................................................................................................................................... 53 Health state 1: Family physician assessment (initial health state) ............................................. 53 Health state 2: Specialist assessment 2nd line ........................................................................... 55 Health state 3: Specialist assessment 3RD LINE ......................................................................... 64 Health states 4 through 7: Maintenance therapy health states .................................................. 65 Health state 8: Treatment discontinued ...................................................................................... 65 Health state 9: Death (absorbing state) ...................................................................................... 66 Treatment algorithm ................................................................................................................... 66 Dosing ......................................................................................................................................... 69
Clinical Variables ......................................................................................................................... 69 Intolerance .................................................................................................................................. 72 Remission and response ............................................................................................................ 72 Treatment discontinuation .......................................................................................................... 74 Relapse ....................................................................................................................................... 75 Spontaneous remission .............................................................................................................. 76
Resource use and costs ............................................................................................................. 76 Physician costs ........................................................................................................................... 76 Drug costs ................................................................................................................................... 77 Dispensing fees .......................................................................................................................... 77 Hospitalization costs ................................................................................................................... 78 Treatment seeking behaviour ..................................................................................................... 78 Indirect costs ............................................................................................................................... 78 Discounting ................................................................................................................................. 80 Distributions ................................................................................................................................ 80
Outcomes ..................................................................................................................................... 81
Variability and uncertainty .......................................................................................................... 82
Committee members ................................................................................................................... 82
Section 3: Results ............................................................................................. 84
Cost of illness .............................................................................................................................. 84 Reference case ........................................................................................................................... 84 Breakdown cost analysis ............................................................................................................ 85 Epidemiologic parameters .......................................................................................................... 86 Debugging .................................................................................................................................. 86 Sensitivity analysis...................................................................................................................... 88 Sub analysis ............................................................................................................................... 91
vi
Meta-analysis ............................................................................................................................... 92 CPA literature included in meta-analysis .................................................................................... 92 Remission rates .......................................................................................................................... 93 Response rates ........................................................................................................................... 93 Mean changes in HAM-A scores ................................................................................................ 99 Assessment of heterogeneity ..................................................................................................... 99
Section 4: Discussion, Conclusions, Recommendations ........................... 101
Summary of findings ................................................................................................................. 101
Structure ..................................................................................................................................... 101 Analytic framework ................................................................................................................... 101 Patients seeking treatment ....................................................................................................... 102 Baseline HAM-A scores ............................................................................................................ 103 Treatment Algorithm ................................................................................................................. 104
Process ....................................................................................................................................... 104 Clinical variables ....................................................................................................................... 104 Resource utilization and costs .................................................................................................. 107
Limitations .................................................................................................................................. 109 Intolerance ................................................................................................................................ 109 Remission and response .......................................................................................................... 110 Treatment discontinuation ........................................................................................................ 111 Resource use and costs ........................................................................................................... 111 Meta-analysis ............................................................................................................................ 112
Conclusions ............................................................................................................................... 112 Estimated Burden of Illness ...................................................................................................... 113
Recommendations .................................................................................................................... 114
References ...................................................................................................... 115
List of publications and abstracts ................................................................. 124
vii
LIST OF FIGURES FIGURE I: LITERATURE SEARCH AND DISPOSITION OF IDENTIFIED ARTICLES ................ 17 FIGURE II: BUBBLE DIAGRAM SHOWING THE HEALTH STATES AND TRANSITION
PATHWAYS FOR THE MARKOV MODEL ........................................................................... 52 FIGURE III: DISTRIBUTION OF THE AGE (18-80) OF ONSET FOR GAD USED FOR THIS
STUDY ................................................................................................................................... 54 FIGURE IV: DECISION TREE WITHIN FAMILY PHYSICIAN ASSESSMENT HEALTH STATE 56 FIGURE V: DECISION TREE WITHIN SPECIALIST ASSESSMENT 2ND LINE HEALTH STATE57 FIGURE VI: DECISION TREE WITHIN SPECIALIST ASSESSMENT 3RD LINE HEALTH STATE
............................................................................................................................................... 58 FIGURE VII: DECISION TREE WITHIN MAINTENANCE 1ST LINE THERAPY HEALTH STATE 59 FIGURE VIII: DECISION TREE WITHIN MAINTENANCE BENZODIAZEPINE MONO-THERAPY
HEALTH STATE .................................................................................................................... 60 FIGURE IX: DECISION TREE WITHIN MAINTENANCE 1ST LINE &
BENZODIAZEPINETHERAPY HEALTH STATE .................................................................. 61 FIGURE X: DECISION TREE WITHIN MAINTENANCE 3RD LINE THERAPY HEALTH STATE 62 FIGURE XI: DECISION TREE WITHIN TREATMENT STOPPED HEALTH STATE ................... 63 FIGURE XII: DISTRIBUTION OF COI VALUES FROM MONTE-CARLO SIMULATION ............. 85 FIGURE XIII: HEALTH STATE CUMULATIVE TRANSITION PROBABILITIES GRAPH ............ 87 FIGURE XIV: SENSITIVITY ANALYSIS OF COST VARIABLES ................................................. 89 FIGURE XV: SENSITIVITY ANALYSIS OF CLINICAL VARIABLES ............................................ 90 FIGURE XVI: LITERATURE SEARCH RESULTS FOR A META-ANALYSIS OF CLINICAL
OUTCOMES FOR GAD. ....................................................................................................... 94
viii
LIST OF TABLES TABLE 1: PHARMACOTHERAPEUTIC TREATMENT GUIDELINES FOR GAD .......................... 7 TABLE 2: DESCRIPTION OF INCLUDED STUDIES IN FOR REVIEW ....................................... 18 TABLE 3: FULL ECONOMIC EVALUATIONS: PARAMETERS AND OUTCOMES ..................... 21 TABLE 4:FULL ECONOMIC EVALUATIONS: QUALITY ASSESSMENT .................................... 23 TABLE 5: RESOURCE UTILIZATION STUDIES: PARAMETERS, AND PRIMARY OUTCOMES
............................................................................................................................................... 27 TABLE 6: IMPAIRMENT STUDIES: PARAMETERS AND PRIMARY OUTCOMES .................... 34 TABLE 7:QUALITY-OF LIFE-STUDIES: PARAMETERS AND PRIMARY OUTCOMES ............. 38 TABLE 8: INITIAL, MAINTENANCE AND TITRATED DOSES BY TREATMENT LINE FOR THIS
STUDY ................................................................................................................................... 70 TABLE 9: CLINICAL VARIABLES AND THEIR VALUE INPUTS (I.E., RATES) USED TO
POPULATE THE MARKOV MODEL ..................................................................................... 71 TABLE 10: ECONOMIC INPUT VARIABLES AND ASSOCIATED COSTS*................................ 79 TABLE 11:REFERENCE CASE RESULTS* ................................................................................. 84 TABLE 12:EPIDEMIOLOGIC PARAMETERS OF STUDY COHORT .......................................... 88 TABLE 13: SUB-ANALYSIS OF COST OF ILLNESS WITH 10 SESSION OF CBT THERAPY .. 91 TABLE 14: STUDIES INCLUDED AS 1ST LINE EVIDENCE AND RELATED HAM-A SCORES 95 TABLE 15:STUDIES INCLUDED AS SECOND LINE EVIDENCE (BENZODIAZEPINE
MONOTHERAPY) AND RELATED HAM-A SCORES .......................................................... 96 TABLE 16:STUDIES INCLUDED AS THIRD LINE EVIDENCE AND RELATED HAM-A SCORES.
............................................................................................................................................... 97 TABLE 17: META-ANALYTIC RESPONSE AND REMISSION RATES BY TREATMENT LINE . 98 TABLE 18:SUMMARY HAM-A SCORES BY TREATMENT LINE, DRUG CLASS AND DRUG 100 TABLE 19:PAST MONTH’S WORK IMPAIRMENT ASSOCIATED WITH GAD AND MDD
REPORTED BY KESSLER ................................................................................................. 108 TABLE 20: DAILY COST OF AGENTS EXCLUDED FROM 2ND LINE THERAPY ................... 109
ix
List of abbreviations and acronyms AL A German-language health-related quality-of-life measurement APA American Psychiatric Association BGB Basil Gregory Bereza BDQ Brief Disability Questionnaire BPA British Association of Psychopharmacology CBA Cost-benefit analysis CBT Cognitive behaviour therapy CEA Cost-effectiveness analysis CGI-I Clinical global impressions-improvement CGI-S Clinical global impressions- severity CIDI Composite International Diagnostic Interview CIHI Canadian Institutes of Health Information Cl Confidence interval CMA Cost-minimization analysis COI Cost of illness CPA Canadian Psychiatric Association CUA Cost-utility analysis ECG Electrocardiogram ED Emergency department DALY Disability adjusted life year DSM Diagnostic and statistical manual of mental disorders GAD Generalized anxiety disorder
GHS German National Health Interview and Examination Survey, Mental Health Supplement
HAM-A Hamilton rating scale for anxiety ICD International classification of disease and related health problems ICER Incremental cost-effectiveness ratio LSAS Liebowitz Social Anxiety Scale LSRDS Liebowitz Self-Rated Disability Scale MD Major depression MDD Major depressive disorder MDE Major depressive episode MDUSS Midlife Development in the United States Survey MM Márcio Machado MoH Ministry of Health MOHLTC Ministry of Health and Long Term care NCS National Comorbidity Survey NHRDP National Health Research Development Program; Health NICE National Institute for Clinical Excellence NIDA National Institute of Drug Abuse NIMH National Institute of Mental Health ODB Ontario Drug Benefit OPOT Ontario Program for Optimal Therapeutics
x
OR Odds ratio PD Panic disorder Q-DIS-III-R Quick Diagnostic Interview Schedule-III-R Q-LES-Q Quality of Life, Enjoyment, and Satisfaction Questionnaire QOLI Quality of Life Index RCT Randomized controlled trials SD Standard deviation SDS Sheehan Disability Scale SF-12 Short-Form Health Survey – 12 item SF-36 Short-Form Health Survey – 36 item SFD Symptom free day SOC Societal SSRI Selective serotonin reuptake inhibitors SNRI Serotonin norepinephrine reuptake inhibitors TRE Thomas Ray Einarson VAHSRD Veterans Affairs Health Services Research and Development. WFSBP World Federation of Societies of Biological Psychiatry WHO World Health Organisation XL,XR Extended release
1
SECTION 1: INTRODUCTION
OVERVIEW The resources needed to meet the demand for healthcare services, including those
required to manage mental health, are scarce and are becoming more so over time.
Decision makers are increasingly relying on economic evaluations to help make choices
under budget constraints.[1] Generalized anxiety disorder (GAD) is a chronic disease
with waxing and waning of symptoms; characterized by excessive, uncontrolled, and
often irrational and disproportionate concern about everyday issues.[2,3] Patients
diagnosed with GAD experience considerable impairment and disability, hampering
economic productivity and contributing to health care service utilization.[4-6] However,
despite the chronicity and the course of the disorder, economic evaluations related to
GAD have been limited to time horizons of ≤18 months; leaving the societal lifetime cost
of the disorder to be determined.[6] This thesis presents the mean lifetime cost-of-
illness (COI) per GAD patient (direct and indirect costs) in Canada, quantified through
the development of a dynamic decision model.
GENERALIZED ANXIETY DISORDER The symptoms experienced by patients with GAD are out of proportion to the impact of
the issue related to the anxiety or worry.[7] The disorder is also accompanied by
physiologic symptoms such as restlessness, muscle tension, lack of concentration and
disturbed sleep.[2,3]
Diagnoses of GAD are commonly established through either the International
Classification of Disease and Related Health Problems (ICD code F41.1) published by
2
the World Health Organisation [WHO] or the Diagnostic and Statistical Manual of Mental
Disorders (DSM code 300.02) published by the American Psychiatric Association (APA).
The ICD classification is generally the prevalent choice in Europe, while the DSM criteria
are preferred in the United States and in research.[8] Several differences have been
noted between the diagnostic classifications. While the DSM-IV specifies that symptoms
be present for more days than not, over a period of at least 6 months; the ICD-10
classification states that the individual have “primary symptoms of anxiety most days for
at least several weeks at a time and usually for several months.” Furthermore, while, the
ICD-10 criteria require fewer physical symptoms than DSM-IV, some authors suggest
that the ICD-10 classification has more relaxed diagnostic criteria and is therefore more
realistic and meaningful.[9]
Features that differentiate GAD from non-pathological worry include: controllability,
severity and associated somatic symptoms. Individuals with GAD find it difficult to
control their worry, thereby contributing to the impairment of their daily functioning,
whereas people without GAD are able to control their worries with no significant
interference in daily life.[2] Furthermore, pathological worry is more pervasive,
pronounced, lasts longer and occurs without precipitants when compared to non-
pathological worry. Another distinguishing feature is the associated somatic symptoms
with GAD such as excessive fatigue, restlessness, feeling on edge and irritability.
However, this last feature is less likely to be true of children with GAD.[2]
The severity of the disorder (as well as measurement of a response to therapy) may be
assessed with patient- or physician-reported tools such as the Hamilton Rating Scale for
Anxiety (HAM-A), clinical global impressions- severity (CGI-S) or clinical global
3
impressions-improvement score (CGI-I), the Depression Anxiety Stress Scale, the Penn
State Worry Questionnaire or the Generalized Anxiety Disorder Questionnaire-IV.[10,11]
EPIDEMIOLOGY, NATURE AND COURSE The course of GAD is difficult to map due to the lack of prospective epidemiologic
studies;[12] however, retrospective studies indicate that GAD is a chronic disease with
fluctuating symptoms.[7,12] Onset of the disorder may occur as early as childhood or in
patients into their eighties.[13] The distribution for age of onset in patients with GAD is
bi-modal, with the probabilities decreasing from childhood through the thirties, increasing
into the fifties then decreasing thereafter.[13] To further describe the distribution of
onset in patients with GAD, 27 years separate the 25th and 75th percentile age of onset
for GAD, which is greater than in other anxiety disorders and comparable to distributions
observed with mood disorders.[14]
The estimated lifetime prevalence of GAD ranges between 2.4% and 5.7% in the
general population, while the estimated 12-month prevalence is between 1% and
3.8%.[14-16] The disorder is diagnosed more frequently in women than in men (55-60%
of those presenting with GAD are female).[2] In epidemiological studies, the gender
proportion is approximately two-thirds female.[2] The lifetime prevalence estimate is
6.6% in females and 3.6% in males while the 12 month prevalence rate is 4.3% in
females and 2% in males.[14] Furthermore, the lifetime and 12-month prevalence rates
for GAD are higher in older than in younger patients.[14] In the National Comorbidity
Survey (NCS), the lifetime prevalence in the 15-24 year age was lower than in the 45-55
year group (2% versus 6.9%).[19]
4
GAD is frequently co-morbid with depression and other anxiety disorders.[12] The
German National Health Interview and Examination Survey, Mental Health Supplement
(GHS) found that approximately 40% of GAD patients had a co-morbid current major
depression, and 59-60% had 12 month or lifetime major depression.[17] Co-morbidities
in patients with lifetime GAD and other anxiety disorders include: dysthymia (39%), panic
disorder (23%), and social phobia (34%), alcohol (34%) and substance abuse (28%).[18]
Patients with GAD tend to present with somatic symptoms as their primary complaint. In
a study by Wittchen et al., 13% of GAD patients in primary care presented with anxiety,
while the remaining patients presented with pain, insomnia, cardiac and gastrointestinal
symptoms as their primary complaint.[19] GAD was the primary diagnosis in 50% of
patients seeking a cardiac evaluation and 13-25% of patients with irritable bowel
syndrome.[20]
MANAGEMENT OF GAD The Canadian Psychiatric Association (CPA) guidelines [21] for the management of
GAD include both psychological and pharmacologic treatment. The recommended first
line non-pharmacologic treatment for patients with GAD is cognitive behaviour therapy
(CBT). The recommended first line pharmacotherapeutic agents are either selective
serotonin reuptake inhibitors (SSRIs) or serotonin norepinephrine reuptake inhibitors
(SNRIs). The recommended SSRIs for 1st line treatment include paroxetine, sertraline,
or escitalopram. Venlafaxine is the sole SNRI recommended in the more recent 2006
CPA guidelines. Recommendations for second line agents include: benzodiazepines,
bupropion, buspirone, pregabalin, and imipramine. Adjunctive olanzapine or risperidone,
hydroxyzine, mirtazapine, trazodone and citalopram are recommended as third line
5
options or for treatment resistance individuals.[21] It is worthwhile to note that CPA
guidelines state that there is currently no evidence to support the routine combination of
CBT and pharmacotherapy; and therefore make no such recommendation.[21]
In clinical trials observing an underlying population with GAD, response is often defined
as a CGI-I of 1 or 2 or a 50% reduction in HAM-A scores.[21] Remission of symptoms
is usually defined as a HAM-A score of ≤7 (no or minimal anxiety).[21] GAD is a
treatable disorder; using paroxetine as an example, response rates range from 62-68%
and remission rates range from 30-34%.[21] 1
Treatment guidelines set forth by the CPA are similar to guidelines recommended by the
British Association of Psychopharmacology (BAP),[22] the Ontario Program for Optimal
Therapeutics (OPOT)[23], the National Institute for Clinical Excellence;[24] or the World
Federation of Societies of Biological Psychiatry (WFSBP). [25] 2 The CPA, OPOT and
NICE guidelines all recommend CBT as 1st line non-pharmacotherapeutic treatment.
The NICE guidelines also recommend self help bibliography treatment based on CBT
principles. All of the mentioned guidelines recommend the use of SSRIs as 1st line or
long term treatment; however, recommendations for the use of the SNRI, venlafaxine
vary. While the CPA guideline recommends SNRI’s as 1st line treatment, the NICE
guidelines state that venlafaxine should be administered by a specialist; however no
reason is provided in the NICE guideline. While the CPA includes benzodiazepines as
2nd line therapy, NICE guidelines refer to the use of benzodiazepines in primary care
and on a short term (2-4 week) basis. In contrast, BAP guidelines recommend the use
1 Meta analytic probabilities for response and remission in GAD patients; based on CPA evidence were derived for this study and are presented in section 3. 2 Refer to Table 2 on page 17
6
of benzodiazepines as well as venlafaxine and imipramine as long term therapy or for
patients not responsive to SSRIs.
The 2006 CPA guidelines for the management of anxiety disorders were developed
using evidence-based criteria. Recommendations were based on the strength of
evidence attributed to each study and dependant on the quality of the study design.
‘Level 1’ evidence was assigned to either meta-analyses or replicated
randomized controlled trials (RCTs) that included a placebo condition. ‘Level 2’
evidence was assigned to RCTs with placebo or active comparison condition.
Uncontrolled trials with at least 10 subjects were attributed a ‘Level 3’ status while
anecdotal reports or expert opinion were assigned to ‘Level 4’ evidence.[21] According
to the CPA guidelines; recommendations for 1st line treatment were based on ‘Level 1’
or ‘Level 2’ strength of evidence in addition to clinical support for efficacy and safety, 2nd
line treatment recommendations were based on ‘Level 3’ evidence or higher plus clinical
support for efficacy and safety while 3rd line treatment was based on ‘Level 4’ evidence
or higher plus clinical support for efficacy and safety. ‘Level 1’ of ‘Level 2’ evidence for
lack of efficacy was used for agents that were not recommended by the CPA
guideline.[21]
Meta-analyses of outcomes related to the psychological management of GAD patients,
and cited by the CPA guidelines, reported that CBT is more effective that non-specific
psychological treatment methods for GAD and produced comparable magnitude of
benefits to antidepressant drugs.[26-30] Meta-analyses have also been performed
using selected studies on guideline recommended pharmacotherapeutic treatment.[31-
34] Two studies reported the pooled results of either venlafaxine or escitalopram
trials.[31,32] Another systematic review focused on benzodiazepines, and included a
7
TABLE 1: PHARMACOTHERAPEUTIC TREATMENT GUIDELINES FOR GAD
CBT = Cognitive behaviour therapy; CPA = Canadian psychiatric association; BAP = British association of psychopharmacology; OPOT=Ontario program for optimal therapeutics; NICE = National institute for clinical excellence; WFSBP = World federation of societies of biological psychiatry; XL = Extended release; XR = Extended release
OPOT (2000)
CPA (2006)
WFSBP (2002)
NICE (2004)
BAP (2006)
1st Line non-pharmacotherapeutic
CBT CBT Not stated CBT Acute: Higher doses of
SSRIs or venlafaxine
Long term: Escitalopram or
paroxetine Non-response:
Switch to venlafaxine, imipramine or a benzodiazepine
CBT & Pharmaco-therapy combination
1st Line pharmacotherapeutic
Buspirone, fluvoxamine, paroxetine, venlafaxine XR
Paroxetine, escitalopram, sertraline, venlafaxine XR
SNRI’s or SSRI’s (venlafaxine &
paroxetine)
SSRIs (paroxetine)
2nd Line pharmacotherapeutic
Clomipramine, imipramine
Alprazolam, bromazepam, lorazepam, diazepam, buspirone, imipramine, pregablin, bupropion XL.
imipramine Switch to another SSRI
3rd Line pharmacotherapeutic or adjunctive therapy
Alprazolam, bromazepam, clonazepam, diazepam, oxazepam
Mirtazapine, citalopram, trazodone, hydroxyzine, adjunctive olanzapine, risperidone
benzodiazepines venlafaxine (inferred)
8
single SSRI arm. [33] While the Kapczinski meta-analysis included a wider range of
pharmacologic treatments such as paroxetine, sertraline, venlafaxine and imipramine,
the study did not include escitalopram or other pharmacotherapies such as
benzodiazepines, pregabalin or buspirone.[34] Furthermore, no quantitative synthesis of
the comprehensive evidence used to develop the treatment guideline for GAD was
reported by the CPA. The allocation of specific agents to treatment lines was based on
the strength of evidence (as well as clinical support for efficacy and safety); however, the
size of clinical effect for each treatment line based on the evidence is unknown.
RELAPSE Research studies have reported that GAD patients were at high risk to experience
relapse events within the first 2 years of observation.[35,36] Relapse rates ranged from
20-40% within 6 months to 12 months after treatment discontinuation.[35,36,39] In an
8-year follow up study, cumulative relapse rates at year 1 and year 8 were 19% and 43%
for men; and 6% and 36% for women.[37] These studies suggest that patients with GAD
may require long term treatment to obtain full benefits. [38]
ECONOMIC EVALUATIONS Economic evaluations of health care services or products may be categorized as either
‘full’ or ‘partial’ depending on the scope of analysis.[40] A full economic evaluation is
the comparative analysis of both the costs and consequences of two or more
alternatives for a specific disease or condition. Examples of full economic evaluations
include: cost-benefit analysis (CBA), cost-effectiveness analysis (CEA), cost-
minimization analysis (CMA) and cost-utility analysis (CUA).[40] Parameters of an
9
economic evaluation include: costs, consequences and the subsequent comparison of
those costs and consequences through an incremental cost-effectiveness ratio (ICER)
between targeted comparators. Studies limited to the economic component of a full
economic evaluation have been referred to as ‘partial’ economic evaluations.[40]
Examples of such studies include: cost-description studies (cost or burden of illness,
resource utilization), cost-outcome description studies (single service or program), and
cost-comparison.[40] All of these approaches have been used previously in psychiatry.
DECISION ANALYTIC MODELING Decision analytic modeling synthesizes evidence on health consequences and costs,
(usually of alternative and mutually exclusive healthcare strategies) and informs health
care budget decision makers about resource allocations.[41] However, the purpose of
such models is not to make unconditional claims, but rather to serve as an aid to
decision makers and reveal relationships between assumption and outcomes.[42]
Decision analysis is a mathematical model that incorporates a systematic and
quantitative approach to decision making under conditions of uncertainty. The
framework for decision analysis is based on research by von Neumann and
Morgenstern, known as the theory of expected utility.[42,43] The premise of this theory
is that rational decision makers would choose an option that maximizes their expected
utility.[43] However, given the abundance or complexity of information, the ability of
decision makers to process and arrive at a rational decision may be subject to bias.[44]
The structured approach afforded by decision analysis model enables the decision to be
based on a more extensive range of data and a formal synthesis of the information;[45]
thereby supplementing reasoning abilities and reducing bias.[46]
10
Information contributing to decision analysis may be grouped into three categories:
choices or relevant options that are available to patients and the related outcomes,
chances or the probability that any of the outcomes will occur, and values or a
quantitative expression of the different outcomes.[47] In the context of health
economics, the quantitative values are referred to as utilities, which may be expressed in
monetary terms or other indices of value and may be subjective in nature.[42]
A decision analysis may be performed using either a conventional or dynamic (Markov)
modeling technique. In both conventional and dynamic models, a decision is depicted
as a square node, and chance is depicted as a circular node. The line connecting the
two nodes represents the passage of time. A terminal node represents the outcome at
the end of each pathway. Once a decision is taken, the outcome is a matter of chance.
A conventional decision tree describes the path of a cohort from one health state to
another over a fixed period of time. This approach works well for limited recursion
events or a limited time horizon; and is therefore usually employed for short-term health
consequences. However, under conditions where patients experience recurring events
over the span of a lifetime, additional branches within a conventional tree may become
unmanageable. For example, probability values may vary over time and may not be
valued realistically under a conventional model. Conventional decision models may not
easily capture the dynamic quality of long term clinical implications.[48]
The analyst begins the dynamic model by identifying mutually exclusive health states
that are likely to be experienced by the patient.[48] Once these health states are
determined, the analyst creates all possible pathways within each state that maps out
11
patient behaviour that will lead them to the next health state. The time spent by patients
in a health state before transitioning to the next health state is referred to as the ‘cycle
length’. The length of a cycle time horizon is chosen to reflect a time interval that is
clinically meaningful. For example consider the health states ‘well’, ‘sick’ and ‘dead’.
Patients may enter the model through the ‘well’ health state. From there, pathways
within the ‘well’ health state may transition the patients back into the ‘well’ state, move to
the ‘sick’ health state or to the ‘dead’ health state (sudden accident). The ‘dead’ health
state is referred to as an absorbing state; one where patients exit the model. While the
conventional model considers health state transitions over a fixed period of time, the
dynamic approach considers the transition of health states over a series of cycles.[48]
Although several best practice guidelines have been published, there is no consensus
on the quality assessment of a decision analytic model.[41] Overall, guidelines agree
that all assumptions and data used to develop the structure of the model, including time
horizon and cycle length should be clear, transparent and justified. Contradictions
between guidelines include the quality of the data that should be incorporated and how
uncertainty should be assessed.[41] Most guidelines recommend incorporating time
horizons that are long enough to reflect the essential features of the disease.
Furthermore, lack of available data should not be used as a rationale for failing to
incorporate a time horizon that reflects the course of the underlying condition.[41,42]
REVIEW OF LITERATURE A systematic literature review was performed to summarize the results, and assess the
quality of economic evaluations and humanistic studies related to GAD patients. The
objective of the review was to identify research requirements in this area.
12
METHODS
Acceptability criteria Peer-reviewed, published, original, full-text articles reporting full or partial economic
evaluations or humanistic outcomes related to patients diagnosed with GAD were
included for review. Types of full economic evaluations included in the assessment were
CBA, CEA, CMA, and CUA. Studies reporting partial economic evaluations, such as cost
comparison, cost description (cost or burden of illness, resource utilization), or cost-
outcome description (single service of program), as well as studies reporting on
humanistic outcomes such as preference, utility, and willingness to pay, were also
considered.
Studies that included a sample population meeting the criteria for GAD set forth by the
DSM published by the APA and the ICD published by the World Health WHO), were
considered.[2,3] Only studies reporting results attributable to a population diagnosed
with GAD were included. Studies reporting results in patients who had anxiety-related
symptoms but who were not diagnosed with GAD were excluded. Studies where patients
were diagnosed with GAD in addition to associated comorbid conditions were
considered. No restrictions were made on specific demographic attributes (ie, age,
gender, disease duration) of the underlying population diagnosed with GAD.
One reviewer (BGB) performed a search of the following databases: EMBASE (1980 to
2008 week 16), EBM Reviews (1995 to second quarter of 2008), MEDLINE (1950 to
April 2008 week 4), and HealthSTAR (1966 to April 2008). Another reviewer (MM)
verified the appropriateness of the search by reviewing the search terms and ensuring
that each database was searched separately and that the combination of search terms
13
was used properly. The following search terms, mapped to subject headings of each
database, were used: economic evaluation, cost, burden, generalized anxiety disorder,
resource utilization, budget, outcomes, preference, utility, willingness to pay,
effectiveness, and efficacy.
Two reviewers [BGB, MM] independently assessed titles or abstracts of potentially
relevant studies based on the inclusion criteria. Disagreements were adjudicated by a
third reviewer. Articles of the selected titles and abstracts were retrieved for full-text
review.
Retrieved articles were excluded for the following reasons: no economic evaluation
reported, inappropriate patient population, or results not attributable to GAD population.
Furthermore, publication types other than original studies (eg, comments, letters to
editors, abstracts, reviews) were excluded.
The reference lists of retrieved articles were searched manually for relevant studies that
may have not been captured by the established article search (ie, electronic database
search). Studies identified through the manual search were subject to the previously
described inclusion criteria.
Presentation of literature review Data were extracted by 1 reviewer [BGB] onto an electronic data sheet created a priori
and assessed for accuracy by a second reviewer [MM]. The systematic review was
presented through an overall description of articles followed by a summary of study
parameters, results, and quality assessments (where applicable). Study descriptors
14
included: first author, year of publication, economic/humanistic outcomes presented,
diagnostic criteria, country of sampled population, and source of funding. Articles were
categorized as either full or partial economic evaluations.
Following the overall description of the articles, study parameters and results by reported
outcome were also presented for full (ie, cost-effectiveness studies), and partial
economic evaluations (ie, cost or humanistic studies). For studies reporting full economic
evaluations, study parameters such as description of the incremental outcome,
interventions, study perspective, description of GAD population, time horizon, and type
of economic model were recorded, along with results reported and a quality assessment.
Results reported by partial economic evaluations were grouped under the following
categories: resource utilization, impairment, and quality of life. Resource utilization
outcomes were categorized under prevalence, frequency or cost of illness. Sub-
categories included: professional help-seeking, primary care, specialist, hospitalization,
diagnostic assessment, and utilization of medication. Impairment outcomes were
detailed as overall measures of impairment, family impairment, mental health, physical
health, social impairment, and work impairment. Quality-of-life results were not stratified
into smaller categories. Partial economic evaluation study parameters and outcomes
included overall outcome, results, first author and year, study design, description of GAD
population, comorbid condition, and mean age.
Quality Assessment The reporting quality of full economic evaluations were assessed using the checklist
created by the Panel on Cost-Effectiveness in Health and Medicine.[49] The checklist
contains 38 questions, which were answered with a dichotomous (yes–no) choice. The
15
questions were grouped as follows (number of questions): framework (8), data and
methods (14), results (9), and discussion (7). Two reviewers (MM and BB) independently
assessed the quality of the full economic evaluations. The third author (TRE) adjudicated
any disagreements. To our knowledge, no checklist has been created to assess the
reporting or analytic quality of partial economic evaluations; therefore, no such
assessment was performed for those studies
RESULTS
Literature search The literature search yielded 784 abstracts, of which 137 met the inclusion criteria and
were retrieved for full-text review. From the articles retrieved, 106 were excluded for the
following reasons: duplicates between databases (n = 32), inappropriate condition (n =
31), GAD-specific results not extractable (n = 29), and no related economic evaluation
reported (n = 14). The references from the remaining 31 articles were manually
searched for relevant articles not identified through the databases. Five additional
articles were identified and retrieved through the manual search, bringing the total
number of articles reviewed to 36. The search tree is presented in Figure I.
Description of Included Studies A description of included articles is presented in Table 2. Five articles reported full
economic evaluations [50-54] and 31 [13,55–84] presented partial economic evaluations.
Articles included in this review were published between 1990 and 2008. The countries in
which the economic evaluations were carried out included the United States (n = 16),
Canada (n = 6), Australia (n = 3), Germany (n = 3), the United Kingdom (n = 2),
Denmark, Finland, France, and Sweden (1 study each). Two studies were sponsored by
the WHO and included data from Brazil, Chile, China, France, Germany, Greece, India,
16
Italy, Japan, the Netherlands, Turkey, the United Kingdom, and the United States. One
study did not report the country in which the analysis was conducted.[66]
Seven of the 36 studies included did not specify sources of funding. Of the 29 studies
that reported funding sources, 57% were funded by public agencies, 29% by the
pharmaceutical industry, and 14% by nonprofit or private contract research
organizations. Thirty studies used DSM criteria, of which 16 used the DSM, Fourth
Edition (DSM-IV), 11 used the DSM, Third Edition, Revised (DSM-III-R), and 4 used the
DSM, Third Edition (DSM-III). Seven studies used ICD criteria, with 6 using the ICD-10
and 1 study using the ICD-9. None of the studies using ICD criteria specified whether the
clinical or research version was used.
17
Figure I: LITERATURE SEARCH AND DISPOSITION OF IDENTIFIED ARTICLES
EMBASE (110), EBM Reviews (236), EBM Health Technology Assessment (12)Ovid MEDLINE (R)(229), Ovid Healthstar (197)
Total number of titles or abstracts reviewed: 784
Inclusion criteria:
Abstracts selected for full text retrieval: 137
Full text retrieval from manual search: 5
Exclusion Criteria: Inappropriate condition (31), no economic evaluation (14), results not attributable to GAD
(29) duplicates (32): Total (106)
Final number of articles included in review: 36
Number of full economic articles included in review:
5
Number of partial economic and humanistic outcomes articles
included in review: 31
18
TABLE 2: DESCRIPTION OF INCLUDED STUDIES IN FOR REVIEW Evaluation Type Economic or Humanistic
Outcomes Diagnostic Criteria
Location of Study Funding
Full economic evaluation Guest et al50 (2005) Cost effectiveness DSM-IV United Kingdom Wyeth Pharmaceuticals
Heuzenroeder et al51 (2004) Cost effectiveness ICD-10 Australia Australian Department of Health and Ageing
Iskedjian et al52 (2008) Cost effectiveness DSM-IV Canada H Lundbeck A/S Issakidis et al53 (2004) Cost effectiveness ICD-10 Australia Not stated Jørgensen et al54 (2006) Cost effectiveness DSM-IV United Kingdom H Lundbeck A/S Partial economic evaluations Ansseau et al55 (2005) Impairment, resource utilization DSM-IV Belgium, Luxemburg Not stated Bélanger et al56 (2005) Resource utilization DSM-IV Canada Not stated Bland et al57 (1998) Resource utilization DSM-III Canada Alberta Mental Health
Research Fund Boerner et al58 (2003) Quality of life ICD-10 Germany Lichter Pharma AG Cramer et al59 (2005) Quality of life DSM-III-R Norway Not stated Dahl et al60 (2005) Quality of life DSM-IV Australia, Canada,
Denmark, Norway, Sweden Pfizer Inc
Essau et al61 (2000) Resource utilization DSM-IV Germany German Research Foundation
Fournier et al62 (1997) Resource utilization DSM-III Canada NHRDP Grant et al63 (2005) Impairment DSM-IV United States NIDA Henning et al64 (2007) Impairment, Quality of life DSM-IV United States Not stated Katon et al65 (1990) Resource utilization DSM-III-R United States NIMH Kennedy and Schwab66 (1997) Resource utilization DSM-III-R Not stated Alberta Mental Health
Research Fund, NHRDP Kessler et al67 (1999) Impairment DSM-III-R,
ICD-10 United States NIMH; MacArthur
Foundation; Wyeth-Ayerst Le Roux et al13 (2005) Impairment DSM-IV United States NIMH
19
Maier et al68 (2000) Impairment ICD-10 Brazil, Chile, China, France, Germany, Greece, India, Italy, Japan, Netherlands,Turkey, United Kingdom, United States
WHO
Marciniak et al69 (2005) Cost of illness ICD-9 United States Not stated Massion et al70 (1993) Quality of life DSM-III-R United States Upjohn Company McCracken et al71 (1990) Resource utilization DSM-III United States NIDA Mennin et al72 (2000) Impairment DSM-III-R United States NIMH Olfson and Gameroff73 (2007) Resource utilization DSM-IV United States Eli Lilly and Company Ormel et al74 (1994) Impairment DSM-III-R,
ICD-10 Brazil, Chile, China, France, Germany, Greece, India, Italy, Japan, Netherlands, Turkey, United Kingdom, United States
WHO, NIMH
Saarni et al75 (2007) Quality of life DSM-IV Finland Signe and Ane Gyllenberg Foundation
Schonfeld et al76 (1997) Impairment DSM-III-R United States Upjohn Company
Souêtre et al77 (1994) Resource utilization DSM-III-R France Not stated
Stanley et al78 (2003) Quality of life DSM-IV United States NIMH Steffans et al79 (2005) Resource utilization DSM-IV United States NIMH Stein and Heimberg80 (2004) Quality of life DSM-III-R Canada Province of Ontario, NIMH Sullivan et al81 (2006) Resource utilization DSM-III United States VAHSRD, Greenwall &
RW Johnson Foundation Wetherall et al82 (2003) Quality of life DSM-IV United States NIMH Wittchen et al83 (2000) Impairment, quality of life,
resource utilization DSM-IV Germany Wyeth Pharmaceuticals
Wittchen et al84 (1994) Impairment, resource utilization DSM-III-R United States US Alcohol, Drug Abuse, and Mental Health Administration, Grant Foundation
DSM = Diagnostic and Statistical Manual of Mental Disorders; ICD = International Classification of Disease and Related Health; NHRDP = National Health Research Development Program; NIDA = National Institute of Drug Abuse; NIMH = National Institute of Mental Health; WHO = World Health Organization; Q-DIS-III-R = Quick Diagnostic Interview Schedule-III-R, VAHSRD = Veterans Affairs Health Services Research and Development.
20
Full Economic Evaluations A summary of study parameters and outcomes related to the full economic evaluations
included in this review is presented in Table 3. The following interventions were used to
treat GAD in these studies: cognitive-behavioral therapy,[51] diazepam,[50]
escitalopram,[52] paroxetine,[54] and venlafaxine.[50] Furthermore current care (care
provide to those already in the health care system) was compared to evidence base
optimal care for GAD patients in Australia.[53] Psychiatric co-morbidities were not
consistent between studies. Full economic evaluations considered patients with either
‘pure’ GAD (i.e. no other psychiatric diagnosis) or patients with GAD comorbid with other
anxiety or mood disorders.
All 5 studies reported an ICER as an outcome.[50-54] No studies reported CBA, CMA or
CUA analyses. Four studies reported costs from the public-payer perspective; [50–53] 1
study also reported patient out-of-pocket expense,[51] and 2 from a societal
perspective.[52,54] All studies included direct costs such as medication and
professional health care costs. Absenteeism [54] and symptom free days,[52] were used
as a proxy for indirect costs in studies reporting outcomes from a societal perspective. A
population summary design was used in 2 studies,[51,53] while a conventional decision-
tree model was used in the 3 remaining studies.[50,52,54] Time horizons of the full
economic evaluations varied between 6 and 12 months.
One cost effectiveness study concluded that compared to current practice (i.e.
interventions that included either private psychologists, private or public psychiatrists,
21
TABLE 3: FULL ECONOMIC EVALUATIONS: PARAMETERS AND OUTCOMES Study Incremental
Outcome Interventions and Comparator
Perspective Primary Outcome
GAD Sample Population
Time Horizon, Months
Economic Model
Guest et al50 (2005)
Cost per successfully treated patient, cost per averted relapse
Extended-release venlafaxine; diazepam
Public payer £380 per successful treatment; £295 per averted relapse with venlafaxine
Nondepressed patients with GAD
6 Conventional decision tree
Heuzenroeder et al51 (2004)
Cost per DALY
Evidence based, current practice
Public payer and out-of-pocket expenses
Aus $6900 per DALY saved treated when treated by public salaried psychologist
Panic disorder 12 Population summary
Iskedjian et al52 (2008)
Cost per SFD Escitalopram, paroxetine
Public Payer, Societal
Can $2362 per symptom free year (MoH); dominated (SOC)
GAD only 6 Conventional decision tree
Issakidis et al53 (2004)
Cost per YLD Evidence based practice, current care
Public payer Aus $2713 saved per YLD averted with evidence-based treatment
Panic disorder, social phobia, avoidant personality
12 Population summary
Jørgensen et al54 (2006)
Cost per successfully treated patient
Escitalopram, paroxetine
Societal £1408 cost savings per successful treatment with escitalopram
Not specified 9 Conventional decision tree
DALY = disability-adjusted life-year; MoH = Ministry of Health perspective; SFD = symptom-free day; SOC = societal perspective
22
serotonin or noradrenaline reuptake inhibitors), CBT administered to GAD patients by a
public-salaried psychologist generated cost savings of year-2000 US $6900 per
disability-adjusted life-year (DALY).[51] Another study reported that evidence-based
care in treating GAD resulted in greater population health gains at a cost that was similar
to that of current care, generating an estimated cost savings of year-1997–1998 US
$2713 per year lived with disability (YLD) averted.[53]
The remaining studies examined the cost-effectiveness of drug interventions in the
treatment of GAD. Guest et al[50] reported that extended-release venlafaxine was cost-
effective compared with diazepam, based on an analysis that used Clinical Global
Impression improvement (CGI-I) scores as a measure of effectiveness. Two ICERs were
reported in the study; the estimated additional cost for each additional patient
successfully treated with venlafaxine was £380 (year 2000-2001), and the additional
cost for an averted relapse in patients treated with venlafaxine was £295.50 Two studies
reported the cost-effectiveness of escitalopram compared to paroxetine, primarily as a
result of greater productivity gains.[52,54] Escitalopram treatment resulted in total cost
savings of £1408 (2004) per successfully treated patient compared with paroxetine in a
UK study; the savings were largely attributed to the reduction in the number of missed
days of work.[54] In a Canadian study, escitalopram was cost-effective compared with
paroxetine as a result of a greater number of expected symptom-free days for patients
treated with escitalopram.[52]
Quality Assessment of Full Economic Evaluations The number of articles that met the criterion for each checklist item is presented in Table
4. The framework of the 5 studies was well presented, addressing most of the items on
23
TABLE 4:FULL ECONOMIC EVALUATIONS: QUALITY ASSESSMENT Category Item No. of Articles
That Met the Criterion
Framework Background of problem 5 General framing and design of the analysis 5 Target population for intervention 4 Other program descriptors (eg, care setting, mode of delivery) 1 Description of comparator programs 5 Boundaries of the analysis 5 Time horizon 5 Statement of the perspective of the analysis 5 Data & Methods Description of event pathway 3 Identification of outcomes of interest in analysis 5 Description of model used 3 Modeling assumptions 3 Diagram of event pathway (model) 3 Software used 5 Complete description of study parameters 4 Methods for obtaining estimates of effectiveness, costs, and preferences 5 Critique of data quality 1 Statement of year of costs 5 Statement of method used to adjust costs for inflation 0 Statement of type of currency 4 Source and methods for obtaining expert judgment 0 Statement of discount rates 2 Results Results of model validation 0 Reference case results (discounted at 3% and undiscounted): Total costs
and effectiveness, incremental cost and effectiveness, and incremental cost-effectiveness ratios
5
Results of sensitivity analysis 5 Other estimates of uncertainty 0 Graphical representation of cost-effectiveness results 3 Aggregate cost and effectiveness information 4 Disaggregated results as relevant 1 Secondary analysis using 5% discount 0 Other secondary results as relevant 0 Discussion Summary of reference case results 3 Summary of sensitivity of results to assumptions and uncertainties 3 Discussion of analysis assumptions having ethical implications 3 Limitations of study 5 Relevance of study results for specific policy questions or decisions 3 Results of related cost effectiveness analyses 3 Distributive implications of an intervention 3
24
the checklist within this category. Data and methods were also well presented in most of
the full economic evaluations. All articles identified: the outcomes of interest,
the software used and a stated the year in which costs were valued. Three articles
provided a pathway diagram. Overall the articles failed to critique the quality of the data
used in the analysis. Furthermore two articles provided statements of discount rates and
none provided a method for adjusting costs for inflation. (This omission was of little
consequence since none of the studies covered a time horizon >1 year.) None of the
articles presented any validation of the models used. All articles presented their
reference and sensitivity results. Disaggregated results, secondary results or other
estimates of uncertainty were generally not provided. Each of the 5 studies addressed
limitations of their research. Three of 5 articles included a discussion of related cost
effectiveness analyses, and relevance of the study results for specific policy questions or
decisions.
A 100% consensus was reached between the reviewers. No adjudication was required.
The checklist does not provide for an overall quality rating. However, the percentage of
‘yes’ answers attributed to each article was calculated and ranged from 55.3% to 68.4%
of the checklist items.
Partial Economic Evaluations
Resource Utilization
Of the articles reviewed, 14 reported resource utilization outcomes.[55–57,61,62,66,69–
71,73,77,79,81,84] Study parameters and results for studies reporting resource
utilization outcomes are presented in Table 5.
25
Three studies reported the prevalence of treatment-seeking behavior among GAD
patients.[57,61,84] In a Canadian study of 3,956 adult residents, 26% of those meeting
diagnostic criteria for GAD alone sought treatment for the disorder, while 43% of those
meeting the diagnostic criteria for GAD and a comorbid psychiatric condition sought
treatment.[57[ Bland et al.[57] reported an odds ratio of 2:1 in seeking professional help
for GAD patients with a co-morbidity compared to GAD patients with no psychiatric co-
morbidity; however, these results were not statistically significant. Prevalence rates
were higher in a larger US study that assessed the general population ≥15 years of age,
reporting that 48% of responders meeting the diagnostic criteria for GAD alone and 68%
of those meeting the criteria for GAD and a comorbid condition sought professional
help.[84] In a study that focused on anxiety disorders among those aged 14 to 17 years,
of the 4 adolescents with GAD, one sought treatment.[61]
Resource utilization studies show that patients with GAD were high utilizers of primary
care services.[56,62,66,77] Patients with GAD were approximately 4 times more likely
to visit a physician ≥5 times per year, compared with patients who had not been
diagnosed with GAD.[56] The probability that patients would visit a primary care
physician several times a year was 2 to 4 times higher among patients with GAD than
those without GAD.[56] Furthermore, the prevalence of GAD among patients in a
primary care practice was higher than that of any other psychiatric disorder.[62] Patients
with comorbid condition with GAD showed statistically significant higher primary care
utilization than patients with a diagnosis of GAD alone.[84] However, it is important to
26
note that patients may not necessarily present with anxiety, but rather with underlying
somatic conditions.[8,91]
27
TABLE 5: RESOURCE UTILIZATION STUDIES: PARAMETERS, AND PRIMARY OUTCOMES Outcome Author (Year)
Results Study Design Sample Population
Comorbid Conditions
Age, Years
Prevalence of utilization Professional help-seeking Bland et al57 (1999) GAD alone, 25.8%; comorbid
GAD; 42.7%; P = NS Cross-sectional Population Other psychiatric
diagnosis ≥18
Essau et al61 (2000) GAD, 25% Cross-sectional Schools Not specified 14–17 Wittchen et al84 (1994) GAD alone, 48.2%; comorbid
GAD, 67.9%; P=Not specified Cross-sectional Nationally
representative Anxiety, depression, substance abuse
≥15
Primary care utilization (Family physician, internal medicine) Fournier et al62 (1997) GAD, 35.3%; any psychiatric
disorder, 19.5%; P=Not specified
Cross-sectional Nationally representative
Not specified ≥18
Katon et al65 (1990) Current GAD; 21.8%; lifetime GAD, 40.3%; P=Not specified
RCT Distressed high utilizers of medical care
Not specified 18-75
Kennedy and Schwab66 (1997)
GAD,60%: PD, 83%: OCD, 28% P=Not specified
Cross-sectional General population
Not specified Not specified
Souêtre et al77 (1994) GAD alone, 60.6%; comorbid GAD, 59.9%; P = NS
Cross-sectional Ambulatory care setting
Anxiety, depression, substance abuse
18-65
Other specialist utilization Kennedy and Schwab66 (1997) [Gastroenterologist]
23% of GAD patients; P < 0.05
Cross-sectional General population
Not specified Not specified
28
Souêtre et al77 (1994) [Psychiatrists]
GAD alone, 57.2%; comorbid GAD, 61.8%;P = NS
Cross-sectional Ambulatory care setting
Anxiety, depression, substance abuse
18-65
Souêtre et al77 (1994) [Other specialists-not specified]
GAD alone, 14.7%; comorbid GAD, 29.0%; P < 0.001
Cross-sectional Nationally representative
Anxiety, depression, substance abuse
18-65
Wittchen83 (2000) [Specialists-not specified]
No GAD/no MDE, 5.4%; GAD alone, 18.6% (OR, 4.0; 95% CI, 3.2–5.0); GAD/MDE, 25.3% (OR, 6.0; 95% CI, 4.4–8.1)
Cross-sectional Primary care patients
MDE 15-54
Hospital utilization
Souêtre et al77 (1994) [Hospitalization]
GAD alone, 5.1%; comorbid GAD, 11.8%; P < 0.001
Cross-sectional Ambulatory care setting
Anxiety, depression, substance abuse
18-65
Massion et al70 (1993) [Hospitalization]
GAD 35% Panic disorder 26-33% P=.005
Cross sectional Anxiety disorder patients
Anxiety, depression
≥18
Souêtre et al77 (1994) [ED + surgery]
GAD alone, 1.8%; comorbid GAD, 4.0%; P < 0.05
Cross-sectional Ambulatory care setting
Anxiety, depression. substance abuse
18-65
Diagnostic assessment
Souêtre et al77 (1994)[Lab tests]
GAD alone, 25.8%; comorbid GAD; 39.7%; P < 0.001
Cross-sectional Ambulatory care setting
Anxiety, depression, substance abuse
18-65
Souêtre et al77 (1994)[ECG & Cardio]
GAD alone, 7.6%; comorbid GAD, 15.7%; P < 0.001
Cross-sectional Ambulatory care setting
Anxiety, depression, substance abuse
18-65
29
Medication use Ansseau et al55 (2005)[Antidepressants]
GAD alone, 24%; GAD & MD, 42%; P < 0.001
Cross-sectional Primary care patients
GAD with or without MD
≥18
Steffans et al79 (2005)[Benzodiazepines]
GAD & MD, 58.7%; MD only, 36.2%; P < 0.003
Prospective cohort
Depressed patients at research centre
Major depression ≥60
Ansseau et al55 (2005) [Hypnotics]
GAD alone, 23%; GAD & MD: 31%; P = 0.025
Cross-sectional Primary care patients
Major depression ≥18
Ansseau et al55 (2005) [Tranquilizers]
GAD alone, 44%; GAD & MD: 55%; P < 0.0001
Cross-sectional Primary-care patients
Major depression ≥18
Souêtre et al77 (1994) [Anxiolytics]
GAD alone, 74.2%; comorbid GAD, 82.8%
Cross-sectional Ambulatory care setting
Anxiety, depression, substance abuse
18-65
Sullivan et al81 (2006) [Opioids]
GAD patients, 3.4% Cross-sectional Nationally representative
Not specified Not specified
Wittchen et al84 (1994) [Medication not specified]
GAD alone, 24.1% (0.08); comorbid GAD, 46.2% (0.04)
Cross-sectional Nationally representative
Anxiety, depression, substance abuse
≥15
Frequency of utilization Bélanger et al56 (2005) [Number of annual visits*]
Mean (SD): GAD, 5.5 (5.9); no GAD, 3.4 (3.8); P < 0.001
Cross-sectional Primary-care patients
Not specified ≥18
Bélanger et al56 (2005) [≥5 visits]
GAD vs no GAD: OR, 3.95; P < 0.001
Cross-sectional Primary care patients
Not specified ≥18
Kennedy and Schwab66 (1997) [Number of annual visits*]
0 visits: 40%; 1–2 visits: 50%; 3–5 visits: 10%
Cross-sectional General population
Not specified Not reported
Wittchen83 (2000)[Number of annual primary care visits]
No GAD/no MDE, 7.2;GAD alone, 13.8; GAD/MDE, 14.6
Cross-sectional Primary care patients
MDE 15-54
30
Wittchen83 (2000) [Number of annual specialist visits]
No GAD/no MDE, 1.8; GAD alone, 3.4; GAD + MDE, 4.2
Cross-sectional Primary care patients
MDE 15-54
Medication Ansseau et al55 (2005) [Number of drugs per day]
Mean (SD): GAD alone, 2.3 (2.0); GAD & MDD: 2.7 (2.1); P < 0.007
Cross-sectional Primary care patients
With or without MD
≥18
Cost of illness Marciniak et al69 (2005) Marginal effect of GAD on medical costs
Mean (95% CI): US $2138.14 ($1640.96 to $2631.98)
Retrospective, multivariate analysis, incidence based
Patients in private payer database
Not specified Mean (SD): 39.0 (13.6)
Olfson and Gameroff73 (2007)
Median annual health care costs GAD, US $2375; no GAD, US $1448; P = 0.006
Pre–post index visit
Primary care adult patient
Not specified 18-70
Olfson and Gameroff73 (2007) Patients with none to moderate pain interference:
Annual median cost: GAD, US $6932; no GAD, US $3207; P = NS
Pre–post index visit
Primary care adult patient
Not specified 18-70
Olfson and Gameroff73 (2007) Patients with quite a bit to extreme pain interference:
Annual median cost: GAD, US $42,620; no GAD, US $9601; P < 0.05
Pre–post index visit
Primary care adult patient
Not specified 18-70
ECG = electrocardiogram; ED = emergency department; MD = major depression; MDE = major depressive episode; RCT = randomized clinical trial.*Health care service not specified.
31
Patients with GAD were also high utilizers of other specialists' services.[66,77] Patients
with GAD had a higher rate of visits to gastroenterologists than those with obsessive-
compulsive disorder or panic disorder.[66] Comorbidity with other anxiety disorders or
major depressive disorder was associated with significant increases in rates of specialist
utilization compared with GAD alone (16% panic disorder, 3% obsessive compulsive
disorder p<0.05). Patients with comorbid GAD had greater use of internal medicine
services (4.5% vs 1.8% p<0.05), as well as a higher proportion of specialist
consultations, than patients with GAD alone (29.0% vs 14.7%p<0.001).[77]
Furthermore, Wittchen et al.[84] reported that the ratio of primary care to specialty-care
help-seeking was higher among patients with a diagnosis of GAD alone (43.3% vs
29.2%; 1.48) compared with patients with a comorbid GAD diagnosis (50.8% vs 49.6%;
1.02).
GAD patients had higher rates of hospitalization and suicide attempts than did patients
with panic disorder (with or without agoraphobia).[70] Hospitalizations were also
significantly higher in GAD patients with a comorbid condition than patients with GAD
alone (p<0.01).[77]
Diagnostic and laboratory tests were also significantly more commonly used in GAD
patients with comorbidity than GAD alone (p<0.001).[77] This was also true of
biochemistry or hematology tests, X-rays, and computed tomography scanners, as well
as electrocardiogram and other cardiovascular tests.[77]
Drug utilization was usually assessed by comparing patients with no comorbid condition
with those who had a comorbidity.[55,71,77,79,81] Patients with GAD and a comorbid
32
psychiatric condition had higher rates of use of anxiolytics, hypnotics, tranquilizers, and
antidepressants than GAD patients without comorbidities.[55,77,79] Use of
benzodiazepines was significantly greater among patients with both GAD and
depression than among patients with depression alone(p<0.0026).[79] In a study of 14
patients with GAD, participants choose diazepam on 38% of all possible
opportunities.[71]
Two studies reporting cost of illness related to GAD[69,73] were included in this review.
Primary care patients with GAD incurred higher medical costs than patients without
GAD.[69,73] The mean marginal increase in total health care costs incurred by patients
with GAD was US $2138 (1999) compared with other anxiety disorders.[69] Age, sex,
and the severity of the disorder (represented by co-morbidities) were positively
correlated with total medical costs in patients with anxiety disorders.[69] Meanwhile
Olfson and Gameroff reported the median annual health care costs for patients with
GAD was US $2375 compared with US $1448 for patients without GAD.[73] This study
also examined the cost of pain interference in patients with and without GAD. Patients
with GAD who had a high degree of pain interference were predicted to incur a
significantly greater health care costs than patients with high pain interference but no
GAD.73 The median annual predicted cost for GAD patients with quite a bit, to extreme
pain interference was US $42,620 versus US $9601 without GAD. [73]
Impairment A number of studies found GAD was disabling, impairing normal levels of mental and
physical health as well as levels of functioning at work, at home, and in social
settings.[13,55,61,63,64,67,68,72–74,76,77,84] The level of severity of GAD was
33
associated with the level of impairment.[55,63,74] Patients diagnosed with co-morbid
GAD and major depression or major depressive episodes experienced a higher degree
of disability than patients with GAD alone or a mood disorder alone.[55,61,63–
65,77,83,84] Women with GAD had a higher level of impairment at work than did men
(p<0.0001).[55] Patients aged >50 years who reported that the onset of GAD occurred
early in life had a greater degree of impairment in role functioning due to physical
problems than those whose GAD developed later in life (p<0.04).[13] Impairment in
social functioning of patients with GAD alone was equivalent with that of patients
diagnosed with mood disorders alone and greater than in patients with other anxiety or
personality disorders.[63,67] A strong association existed between occupational
dysfunction, productivity, and a diagnosis of GAD.[74,77] Although no significant
difference existed between patients with GAD alone and those with comorbid GAD in
terms of the length of absenteeism, the prevalence of absenteeism was significantly
higher among patients with co-morbidities (33.6% vs 26.6%, p<0.05).[77] Study
parameters and impairment outcomes are presented in Table 6.
Quality of life GAD was associated with lower quality-of-life scores when compared with other anxiety
disorders or chronic somatic disorders (Table 7).[58–60,64,70,75,78,80,82,84] The
impact of GAD on a patient’s quality of life was highlighted in a utility study by Saarni et
al.[75] Utility scores were derived from an individual and population perspective using
the 15D[85] and EuroQoL EQ-5D[86] instruments (mean [SE] values for comorbid GAD:
15D, 0.783 [0.019]; EQ-5D, 0.589 [0.038]).[75] Utility scores derived from GAD patients
were similar to those reported by individuals 20 years older and reporting somatic
conditions such as Parkinson’s disease or heart failure.[75] GAD was associated with
34
TABLE 6: IMPAIRMENT STUDIES: PARAMETERS AND PRIMARY OUTCOMES Outcome Result Study Design Sample
Population Comorbid Conditions
Age, Years
Prevalence of impairment Global impairment Maier et al68 (2000) [Moderate-severe impairment SDS-Global]
GAD alone, 24.9%; comorbid GAD, 46.3% P=Not reported
Cross-sectional Primary care patients
Other psychiatric disorders
Not reported
Social impairment Essau et al61 (2000) CIDI psychosocial impairment-somewhat severe
50% Cross-sectional Schools Anxiety 14–17
Wittchen et al84 (1994) CIDI Social Impairment (Interference with Daily Activities
GAD alone, 28.1% (SE, 0.08); comorbid GAD, 51.2% (SE, 0.04)
Cross sectional Nationally representative
Anxiety and major depression disorders, substance abuse
15-54
Physical impairment Ormel et al74 (1994) [Physical health BDQ moderate to severe]
GAD alone, 53%; comorbid GAD, 59%
Cross-sectional WHO collaborative
Anxiety and depression disorders
15–65
Work impairment Ormel et al74 (1994) Prevalence of moderate to severe impairment on SDS-work scale
GAD alone, 26%; comorbid GAD, 38%; P = NS
Cross-sectional WHO collaborative
Depression and anxiety disorder
15–65
Souêtre et al77 (1994)Absenteeism: 3-Month prevalence
GAD alone, 26.6%; comorbid GAD, 33.6%; P < 0.05
Cross-sectional Nationally representative sample
Anxiety, depression, and substance abuse
18-65
Souêtre et al77 (1994) Absenteeism: 30-Day absenteeism prevalence
GAD alone, 8.2%; comorbid GAD, 11.1%; P = NS
Cross-sectional Nationally representative sample
Anxiety, depression, and substance abuse
18-65
35
Wittchen83 (2000) [% disabled 3+ days]
No GAD/No MDE, 3.1%; GAD/No MDD, 16.9%; MDD/No GAD, 19.4%; GAD + MDD, 33.7%
Cross-sectional Primary care patients
MDE 15-54
Wittchen83 (2000) [10% reduction in productivity ]
GAD/No MDD, 34%; MDD/No GAD, 21%; GAD + MDD, 48%
Cross-sectional Primary care patients
MDE 15-54
Impairment Scores Social impairment Ansseau et al55 (2005) SDS–Family/Home
Mean (SD): GAD alone, 5.8 (2.4); comorbid GAD, 7.3 (2.0); P < 0.001
Cross-sectional Primary care patients
Major depression ≥18
Henning et al64 (2007) SDS–Family/Home
Mean (SD): GAD alone, 3.21 (3.26); comorbid GAD, 4.70 (2.63)
Cohort Treatment-seeking individuals
Anxiety, major depressive disorder
Mean (SD): GAD patients, 33 (12.3); control patients, 30.1 (10.4)
Ansseau et al55 (2005) SDS–social
Mean (SD): GAD, 5.8 (2.3); GAD & MD, 7.3 (1.9)
Cross-sectional Primary care patients
GAD, MD, GAD & MD
≥18
Henning et al64 (2007) SDS–social
Mean (SD): GAD alone, 3.89 (2.38); comorbid GAD, 6.3 (1.91)
Cohort Treatment-seeking individuals
Anxiety and major depressive disorders
Mean (SD): patients, 33 (12.3); control, 30.1 (10.4)
Grant (2005)63 SF-12 social functioning scale*
β (range): –3.7 (–8.7 to –1.3); P < 0.01
Logistic regression
Nationally representative
Anxiety and personality disorders
≥18
Kessler et al67 (1999) Social role impairment GAD beyond MDD
OR (95% CI): MDUSS, 3.5 (2.4–4.6); NCS, 0.8 (0.5–1.5)
Cross-sectional Nationally representative
MDD Not reported
Mennin et al72 (2000 LSAS)Performance fear rating
Mean (SD): social phobia alone, 16.4 (4.9); social phobia and GAD, 19.6 (6.2); P = 0.005
Cross sectional Treatment-seeking anxiety patients
Social phobia with or without GAD
18–65
Mennin et al72 (2000) LSAS Mean (SD): social phobia Cross sectional Treatment- Social phobia 18–65
36
Social interaction fear rating alone, 15.0 (6.2); social phobia and GAD, 18.8 (6.1); P = 0.005
seeking anxiety patients
with or without GAD
Work impairment Ansseau et al55 (2005) SDS–Work
Mean (SD): GAD, 5.8 (2.3); GAD & MD, 7.1 (1.9); P < 0.001
Cross-sectional Primary care patients
Major depressive disorder
≥18
Henning et al64 (2007) SDS–Work
Mean (SD): GAD alone, 4.79 (3.14); comorbid GAD, 5.33 (2.56)
Cohort Treatment-seeking individuals
Anxiety and major depressive disorders
Mean (SD): patients, 33 (12.3); control group, 30.1 (10.4)
Physical impairment Grant (2005) 63 SF-12 Short Health Survey Role Functioning Scale*
β, (range): –1.8 (–7.0 to –3.3); P = NS
Logistic regression
Nationally representative
Anxiety and personality
≥18
Le Roux et al13 (2005) SF-36 physical role score
Mean (SD): early onset, 48.6 (40.2); late onset, 28.4 (37.0)
Cross-sectional Hospital programs, senior centers, and general population
Anxiety, depression, dysthymia
≥55
Olfson and Gameroff73 (2007) SF-12 Short Health Survey Physical Component
Mean (SD): GAD, 34.7 (10.4); no GAD, 40.4 (11.3); P < 0.001
Pre–postindex visit
Adult patients seeking primary care
Not specified 18-70
Schonfeld et al76 (1997) SF-36 physical functioning score†
Mean:Baseline: 91.9; GAD effect –6.5); P < 0.05
Regression analysis
Primary care patients
Anxiety and depression
21–64
Schonfeld et al76 (1997) SF-36 vitality score†
Mean: Baseline 59.9 GAD effect (–7.9); P < 0.05
Regression analysis
Primary care patients
Anxiety and depression disorders
21–64
Productivity Kessler et al67 (1999) Work impairment beyond MDD
OR (95% CI): MDUSS, 3.5 (1.7–7.2); NCS, 1.5 (0.8–2.7)
Cross-sectional Nationally representative
Depression Not reported
37
Ormel et al74 (1994) Disability days in past month
Mean: GAD alone, 4.4; comorbid GAD, 6.3
Cross-sectional WHO collaborative
Depression and anxiety disorder
15–65
Souêtre et al77 (1994) Work days lost per year
Mean (SD): GAD alone, 45.0 (70.5); comorbid GAD, 56.8 (84.1); P = NS
Cross-sectional Nationally representative
Anxiety, depression, and substance abuse
18-65
ED = Emergency department; LSRDS = Liebowitz Self-Rated Disability Scale (11-item tool with a 0–4 rating scale, where 4 = problem limits function severely); SDS= Sheehan Disability Scale; MDE = major depressive episode; OR = odds ratio; SF-12 = 12-Item Short-Form Health Survey; BDQ = Brief Disability Questionnaire (8-item scale adapted from Medical Outcomes Study Short-Form general health survey); WHO = World Health Organization; SF-36 = 36-Item Short-Form Health Survey; CIDI = Composite International Diagnostic Interview ; LSAS = Liebowitz Social Anxiety Scale (24-item scale); MDUSS = Midlife Development in the United States Survey; NCS = National Comorbidity Survey;. *Reference group was those with GAD alone; regression estimates were for comorbidity with MDD. †Reference group was women with no disorder.
38
TABLE 7:QUALITY-OF LIFE-STUDIES: PARAMETERS AND PRIMARY OUTCOMES Author (year) Instrument
Results Study Design
Sample Population
Comorbid Conditions
Age, Years
Overall quality of life Boerner et al58 (2003) [AL questionnaire]
Mean (SD): baseline, 140.1 (18.9); 8 weeks, 168.1 (24.7); P = NS
RCT GAD outpatients None 20–71
Boerner et al58 (2003) Self-rated well-being
Mean (SD): baseline, 34.79 (9.64); 8 weeks, 19.86 (11.97)
RCT GAD outpatients None 20–71
Cramer et al59 (2005) Global quality of life
Mean (SD): current GAD, –0.99 (1.03); lifetime GAD, –0.46 (1.08); P < 0.05
Factor analysis
Nationally representative
Not specified 18–65
Dahl et al60 (2005) Q-LES-Q
Mean (SD): baseline, 62.6 (62.6) RCT GAD outpatients None ≥18
Henning et al64 (2007) QOLI
Mean (SD): end point, 9.9 (4.3); nonanxious controls, 2.5 (1.3); GAD alone, 0.8 (1.9); P < 0.05; comorbid GAD, –0.4 (1.4); P < 0.05
Cohort Treatment-seeking individuals
Anxiety Mean (SD): patients, 33 (12.3); control, 30.1 (10.4)
Saarni et al75 (2007) EuroQoL-5D
Mean (SE): GAD alone, 0.654 (0.046); comorbid GAD, 0.589 (0.038)
Cross-sectional
General population
MDD and psychiatric conditions
Mean: GAD alone, 50.9; comorbid GAD, 52.9
Saarni et al75 (2007)15D
Mean (SE): GAD alone, 0.864 (0.018); comorbid GAD, 0.783 (0.019)
Cross-sectional
General population
MDD and psychiatric conditions
Mean: GAD alone, 50.9; comorbid GAD, 52.9
Stanley et al78 (2003) QOLI
Mean (SD): CBT responders, 2.4 (1.7); CBT nonresponders, 0.8 (1.7); P <0.01
RCT General population
None ≥60
Stein et al80 (2004) Overall well-being
Low self-perception compared with no GAD: OR, 4.85
Cross-sectional
General population
MDD Not specified
39
Wetherall et al82 (2002) SF-36–General Health
Mean (SD): GAD alone, 56.6 (18.2); comorbid GAD: 48.1 (20.2); healthy controls: 77.7 (17.8); P < 0.001
Survey Treatment-seeking older adults
Mean (SD): GAD alone, 68.4 (8.2); comorbid GAD, 65.8 (8.1); controls, 67.8 (7.1)
Satisfaction Stein et al80 (2004) [Life satisfaction
questionnaire]
Difference in degree of dissatisfaction between GAD and no GAD: OR, 5.15
Cross-sectional
General population
MDD ≥15
Sleep Boerner et al58
(2003) [Self-rated sleep questionnaire]
Mean (SD): baseline, 2.65 (0.68); 8 weeks, 3.60 (0.84)
RCT GAD outpatients None 20–71
Suicide Massion et al70 (1993) Suicide attempts
GAD, 13%; PD, 6%; P < 0.005 Cross-sectional
Patients with anxiety disorder
Anxiety, depression
≥18
Wittchen83 (2000) Suicidal ideation
No GAD/MDE, 6.3%; GAD alone (OR [range]), 25.4% (4.8 [4.0–5.7]); GAD/MDE (OR [range]), 64.0% (26.3 [20.5–333.8])
Cross-sectional
Primary care patients
MDE 15-54
AL = a German-language health-related quality-of-life measurement; ED=Emergency department; Q-LES-Q = Quality of Life, Enjoyment, and Satisfaction Questionnaire; QOLI = Quality of Life Index; CBT = cognitive-behavioral therapy; RCT = randomized controlled trial; MDD = major depressive disorder; SF-36 = 36-Item Short-Form Health Survey; PD = panic disorder
40
lower utility scores compared with other anxiety disorders, major depression disorder,
and alcohol dependence, but did not have a lower utility score than dysthymia.
Compared with non-anxious individuals who sought treatment, patients with GAD also
reported overall lower self-perception, well-being, and satisfaction with their quality of life
across most domains (health, standard of living, friendships, relationships with family,
community), with the exception of satisfaction regarding children.[64,70,80] Individuals
with GAD and ≥1 comorbid disorder reported a reduction in their quality of life compared
with individuals with no comorbid disorder(p=0.06).[64] Lifetime GAD was associated
with a lower impact on quality of life when compared with a diagnosis of GAD in the last
year(p<0.05).[59]
Suicidal ideation occurred >4 times more often among patients with GAD alone and 26
times more often among GAD patients with comorbid major depressive disorder than
among those with no anxiety disorders.[83] One study reported prevalence of suicide
attempts was higher in a sample population with GAD, than with panic disorder (with or
without agoraphobia) (13% vs. 6%; P < 0.005).[70]
Baseline and end point quality-of-life scores were collected in 3 studies assessing
treatment for GAD. In a sample of GAD patients aged ≥60 years, CBT was associated
with significant improvement in quality-of-life scores; however, the gain was not
maintained 6 months after treatment.[78] No significant differences in quality-of-life
scores were observed between kava-kava extract, opipramol, or buspirone.[58]
However, in a 12-week study, 51% of patients treated with sertraline achieved a quality-
of-life score within the normal range, compared with 31% on placebo(p<0.01).[60]
41
LITERATURE REVIEW DISCUSSION DSM-III-R or DSM-IV criteria were used to diagnose GAD in 22 of the 37 studies
reviewed. Diagnoses of GAD in the remaining studies were determined using DSM III,
ICD-9, or ICD-10. Differences exist between the DSM and ICD classification systems, as
well as between the versions of each system. DSM-IV emphasizes "excessive and
uncontrollable" worry over a specified time horizon of 6 months, whereas the clinical
version of the ICD-10 classification requires the presence of more anxiety symptoms and
a "persistent state of 'free-floating anxiety' over several months".[87] The duration
requirement for GAD was increased from 1 month under DSM-III to 6 months under
DSM III-TR, DSM-IV, and DSM-IV-TR. The ICD-10 criteria have a version for research
purposes and another for clinical practice. The research version specifies a duration
requirement of 6 months, whereas the clinical practice version refers to symptoms with
duration "over several months." To our knowledge, the impact on differences in episode
duration on economic or quality of life outcomes has not been addressed. However, a
cross-sectional study using a 2 stage sampling system, assessed the impact of episode
duration on impairment using the Brief Disability Questionnaire (BDQ) and the Sheen
Disability Scale (SDS).[68] In that study, patients were grouped into 3 categories;
patients with anxiety symptoms for <1 month and those with symptoms for 1 to 6 months
were described as having generalized anxiety syndrome (GAS), whereas patients
experiencing episodes for ≥6 months met DSM-IV criteria for GAD. No significant
differences of moderate to severe social impairment were observed between groups.[68]
Since no studies that assess the impact of episode duration on other economic
outcomes or quality of life were found, comparisons of economic outcomes using
different classifications systems and versions should be avoided.
42
The cost-effectiveness studies included in this review suggest that evidence-based
practice and SNRI’s were cost-effective for managing patients GAD, regardless of the
presence of co-morbidities. However, there are significant limitations to these studies.
None of the studies reflect the chronicity of GAD or the waxing and waning of symptoms.
Cost-effectiveness studies were either based on summary population data or
conventional decision-tree models of ≤12 months in length. Understandably, the length
of the time horizons in the pharmacoeconomic studies paralleled those found in the
related clinical trials, from which data were used to populate the decision-tree models.
Unfortunately, this method does not reflect the course of illness of the disorder, thereby
calling into question the usefulness of the cost-effectiveness outcomes. Furthermore,
CUAs were not found and, although under continuous debate, cost per quality-adjusted
life-year (and associated country-specific incremental cost-effectiveness thresholds) has
become the standard unit for assisting decision-makers in many health care
systems.[88] This is particularly unfortunate since Saarni et al. presented utility findings
for GAD in 1997.
Two studies reported cost-of-illness outcomes; 1 estimated the medical costs of anxiety
disorders as well as depression,[69] and the other measured health care charges of pain
interference in GAD patients GAD.[73] Unfortunately, ICD-9 was used as the diagnostic
criteria of 1 of the studies, limiting the generalizability of that study to the current
diagnostic criteria.[69] Furthermore, both cost-of-illness studies assessed billing data
claims used as a proxy for health care costs; a measurement subject to the limitations of
retrieving information retrospectively from a chart review.[69,73] A more accurate
method of quantifying health care costs would be to prospectively count the use of
43
health-service components such as physician or specialist visits, diagnostic and
laboratory tests, hospitalization, and use of medication. Furthermore, productivity or
absenteeism was not incorporated in estimating the cost of GAD in either study.
Several studies reported that GAD was under-diagnosed or under-recognized.[91,92]
However, the literature search for the present review did not find any studies that
reported the costs associated with under-diagnosing or not treating patients with GAD.
Saarni et al.[75] derived utility values in patients with GAD alone, as well as GAD
comorbid with depression, reporting larger reductions in health-related quality of life at
both the individual and population level with these disorders than with episodic anxiety
disorders or somatic conditions such as Parkinson’s disease or heart failure. The
derivation of utility scores for GAD alone and comorbid GAD makes a direct comparison
possible between patients with GAD and other conditions using quality-adjusted life-
years. Thus, cost-utility studies of interventions for GAD may be carried out and
compared to cost-utility studies of interventions for other conditions, thereby enabling
decision makers to make rational health care policy decisions.
Given the number of countries in which economic or humanistic outcomes were
measured, caution must be exercised when comparing outcomes between studies
included in the present review. For example, cultural differences and risk of stigma may
influence the rate of absenteeism related to GAD. Measuring costs from a public-payer
perspective may differ between jurisdictions because of variations in public policy
regarding public health access and health insurance coverage. Furthermore, drug
utilization varies between jurisdictions,[93] which may affect clinical outcomes. The
44
purpose of the present review was not to provide a quantitative transnational comparison
of economic evaluations related to GAD; rather, it sought to report the outcomes that
have measured the burden of GAD and provide insight about which questions remain
unanswered.
The reporting quality of the cost-effectiveness studies was assessed using a checklist
created by the Panel on Cost Effectiveness in Health and Medicine.[49] Although the
assessed studies were published between 2004 and 2008, the panel’s
recommendations were published in 1996. None of the studies fully adhered to the
panel’s guidelines. Although they consistently established the framework of their
research, none provided validation of their models and only 3 of the 5 provided a full
description of the event pathways. A quality assessment using the guidelines reported
by Ruger et al.[94] assessed economic evaluations related to smoking cessation. The
economic studies assessed in that reviewe predated the panel’s guidelines. Because
these quality assessments are subjective and no overall quality score is reported;
comparisons between such studies should be undertaken with caution. However, no
appreciable changes in the proportion of articles meeting the criteria were noted
between the Ruger et al. study and the quality assessment present here.
Two systematic reviews recently reported the economic and humanistic outcomes
related to anxiety disorders.[89,90] The present review represents a quality assessment
of cost-effectiveness studies and highlights the discrepancy between the etiology of
GAD and the model parameters used to support evidence of cost-effectiveness, which
may affect the utility of such studies for policy makers and formulary decision makers.
Furthermore, the present review included studies reporting on utility values for GAD, an
45
integral component in cost-utility studies and a useful measure for decision makers,
which allows comparisons across various conditions.
The extent of interest in cost-effectiveness, partial economic evaluations, and humanistic
outcomes related to GAD that are expressed by health care budget decision makers
may depend on the extent of coverage offered. Cost-effectiveness research from the
private-payer’s perspective, focusing on parameters such as patient-specific coverage,
copayments, or contributions, was not identified through the search methods used in the
systematic review. Furthermore, while full and partial economic evaluations from the
public-payer and societal perspectives have been conducted, the time horizons
incorporated in the analysis do not reflect the full clinical course of GAD. Therefore, a
need exists for a an economic evaluation that reflects the course of GAD.
Limitations It is possible that the total population of articles reporting economic and quality of life
outcomes were not included in this review because of errors in database search
engines, key words used, or errors made in mapping search terms. Databases based on
languages other than English, such as the Scientific Electronic Library Online, were not
searched. Thus, articles from regions other than North America and Europe and
published in languages other than English were not captured by our search strategy.
Nonetheless, given the care taken in proper use of search techniques, a system of
verification, as well as exhaustive use of key words, and no language limitation for the
searched electronic databases the included articles appear to be representative sample
of the entire population of articles.
46
STATEMENT OF PROBLEM To date, economic evaluations related to GAD do not reflect the nature or course of
GAD; thereby calling into question the usefulness of the outcomes to health care
decision makers. The cost-effectiveness studies identified through this systematic
review were designed using summary population data or conventional decision-tree
models and were ≤12 months in length. The cost of illness studies were based on health
care charges over periods of ≤18 months. Furthermore, one cost-of-illness study
assessed GAD patients using ICD-9 criteria.[39] Neither cost of illness studies
incorporated measured of productivity.
PURPOSE OF THE STUDY AND OBJECTIVES The purpose of this is study is to develop a dynamic decision model, reflecting the nature
of the disorder and to quantify the lifetime cost-of-illness (COI) for GAD per patient. A
secondary objective is to pool the results of clinical outcomes presented in the evidence
used to develop treatment guidelines for the CPA. The pooling of data will increase the
power of the remission and response rates of therapeutic options used in the decision
model; thereby helping to narrow the range of the uncertainty around the mean COI.
RESEARCH QUESTION From a societal perspective, what is the incidence based lifetime cost of generalized
anxiety disorder (GAD), of patients in Canada diagnosed with this disorder, and who
seek treatment?
47
SECTION 2: METHODS
OVERVIEW This study used an incidence-based, dynamic decision analytic framework, to model the
relationship between the assumptions related to the course of illness and the
management of GAD, as well as the cost consequences of these assumptions. Analysis
was conducted from a societal perspective. Results from this study would be of interest
to decision makers from both the public and private sectors
The model included 9 health states. Patients transitioned to health states according to
the relevant conditions. Patients were assessed in 3 health states by either a physician
in family practice or a psychiatrist, to determine the appropriate therapeutic option. Four
other health states modeled patients on maintenance therapy. One health state modeled
patients who had discontinued treatment. Patients were removed from the model in the
absorbing state. With the exception of the absorbing state, patients remained in a health
state for 6 months at a time before either, transitioning into another state or re-entering
the same state. Patients, ages 18 through 80 entered the model with a primary diagnosis
of GAD as per the DSM-IV or ICD-10 criteria, regardless of severity. The age at which
patients entered the model was based on the distribution of patients’ ages at disease
onset.
Treatment algorithms for the management of GAD were based on CPA guideline
recommendations as well as expert input from clinicians participating in the present
thesis committee.3 The clinical rates used in this model were: intolerance to medication,
3Please refer to the Thesis Committee section for a list of committee members.
48
response, remission, ‘spontaneous remission’, discontinuation of treatment, and relapse.
A meta-analysis of the CPA based evidence was conducted to derive remission and
response rates for this study. The remaining clinical rates were derived from peer-
reviewed literature.
Direct costs included: family practice and specialist physician fees, drugs and drug
dispensing fees, as well as hospital stay. Direct costs were based on Ontario’s Ministry
of Health and Long Term Care published fee schedules for 2008. Indirect cost was
estimated by measuring foregone wages due to absenteeism. The rate of absenteeism
was taken from published literature, while the average industrial wage rate published by
Statistics Canada was used as a proxy for wages for 2008.
The primary outcome in this study was the mean and standard deviation (SD) of the
lifetime COI per patient diagnosed with GAD. The model also calculated the mean, the
mean age of GAD onset, frequency of relapse per patient, proportion of treatment-
resistant patients, and duration of treatment discontinuation.
PHARMACOECONOMIC MODEL
STRUCTURE
ANALYTIC PERSPECTIVE
The analysis was done from a societal perspective. Both direct and indirect costs were
considered in the analyses. Direct costs included: family practice and specialist
physician fees, drugs and drug dispensing fees, as well as hospital stay. Indirect cost
was estimated by measuring foregone wages due to absenteeism.
49
AUDIENCE
Results from this study would be of interest to decision makers from both the public and
private sectors. Direct costs associated with GAD would inform Ministry of Health
decision makers. Costs from a societal perspective would inform both government health
policy, as well as private insurance decision makers.
Furthermore, although the model developed for this study presents results for the cost of
illness associated with GAD, it may be adapted to deal with research questions
regarding the cost-effectiveness of future treatment options.
TARGET POPULATION
The target population consisted of adults, ages 18 through 80 with a primary diagnosis
of GAD as per the DSM-IV or ICD-10 criteria, regardless of severity. Patients with
comorbid psychiatric disorders such as depression, or somatic conditions such as
irritable bowel syndrome were not included in the analysis. Both males and females
were considered in the analysis.
TYPE OF MODEL
This study used a decision analytic framework to model the relationship between the
assumptions related to the course of illness, the management of GAD, and the cost
consequences of these assumptions. TreeAge® Pro Suite 2009 (TreeAge Software Inc.,
Williamstown, MA) software was used to develop an incidence-based, probabilistic
Markov model.
50
HEALTH STATES OVERVIEW
The model included 9 health states (HS):
• Health state 1: Family physician assessment (initial health state)
• Health state 2: specialist assessment for 2nd line therapy
• Health state 3: specialist assessment for 3rd line therapy
• Health state 4: Maintenance therapy for 1st line
• Health state 5: Maintenance therapy for 2nd line (1st line plus benzodiazepine)
• Health state 6: Maintenance therapy for 2nd line (Benzodiazepine monotherapy)
• Health state 7: Maintenance therapy for 3rd line
• Health state 8: treatment discontinued
• Health state 9: death (absorbing state).
TIME HORIZON
The time horizon used in this model was patients’ life expectancy from onset of illness.
The lower age limit of 18 was chosen as this is the age of majority in Ontario. This was
also the lower limit for the age at which patients were usually recruited into the CPA
cited evidence for management of GAD.[21] Canadian life expectancy published by
Statistics Canada set the upper age boundary for the model.[95] The overall life
expectancy at birth for both males and females was 80.4 years in 2005, and was
rounded down to 80 for the model.[95]
The cycle length within a model, describes the length of time patients remain in a given
health state before transitioning to another health state, or back into the same health
state. The cycle length for this study was based on the treatment algorithm
51
recommended by the CPA. As per these guidelines, adequate duration and dosage of
agents were required to observe clinically important effects of therapeutic agents.[21]
Given an inadequate response to initial treatment, guidelines recommend that dosing be
optimized before switching or consideration of augmentation.[21] Guidelines suggest
that a period of 8-12 weeks is necessary to adequately assess response to an agent.
Switching to another 1st line treatment is recommended prior to considering 2nd line
treatment options.[21] Therefore, the selection of a practical cycle length was based on
an accommodation for patients to respond to two treatment choices within the same
treatment line. Furthermore, while effects may be observed in as little as 1 week for
some anti-depressants, significant improvements may not be seen for 6-12 weeks and
may continue to accrue for 6-12months. Therefore, a cycle length of 6 months was
chosen.
Within each health state, a decision tree was embedded to describe the possible
pathways that patients may have taken while in that health state. The decision tree
within each health state, with the exception of ‘death’, began with the application of an
attrition rate that was equivalent to the probability of death from all causes.
ATTRITION RATE
The attrition rate accounted for at the beginning of each cycle for all health states except
‘DEATH’ was based on ‘all cause mortality’ rates derived from Statistics Canada
complete life tables for 2000-2002.[96] Patients were removed from the model based
on the age-specific attrition rate. Life tables reported data in rates per 100,000 and for a
population from birth to 109 years of age, segregated by gender. In order to derive the
52
FIGURE II: BUBBLE DIAGRAM SHOWING THE HEALTH STATES AND TRANSITION PATHWAYS FOR THE MARKOV MODEL
53
‘all cause mortality’ rate for both sexes combined, all cause mortality rates were
weighted by gender.[97] The weighted data were then added for each year to determine
the age specific, all cause gender-adjusted, mortality rate. A ‘bubble’ diagram depicting
the health states and possible transitions among the health states is presented in Figure
II.
PROCESS
HEALTH STATE 1: FAMILY PHYSICIAN ASSESSMENT (INITIAL HEALTH
STATE)
Patients entered the model through the initial health state, ‘FAMILY PHYSICIAN
ASSESSMENT’. Patients may also have entered this health state after discontinuing
treatment for at least 1 cycle (i.e. 6 months). This health state described the possible
pathways patients may have taken once they were diagnosed with GAD and whose
treatment was managed by a physician in family practice.
The age of disease onset determined the age at which patients entered the model. The
distribution of patients’ ages at disease onset was based on peer reviewed literature.[13]
Leroux et al. reported details of that distribution which was derived from a sample of 67
patients with a mean age of 66 years. That study reported the percentage of patients
who experienced onset of GAD from the second through the eighth decade of life, as
well as during ‘childhood’ and ‘teen’ years.[13] A single rate was reported by Leroux for
each age category. However, in this study, each patient’s age was input according to the
actual number of years and not grouped into age categories. Therefore, in order to use
the Leroux data in this study, the probability of the onset of GAD symptoms for each year
of age was needed. A uniform distribution of percentages was assumed for each age
54
FIGURE III: DISTRIBUTION OF THE AGE (18-80) OF ONSET FOR GAD USED FOR THIS STUDY
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
0.02
18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78
Age of onset
Perc
enta
ge
55
category reported by Leroux. The percentage for each age group was then divided by
10 for rates associated with the second through eighth decades of life. The value for the
teen category was divided by 7 (i.e. ages 13-19). Given the age range considered in this
model, only the values for ages 18 and 19 were used. Figure III is a graphical
representation of the distribution by age of onset of illness used for this study. Upon
entering this health state, an attrition rate was applied to determine the number of
patients that would transition to the absorbing state ‘DEATH’. Patients, who did not
transition into the absorbing state, were prescribed treatment with a 1st line agent.4
Patients were then guided through pathways that included tolerance, response, and
remission; based on the treatment algorithm previously described. At the end of the 6
month cycle, patients who remitted with 1st line agents exited the ‘FAMILY PHYSICIAN
ASSESSMENT’ health state and proceeded to the ‘MAINTENANCE 1ST LINE’ health
state.
The decision tree pathways within each health state are illustrated in Figures IV-XI on
pages 65-72.
HEALTH STATE 2: SPECIALIST ASSESSMENT 2ND LINE
Patients who entered the ‘SPECIALIST ASSESSMENT 2ND LINE’ health state: were
intolerant to 1st line agents, had responded to 1st line medication but had not remitted, or
had neither responded nor remitted to 1st line therapy. Their treatment was managed by
a psychiatrist who prescribed them 2nd line treatment options. After allowing for attrition
rates at the beginning of each cycle, patients proceeded along pathways that considered
4 Please refer to Treatment algorithm section (Page 67) for description of treatment lines
56
FIGURE IV: DECISION TREE WITHIN FAMILY PHYSICIAN ASSESSMENT HEALTH STATE All cause mortalityp_ACM_Table[startage+_stage/2 ]
Death
Intolerantp_intolerance_1stline
Specialist Assessment 2nd Line
Remission p_remission_1stline
Maintenance 1st line
Remission p_remission_1stline
Maintenance 1st line
No remission #
Specialist Assessment 2nd Line
Titrate (Optimize))Response
p_response_1stline-p_remission_1stline
No response #
Specialist Assessment 2nd Line
No remission#
Tolerant#
Switch to another 1st lineIntolerantp_intolerance_1stline
Remissionp_remission_1stline
Maintenance 1st line Remission
p_remission_1stlineMaintenance 1st line
No remission #
Specialist Assessment 2nd Line
Titrate (Optimize)Response p_response_1stline-p_remission_1stline
Intolerant p_intolerance_1stline
Specialist Assessment 2nd Line
Remission p_remission_1stline
Maintenance 1st line
Remissionp_remission_1stline
Maintenance 1st line No remission (1st line + benzo)
#Specialist Assessment 2nd Line
TitrateResponsep_response_1stline-p_remission_1stline
No response #
Specialist Assessment 2nd Line
No remission#
Tolerant#
Switch to another 1st lilneNo response#
No remission#
Tolerant #
1st line pharmacotherapyNo Death#
Family Physician Initial Assessment1
Specialist Assessment 2nd Line0
[+]
Specialist Assessment 3rd line0
[+]
Maintenance 1st line
0 [+]
Maintenance Benzo only0
[+]
Maintenance 1st Line + Benzo0
[+] Maintenance 3rd Line
0 [+]
Treatment stopped0
[+] Death
0
Patient Seeking Treatment
57
FIGURE V: DECISION TREE WITHIN SPECIALIST ASSESSMENT 2ND LINE HEALTH STATE Family Physician Initial Assessment
1 [+]
All cause mortalityp_ACM_Table[startage+_stage/2 ]
Death
Specialist Assessment 3rd lineIntolerant p_intolerance_benzo
Remission p_remission_benzo
Maintenance Benzo onlyRemission
p_remission_benzo Maintenance Benzo only
No remission#
Specialist Assessment 3rd line
Titrate (optimize)Responsep_respose_remission_benzo_truncated
No Response #
Specialist Assessment 3rd line
No Remission#
Tolerant#
Benzodiazepine onlyp_2ndline_benzo
Intolerantp_intolerance_benzo
Specialist Assessment 3rd line
Remissionp_remission_1st_benzo
Maintenance 1st Line + Benzo
Remissionp_remission_1st_benzo
Maintenance 1st Line + Benzo
No remission#
Specialist Assessment 3rd line
Titrate (optimize)Responsep_respose_remission_benzo_truncated
No reponse#
Specialist Assessment 3rd line
No remission#
Tolerant#
1st line + benzo adj #
No death#
Specialist Assessment 2nd Line0
Specialist Assessment 3rd line0
[+] Maintenance 1st line
0 [+]
Maintenance Benzo only0
[+]
Maintenance 1st Line + Benzo0
[+]
Maintenance 3rd Line0
[+]
Treatment stopped0
[+] Death
0
Patient Seeking Treatment
58
FIGURE VI: DECISION TREE WITHIN SPECIALIST ASSESSMENT 3RD LINE HEALTH STATE
Family Physician Initial Assessment1
[+]
Specialist Assessment 2nd Line0
[+]
All cause mortalityp_ACM_Table[startage+_stage/2 ]
Death
Remission p_remission_3rd_line
Maintenance 3rd Line
Remission p_remission_3rd_line
Maintenance 3rd LineNo remission
#Maintenance 3rd Line
TitrateResponseIf(p_response_3rd_line-p_remission_3rd_line<0;0;p_response_3rd_line-p_remission_3rd_line)
10-Day hospitalizationProb_Hospitalization_3rd_line_failure
Specialist Assessment 3rd lineTreatment Stopped
#Treatment stopped
No response#
No remission#
No death#
Specialist Assessment 3rd line0
Maintenance 1st line
0 [+]
Maintenance Benzo only0
[+]
Maintenance 1st Line + Benzo0
[+]
Maintenance 3rd Line0
[+] Treatment stopped
0 [+]
Death0
Patient Seeking Treatment
59
FIGURE VII: DECISION TREE WITHIN MAINTENANCE 1ST LINE THERAPY HEALTH STATE
Family Physician Initial Assessment1
[+]
Specialist Assessment 2nd Line0
[+]
Specialist Assessment 3rd line0
[+]
All cause mortalityp_ACM_Table[startage+_stage/2 ]
Death
Treatment stoppedp_treatmentdisc_1stline
Treatment stoppedRelapse
p_relapseSpecialist Assessment 2nd Line
No relapse#
Maintenance 1st line
Continue treatment#
No death#
Maintenance 1st line
0
Maintenance Benzo only0
[+]
Maintenance 1st Line + Benzo0
[+]
Maintenance 3rd Line0
[+]
Treatment stopped0
[+]
Death0
Patient Seeking Treatment
60
FIGURE VIII: DECISION TREE WITHIN MAINTENANCE BENZODIAZEPINE MONO-THERAPY HEALTH STATE
Family Physician Initial Assessment1
[+]
Specialist Assessment 2nd Line0
[+]
Specialist Assessment 3rd line0
[+]
Maintenance 1st line
0 [+]
All cause mortalityp_ACM_Table[startage+_stage/2 ]
DeathTreatment stoppedp_treatmentstopped_benzo
Treatment stoppedRelapse
p_relapseSpecialist Assessment 2nd Line
No relapse#
Maintenance Benzo only
Continue treatment#
No death#
Maintenance Benzo only0
Maintenance 1st Line + Benzo0
[+]
Maintenance 3rd Line0
[+]
Treatment stopped0
[+]
Death0
Patient Seeking Treatment
61
FIGURE IX: DECISION TREE WITHIN MAINTENANCE 1ST LINE & BENZODIAZEPINETHERAPY HEALTH STATE
Family Physician Initial Assessment1
[+]
Specialist Assessment 2nd Line0
[+]
Specialist Assessment 3rd line0
[+]
Maintenance 1st line
0 [+]
Maintenance Benzo only0
[+] All cause mortalityp_ACM_Table[startage+_stage/2 ]
DeathTreatment stoppedp_treatmentdisc_1stline_Benzo
Treatment stoppedRelapse
p_relapseSpecialist Assessment 2nd Line
No relapse# Maintenance 1st Line + Benzo
Continue treatment#
No death#
Maintenance 1st Line + Benzo0
Maintenance 3rd Line0
[+]
Treatment stopped0
[+]
Death0
Patient Seeking Treatment
62
FIGURE X: DECISION TREE WITHIN MAINTENANCE 3RD LINE THERAPY HEALTH STATE
Family Physician Initial Assessment1
[+]
Specialist Assessment 2nd Line0
[+]
Specialist Assessment 3rd line0
[+]
Maintenance 1st line
0 [+]
Maintenance Benzo only0
[+] Maintenance 1st Line + Benzo
0 [+]
All cause mortalityp_ACM_Table[startage+_stage/2 ]
DeathTreatment stoppedp_treatmentdisc_3rd_line
Treatment stopped
Relapsep_relapse
Specialist Assessment 3rd line
No relapse#
Maintenance 3rd Line
Continue treatment#
No death#
Maintenance 3rd Line0
Treatment stopped0
[+]
Death0
Patient Seeking Treatment
63
FIGURE XI: DECISION TREE WITHIN TREATMENT STOPPED HEALTH STATE Family Physician Initial Assessment
1 [+]
Specialist Assessment 2nd Line0
[+]
Specialist Assessment 3rd line0
[+]
Maintenance 1st line
0 [+]
Maintenance Benzo only0
[+] Maintenance 1st Line + Benzo
0 [+]
Maintenance 3rd Line0
[+] All cause mortalityp_ACM_Table[startage+_stage/2 ]
Death
Treatment seekingp_treatment_seeking
Family Physician Initial Assessment
Non-treatment seeking#
Treatment stopped
Non-Spontaneous remission#
Spontaneous Remissionp_spontaneous_remission
Treatment stopped
No Death#
Treatment stopped0
Death0
Patient Seeking Treatment
64
tolerance, remission and response rates, based on the treatment algorithm for 2nd line
therapy. After 6 months in this health state, patients who remitted with either treatment
option exited this state and transitioned to either the ‘MAINTENANCE THERAPY FOR
2ND LINE (1ST LINE PLUS BENZODIAZEPINE) or MAINTENANCE THERAPY FOR 2ND
LINE (BENZODIAZEPINE MONOTHERAPY). Patients who were intolerant to either
treatment option or who responded but did not remit with 2nd line treatment options
exited the ‘SPECIALIST ASSESSMENT 2ND LINE’ health state and transitioned to the
‘SPECIALIST ASSESSMENT 3RD LINE’ health state.
HEALTH STATE 3: SPECIALIST ASSESSMENT 3RD LINE
The ‘SPECIALIST ASSESSMENT 3RD LINE’ health state received patients who were
intolerant or who did not respond to 1st or 2nd line pharmacotherapy and whose treatment
was managed by a psychiatrist. In this health state, patients were prescribed 3rd line or
agents for treatment resistance. After allowance for attrition, patients then moved along
pathways based on the treatment algorithm for 3rd line therapy. At the end of the 6
month cycle, patients who remitted or responded when managed with 3rd line agents,
exited the ‘SPECIALIST ASSESSMENT 3RD LINE’ health state and transitioned to the
‘MAINTENANCE THERAPY FOR 3rd LINE’ health state. Patients who died from any
cause exited the ‘SPECIALIST ASSESSMENT 3RD LINE’ health state and were
absorbed into the ‘death’ state at the beginning of each cycle.
65
HEALTH STATES 4 THROUGH 7: MAINTENANCE THERAPY HEALTH
STATES
For each of the 4 maintenance therapy health states, patients either continued or
discontinued treatment after allowance was made for attrition. Those who continued in
the model were prescribed maintenance doses of their respective treatment lines.
Patients who continued treatment and relapsed at the end of the 1st or 2nd line
maintenance health states exited their respective health states and transitioned into the
‘SPECIALIST ASSESSMENT 2ND LINE’ health state. Patients who continued treatment
and relapsed while in the ‘SPECIALIST ASSESSMENT 3RD LINE’ health state, exited the
health state and re-entered the ‘SPECIALIST ASSESSMENT 3RD LINE’ health state.
Patients, who discontinued treatment while in their respective maintenance health states,
exited these states and entered the ‘TREATMENT DISCONTINUED’ health state.
Patients who died from any cause exited their respective maintenance health states and
proceeded to the ‘DEATH’ health state at the beginning of each cycle.
HEALTH STATE 8: TREATMENT DISCONTINUED
Patients who discontinued treatment entered the ‘TREATMENT DISCONTINUED’ health
state. As in previous health states, patients were subject to an attrition rate. They were
assumed to have discontinued treatment because either they experienced ‘spontaneous
remission’ or they continued to experience GAD-related symptoms but had decided not
to continue treatment. Patients who experienced ‘spontaneous remission’ or who
continued to suffer from GAD-related symptoms and did not take medication at the end
of 6 months would re-enter this health state. Patients who continued to suffer from GAD-
related symptoms but decided to seek treatment after 6 months in this health state,
66
exited this health state and re-entered the initial health state (‘FAMILY PHYSICIAN
ASSESSMENT’); in effect, beginning the treatment algorithm anew.
HEALTH STATE 9: DEATH (ABSORBING STATE)
‘DEATH’ was the absorbing state; meaning patients who entered this state were
removed from the model.
TREATMENT ALGORITHM
The reference case scenario considered pharmacotherapeutic management of GAD,
while a combination of psychotherapy and pharmacotherapy was assessed in a
sensitivity analysis. Pharmacotherapeutic choices and treatment algorithms for the
management of GAD were based on CPA guideline recommendations as well as expert
input from clinicians participating in the present thesis committee.5
For this study, first line therapy included the use of either an SSRI (paroxetine, sertraline,
or escitalopram) or SNRI (venlafaxine XR). All of these first line agents have the
indication for GAD in Canada, and are listed in Ontario Drug Benefit (ODB)
Formulary.[128] CPA recommendations for the use of paroxetine, escitalopram and
venlafaxine were based on level 1 evidence, while inclusion of sertraline was based on
level 2 evidence. [21] The thesis committee concurred that these agents are prescribed
as 1st line therapy in Ontario.
Second line therapy for this study was either: 1st line treatment augmented with a
benzodiazepine or a benzodiazepine as monotherapy (i.e., alprazolam, bromazepam, 5Please refer to the Thesis Committee section (Page 83) for a list of committee members.
67
diazepam or lorazepam). Adjunctive therapy is recommended by the CPA for patients
who have had inadequate response to 1st line agents after titration. However, it should
be noted that CPA guidelines for 2nd line pharmacotherapy not only included the
benzodiazepines (level 1 evidence) previously mentioned, but also buspirone (level 1
evidence), imipramine (level 1 evidence), pregabalin (level 1evidence), and bupropion
(level 2 evidence). Buspirone, imipramine, pregabalin, and bupropion were not
considered in this study. Buspirone was not considered, given evidence cited by the
CPA guidelines of reduced efficacy in patients who have previously used
benzodiazepines. [129] Imipramine was not considered due to the risk of death from
overdose.[21] Pregabalin was excluded from this study given the agents’ limited clinical
use in Canada and is not listed as a benefit in Ontario. [21] Furthermore, according to
the ODB formulary, bupropion has a limited use authorization for the treatment of
depression and was therefore excluded from consideration in this study. The thesis
committee concurred with the description of 2nd line therapy for this study.
Third line therapy in this study included citalopram (level 4 evidence), mirtazapine (level
3 evidence), trazodone (level 2 evidence) adjunctive olanzapine with fluoxetine (level 2
evidence) or adjunctive risperidone with benzodiazepine (level 2 evidence). Hydroxyzine
(level 1 evidence) was also recommended by the CPA; however, the agent is no longer
covered as a benefit by the ODB formulary and was not considered in this study. The
thesis committee concurred with the description of 3rd line therapy for this study.
For each 1st line treatment option, an initial dose was prescribed by a family physician to
assess the risk for intolerance. According to expert opinion, intolerance to the initial
dose should be assessed after 2 weeks. In this study, it was assumed that those
patients who were tolerant to medication achieved partial response, thereby requiring an
68
increase to a ‘maintenance’ level dosage in order to achieve full response or remission.
[Please refer to dosing section on page 70 for dosing levels and sources used]. Doses
were then titrated to an optimal level for patients who responded to pharmacotherapy but
failed to achieve remission. Patients who were intolerant to the 1st choice of a 1st line
medication were switched to another 1st line agent.
Those patients who were intolerant or who failed to achieve remission to both 1st line
treatment options were prescribed an initial dose of a 2nd line treatment option by a
psychiatrist to assess intolerance to that drug. Patients who tolerated the medication
were assumed to have achieved partial response, thereby requiring an increase to a
‘maintenance’ level dosage in order to achieve full response or remission. Titration to
optimal dosing was prescribed for patients who responded but did not remit with either of
the 2nd line treatment options. Since one of the choices for 2nd line treatment included
the adjunctive use of a 1st line agent and some patients were intolerant to both 1st line
choices; switching between treatment options for 2nd line therapy was not an option.
Patients who were intolerant to, or who did not achieve remission with either 1st or 2nd
line treatment options were prescribed 3rd line or agents for treatment resistance.
Intolerance to 3rd line agents was not considered, given structure of the model. The
incidence of those intolerant to all 3 treatment lines was not considered substantial by
the thesis committee. Patients who did not respond to 3rd line agents were hospitalized
for 10 days prior to re-entering the ‘Specialist assessment 3rd line’ health state or
discontinued treatment altogether. Doses were titrated for patients who responded to
treatment but did not achieve remission. Pharmacotherapy was continued for as long as
physicians continued prescribing the medication or until patients stopped treatment.[21]
69
Cognitive behavioural therapy in combination with pharmacotherapy was considered for
a sensitivity (scenario) analysis. Based on evidence cited by CPA guidelines, patients
were treated with approximately 10 sessions of CBT.[98-100] Although a combination
of drug and psychotherapy is not expressly recommended by the guidelines, available
evidence for combination therapy cited by the CPA was based on diazepam and
cognitive behavioural therapy. Therefore, this combination was considered in a sub-
group analysis. The cognitive behavioural therapy consisted of 10 sessions over a 6
month assessment6 period conducted by a psychiatrist (SPECIALIST ASSESSMENT
FOR 2ND LINE THERAPY) for patients prescribed any benzodiazepine (either mono or
adjunctive therapy.
DOSING
Initial and optimized dosing for 1st and 2nd line pharmacotherapy were taken from the
product monographs where indicated for patients with GAD.[114-121] Maintenance
doses were taken from peer reviewed literature.[109] Doses for 3rd line treatment were
taken from evidence cited by the CPA[21,122-6]. Table 8 lists initial, maintenance and
optimized doses for all treatment lines used in this model.
CLINICAL VARIABLES
Clinical variables and their inputs for this study are presented in Table 9. The clinical
rates used in this model were: intolerance to medication, response, remission,
‘spontaneous remission’, discontinuation of treatment, and relapse. Clinical parameters
were chosen based on the treatment algorithm recommended by the CPA, and by the
clinical experience of Dr. Arun Ravindran, a member of this thesis committee.
6 Refer to Process section (Page 61) for description of health states
70
TABLE 8: INITIAL, MAINTENANCE AND TITRATED DOSES BY TREATMENT LINE FOR THIS STUDY Daily Dose (mg) Initial Maintenance Maximum Sources 1st Line Escitalopram 10.0 15.0 20.0 Product Monograph[119], Vasile[109] Paroxetine 20.0 26.5 50.0 Product Monograph[117]. Vasile[109] Sertraline 25.0 102.5 200.0 Product Monograph[118] Vasile[109] Venlafaxine 37.5 172.0 225.0 Product Monograph[120], Vasile[109] Benzodiazepine Alprazolam 0.25 2.0 3.0 Product Monograph[114], Vasile[109] Bromazepam 2.0 3.5 6.0 Product Monograph[121] Diazepam 2.0 13.0 40.0 Product Monograph[115], Vasile[109] Lorazepam 2.0 2.8 6.0 Product Monograph[116], Vasile[109] 3rd Line Citalopram 10.0 33.0 60.0 Varia and Rauscher[122] Fluoxetine 20.0 20.0 20.0 Pollack[123] Mirtazapine 30.0 30.0 30.0 Gambi[124] Olanzapine 2.5 8.7 20.0 Pollack[123] Risperidone 0.5 1.1 1.5 Brawman-Minzter[125] Trazodone 150.0 400.0 340.0 Rickels[126]
71
TABLE 9: CLINICAL VARIABLES AND THEIR VALUE INPUTS (I.E., RATES) USED TO POPULATE THE MARKOV MODEL
7 Please see Results section (page 92) for complete results from meta-analysis
Probabilities Clinical parameters Mean Low Value High Value Distribution Source 1st Line Treatment Intolerance 0.0900 0.0470 0.1470 Triangular Machado [101] Remission 0.3970 0.3520 0.4410 Triangular Bereza [104] Response 0.6772 0.6409 0.7136 Triangular Bereza [104] Discontinued treatment 0.7138 0.7022 0.7266 Triangular Mullins [106] Benzodiazepine Treatment Intolerance 0.0730 0.0450 0.1010 Triangular Pooled results Remission 0.5350 0.4400 0.6300 Triangular Bereza [104]7 Response 0.5680 0.4640 0.6720 Triangular Bereza [104]5 Discontinued Treatment 0.5260 0.3630 0.6613 Triangular Ohayon [107] 1st Line + Benzodiazepine Treatment Intolerance 0.0900 0.0470 0.1470 Triangular Machado [101] Remission 0.4100 0.3630 0.4560 Triangular Bereza [104] Response 0.6430 0.6020 0.6840 Triangular Bereza [104] Discontinued Treatment 0.5260 0.3630 0.6613 Triangular Vasile [109] 3rd Line Treatment Remission 0.3570 0.2310 0.4820 Triangular Bereza [104] Response 0.5490 0.3970 0.7020 Triangular Bereza [104] Discontinued Treatment 0.5260 0.3630 0.6613 Triangular Vasile [109] Relapse 0.0250 0.0200 0.0500 Triangular Yonkers [110] Spontaneous remission 0.2000 0.2250 0.2500 Triangular Ballenger [111]
72
INTOLERANCE
Intolerance rates to pharmacotherapy were based on peer-reviewed published
literature.[101,102] Meta-analytic rates of adverse drug reactions in clinically depressed
patients were used as a proxy for intolerance of SSRI or SNRI agents.[101] It was thus
assumed that intolerance was not affected by the type of health condition of the patient,
and that the adverse events were solely caused by the drug effect. Rates of intolerance
to benzodiazepines were derived from the statistical pooling of dropout rates due to
adverse events reported by evidence for 2nd line therapy cited in the CPA guidelines.
Where reported, the number of patients assigned to a benzodiazepine group as well as
those who dropped out due to adverse events was extracted from CPA cited evidence.
Data were pooled using the dropout rate itself as an effect size and the binomial theorem
to determine its variance; a technique described by Einarson.[103] Intolerance rates
were not required for patients treated with 3rd line agents, as per model structure
described previously.
REMISSION AND RESPONSE
To determine the remission and response rates for this study, a meta-analysis was
performed using CPA cited evidence and by extension, any additional relevant (original)
articles retrieved from the reference list of meta-analytic evidence considered in the
2006 guidelines for the management of GAD. No further criteria were considered with
respect to the clinical and demographic make-up of the population considered in this
meta-analysis. Evidence was considered separately for each of the three treatment lines
used in this study. Meta-analytic evidence, duplicate publications, and studies with non-
extractable data were excluded. Preliminary results of the analysis have been
73
previously presented.[104] Complete results from the meta-analysis are presented in the
results section in Chapter 3 below.
Neither placebo arms, nor study arms of agents not recommended by the CPA for the
management of GAD were considered. For example, propranolol, although it may have
been used in the past for this indication, is specifically not recommended by the
guidelines; therefore, evidence for that drug was not included.
The following data were extracted from included articles: the number of subjects treated,
number of patients responding or remitting (or corresponding response or remission
rates), baseline and endpoint scores and standard deviations (SDs), baseline to
endpoint or mean reductions in scores of the Hamilton Rating Scale for Anxiety (HAM-A)
as well as standard deviations or standard error values. Where necessary and possible,
data were extracted from figures to obtain a reasonably accurate estimate of relevant
values not otherwise reported in text or tables. Data were extracted by one author
(BGB) and validated by a second author (MM).
When using the inverse variance method for deriving pooled estimates from published
studies, SDs are necessary for calculations and completeness of the data and relies on
the quality of study reporting. In this meta-analysis, missing SD values were derived
based on scores from studies that reported both HAM-A scores as well as their
corresponding SDs. Each reported SD was squared to calculate the variance. A ratio
was then calculated of the mean HAM-A score and the variance from the related study.
This process was repeated for each study that reported both HAM-A and SD values. A
geometric mean was then taken of the ratios; this mean ratio was then multiplied by the
74
mean HAM-A scores of studies that did not report SDs, to derive a variance for that
study. The square root of the variance was then taken to calculate the SD. When both
the HAM-A score and its SD were not reported by a study, then data manipulation was
not possible, thereby excluding the study from the analysis.
The outcomes of interest were the meta-analytic probability of remission (defined as a
score on HAM-A≤7) and response [defined as a 50% reduction in HAM-A scores, or
clinical global impressions-improvement score (CGI-I) of 1 or 2] by line of treatment.
Review Manager® software version 5 (Cochrane Collaboration 2008) was used to
combine results across study arms. A random effects model using the inverse variance
weighting was applied in all calculations.[136] Tau-squared estimated between-study
variance. Heterogeneity of effects was assessed using Chi-squared and I-squared
statistics. Critical p-values for determining heterogeneity were set at <0.1, and >30% for
the I-squared. [105]
TREATMENT DISCONTINUATION
Treatment discontinuation rates were derived or taken directly from peer-reviewed
literature for each treatment line where possible. Discontinuation rates specifically
attributed to GAD patients treated with 1st line agents were not found and were therefore
taken from a retrospective cohort study comparing discontinuation rates for SSRIs
among patients with social anxiety disorder and post-traumatic stress disorder.[106] A
discontinuation rate of treatment for GAD patients treated with benzodiazepines was
also not found. The discontinuation rate for benzodiazepines was taken from a survey
75
conducted in the United Kingdom of 4972 non-institutionalized individuals 15 years of
age or older. In that study, consumption for a period of 12 months or more was reported
by 47.4% of responders who were prescribed anxiolytics. [107] One-hundred percent
minus that rate was used as a proxy for a discontinuation rate in patients treated with
benzodiazepines. Wilson’s score method for deriving a confidence interval from a single
proportion was used to derive upper and lower limits for this variable.[108]
CI = [(2n + Z2) ± Z√ (z2 + 4n*(1 - n/N)]/(2n + 2Z2)
Where:
Z= the standard normal deviate associated with alpha error (i.e., Z=1.96 when α = 0.05,
or 95% confidence); n=observed cases; N=total sample.
The discontinuation rate for adjunctive benzodiazepine therapy with SSRI/SNRI was
taken from a naturalistic longitudinal study of GAD patients.[107] No treatment
discontinuation rates were found for GAD patients treated with 3rd line agents. Treatment
discontinuation for GAD patients on adjunctive benzodiazepine with 1st line agents was
used as a proxy.
RELAPSE
Relapse rates reported from the Harvard/Brown Anxiety Research Project
(HARP), observational, longitudinal study of patients with GAD patients and extrapolated
for the purposes of this model.[108] In the HARP study, relapse was defined when
patients in full remission met full DSM-III-R criteria, severe symptoms, or extreme
impairment or pronounced interference in functioning. Yonkers et al. reported
cumulative relapse rates over an 8 year follow up period segregated by gender.[110]
Cumulative probability for both genders was computed by weighting the cumulative
76
probability of each gender by their respective sample size. Individual yearly rates were
then computed.
SPONTANEOUS REMISSION
A spontaneous remission rate of 22.5% was assigned to a cohort of patients who had
discontinued treatment. This rate was taken from a consensus statement on GAD from
the international consensus group on depression and anxiety. No evidence was given for
this rate nor was spontaneous remission defined in their statement.[111] This rate was
attributed to the ‘Treatment discontinuation’ health state only. This model assumed that
patients who did not ‘spontaneously remit’ were experiencing GAD related symptoms
and may or may not decide to seek treatments.8
RESOURCE USE AND COSTS Economic input variables and their values for this study are presented in Table 10.
Economic parameters included; physician and pharmacist fees, medication costs, as
well as rates related to hospitalization and seeking treatment after treatment
discontinuation.
PHYSICIAN COSTS
Fees for family practice and psychiatry consults were taken from the Government of
Ontario’s Ministry of Health and Long Term Care Schedule of Benefits for Physician
Services.[127] Patients entering the Family physician, Specialist assessment 2nd or 3rd
line treatment health states were subject to an initial physician assessment as well as
follow-up visits to assess intolerance, response or remission. Follow-up visit(s) in the
8 See seeking treatment behaviour section page 81
77
Family physician, Specialist assessment 2nd or 3rd line health states ranged from 2-5, 1-5
and 1-3 respectively per cycle. Maintenance health states were attributed 1 follow up
visit per cycle. No physician costs were attributed to the ‘TREATMENT
DISCONTINUED’ or ‘DEATH’ states. For sub-group analysis, an additional 10 sessions
of CBT with a psychiatrist were attributed to patients in the Family physician, Specialist
assessment 2nd or 3rd line treatment health states.
DRUG COSTS
Costs for pharmacotherapy were based on the Government of Ontario, Ministry of
Health and Long Term Care, Ontario Drug Benefit Formulary/Comparative Drug
Index.[128] Costs for initial, maintenance and optimized daily doses were considered
separately. No resource utilization data was available to determine the proportional
usage of each drug by a cohort of GAD patients. Therefore, in this study it was
assumed that each drug was utilized in equal proportion. The daily cost for each drug
was derived by multiplying the dose in milligrams by the cost per milligram. An arithmetic
mean of the daily cost of the drugs included in the respective treatment line was then
calculated to determine the daily cost for each treatment line. The daily cost was then
multiplied by the number of days the patient was exposed to either an initial,
maintenance or optimized dose.
DISPENSING FEES
The dispensing fee was taken from the Government of Ontario, MoHLTC, ODB:
Dispensing fee web site.[129] Prescribed drugs were assumed to have been filled. A
dispensing fee was charged for the initial prescription, or when medication was titrated
or switched. One prescription fee of $7 CAD was incurred for each 6-month cycle in a
‘maintenance’ health state.
78
HOSPITALIZATION COSTS
The 2002 Health Canada report on mental illness in Canada showed that patients with
anxiety disorders incurred 10 hospitalization days in 1999 as a result of their disorder.
This rate of utilization changed little between 1987 and 1999; ranging from 9.6 to 11.1
days.[113] Cost of hospitalization per day was an estimation of total acute care inpatient
costs for 2004 from the Canadian Institute for Health Information(CIHI) data base
available on their web site.[130] Total hospitalization costs attributed to GAD was
calculated by multiplying the number of hospitalization days by the total acute care
inpatient cost. Inpatient costs were inflated to 2008 dollars using the Consumer Price
Index.[137]
TREATMENT SEEKING BEHAVIOUR
The rate of patients’ seeking treatment was assigned to patients who were once again
seeking treatment after discontinuation of therapy. This rate was based on a nationally
representative household survey of respondent s 18 years and older in the United States
(i.e., the National Comorbidity Survey-Replication). [112]
INDIRECT COSTS
Indirect costs were based on number of days absent from work in a given month
(absenteeism) and the industrial wage rate. Kessler reported absenteeism based on two
surveys: National Co-morbidity Survey (NCS) as well as the Midlife Development in the
United States Survey (MDUSS). Kessler reported probabilities for 0 to 6 days or more of
absenteeism from work per month.[15] The weighted average from both surveys was
rounded to approximately 2 days per month. This rate was multiplied by the average
79
TABLE 10: ECONOMIC INPUT VARIABLES AND ASSOCIATED COSTS* Variable Distribution Low High Value Source Direct costs
Clinician/hospital charges Pharmacist dispensing fee Uniform $6.00 $15.00 $10.50 MOHLTC: Dispensing Fees[129] Initial assessment family practice Uniform $42.53 $78.98 $60.75 MOHLTC: Schedule of Benefits[127] Follow-up family practice Uniform $21.25 $39.46 $30.35 MOHLTC: Schedule of Benefits[127] Initial assessment psychiatry Uniform $113.74 $211.22 $162.48 MOHLTC: Schedule of Benefits[127] Follow-up psychiatry Uniform $57.72 $107.19 $82.45 MOHLTC: Schedule of Benefits[127] Hospitalization cost (10-day period) Uniform $3,416.00 $6,340.00 $4,878.00 Health Canada[113];CIHR[130] Pharmacotherapy (Mean daily cost) Initial 1st line Uniform $0.70 $1.30 $1.00 MOHLTC:ODB: Formulary[128] Initial benzodiazepine adjunctive 1st line Uniform $0.80 $1.49 $1.15 MOHLTC:ODB: Formulary[128] Initial benzodiazepine monotherapy Uniform $0.10 $0.19 $0.15 MOHLTC:ODB: Formulary[128] Initial 3rd line Uniform $0.83 $1.55 $1.19 MOHLTC:ODB: Formulary[128] Maintenance 1st line Uniform $0.96 $1.78 $1.37 MOHLTC:ODB: Formulary[128] Maintenance benzodiazepine adjunctive 1st line Uniform $1.12 $2.07 $1.59
MOHLTC:ODB: Formulary[128]
Maintenance benzodiazepine monotherapy Uniform $0.16 $0.30 $0.23 MOHLTC:ODB: Formulary[128] Maintenance 3rd line Uniform $2.04 $3.79 $2.91 MOHLTC:ODB: Formulary[128] Titrated 1st line Uniform $1.19 $2.20 $1.70 MOHLTC:ODB: Formulary[128] Titrated benzodiazepine adjunctive 1st line Uniform $1.41 $2.63 $2.02 MOHLTC:ODB: Formulary[128] Titrated benzodiazepine monotherapy Uniform $0.16 $0.29 $0.22 MOHLTC:ODB: Formulary[128] Titrated 3rd line Uniform $4.04 $7.50 $5.77 MOHLTC:ODB: Formulary[128]
Indirect costs Absenteeism 6 months (2 days/month) Uniform $1,465.12 $2,719.60 $2,092.72 Kessler[67];Statistics Canada [132]
Resource use Treatment seeking rate Triangular 0.1000 0.3300 0.2600 Wang[112] Hospitalization rate Triangular 0.0000 0.1200 0.0600 Health Canada [113] Discount rate Uniform 0.0000 0.0500 0.0500
* Canadian 2008 dollars CIHR=Canadian Institutes of Health Information MOHLTC= Ministry of Health and Long Term care: Ontario Drug Benefit Formulary/Comparative Index ODB=Ontario Drug Benefit
80
industrial hourly wage rate for 2008 of $21.08 to calculate indirect costs and assumed an
8 hour work day.[132] Costs related to absenteeism were attributed to the assessment
health states (i.e. Family physician, Specialist assessment 2nd line and Specialist
assessment 3rd line), as well as to patients in the ‘TREATMENT DISCONTINUED’ health
state who have GAD related symptoms but are not seeking treatment for the disorder.
(i.e. 1 minus the spontaneous remission rate). It was assumed that patients not
achieving ‘spontaneous remission’ would be subject to impairment. Since indirect costs
were measured as foregone wages, indirect costs were considered only for patients 65
years of age or younger.
DISCOUNTING
In accordance with the Canadian guidelines for economic evaluations of health
technologies, costs were discounted at a rate of 5% each year and altered to 0% in the
sensitivity analysis.[146]
DISTRIBUTIONS
Triangular distributions were assumed for response, remission, treatment
discontinuation, treatment seeking behaviour, relapse rates, and spontaneous remission
and hospitalization rates, where confidence intervals were used to represent minimum
and maximum values, and point estimates represented most likely values. Confidence
intervals for response and remission rates were based on outcomes from the meta-
analysis described earlier. Confidence intervals for the remaining variables were based
on peer-reviewed literature as previously described.
Uniform distributions were assumed for costs, fees, and absenteeism. The range of
values for the variables was determined by multiplying by an arbitrary factor of ± 30%
81
around the likeliest value. Relevant resource utilization data related to GAD in Canada
was not available. As such mean of logs and related standard deviation data were not
computable; therefore use of a log normal (and most likely appropriate) distribution for
costs was not possible.
All cause mortality data were retrieved from published tables with no missing values.
Age of onset distribution was bi-modal, as previously described.
OUTCOMES
The primary outcome was the mean and standard deviation (SD) of the lifetime COI per
patient diagnosed with GAD. COI was reported in 2008 Canadian dollars (CAD$). The
primary outcome was generated from 100,000 iterations of 2nd order Monte-Carlo
simulations. That is to say, that the 2nd order Monte-Carlo simulation randomly drew
from the values of distributions for each model parameter.
Secondary outcomes included epidemiologic parameters collected from the economic
model and included the mean and SD of: the age of GAD onset, frequency of relapse
per patient, proportion of treatment-resistant patients, duration of treatment
discontinuation. Secondary outcomes also include the meta-analytic probability of
remission (defined as score on HAM-A≤7) and response [defined as a 50% reduction in
HAM-A scores, or clinical global impressions-improvement score (CGI-I) of 1 or 2].
82
VARIABILITY AND UNCERTAINTY
Structural validation of the model was performed through a ‘debugging’ process. This
process involved changing one variable at a time in the model, over a range of extreme
values that were wider than a ‘plausible’ range for that variable. [134]. For clinical
variables and probabilities, values from 0 to 1 were input into the model parameters to
determine whether they would produce expected results. For economic variables, value
inputs from 0 to twice the upper limit were input into the model. A one-way sensitivity
analysis was also performed to assess the impact of the parameter on the base case
result. The upper and lower values around the point estimate for each parameter were
used to perform the probabilistic sensitivity analysis; and were presented using tornado
diagrams. Breakdown analysis was also performed to assess the impact of each
variable on the base-case result. Validation for the course of illness and management of
GAD reflected in the Markov model was provided by expert opinion from members of the
thesis committee for this study.
COMMITTEE MEMBERS
Members of the thesis committee for this study were: Dr. Arun V. Ravindran, Dr. Beth
Sproule, Dr. Márcio Machado, Dr. Emmanuel Papadimitropoulos and Dr. Thomas R.
Einarson.
Dr. Ravindran MD, PhD, MRCPsych, FRCPC is a professor in the department psychiatry
at the University of Toronto as well as the Clinical Director of the Mood and Anxiety
Disorders Program at the Centre for Addiction and Mental Health (CAMH). Dr. Sproule
PharmD is an assistant professor in the Leslie Dan Faculty of Pharmacy at the
83
University of Toronto and is an advanced practice pharmacist and clinician scientist at
CAMH. Further validation for the pharmacotherapeutic management of GAD, the clinical
and economic parameters as well as statistical and technical methods used for this
study was provided by Dr Márcio Machado, Dr Emanuel Papadimitropoulos and Dr.
Thomas R. Einarson; experts in the field of systematic reviews, cost analysis, and
decision modeling. Márcio Machado, PharmD, PhD, and Emanuel Papadimitropoulos
PhD are assistant professors in the Faculty of Pharmacy at the University of Toronto.
Thomas Einarson PhD was the academic supervisor for this thesis and is an associate
professor in the Leslie Dan Faculty of Pharmacy at the University of Toronto.
84
SECTION 3: RESULTS
COST OF ILLNESS
REFERENCE CASE
The mean value of the lifetime COI per patient with GAD was (2008 CAD) $31,213
(SD=$9,100). The distribution statistics are presented in Table 10. The distribution is
graphically represented in Figure V. Health state probabilities are presented in Figure VI.
TABLE 11:REFERENCE CASE RESULTS* Statistic Base case
Mean $31,213
SD $9,100
Minimum $1,089
2.5% CL $5,250
10% CL $17,398
Median $33,855
90% CL $39,648
97.5% CL $42,095
Maximum value $50,248
* Values are in 2008 Canadian dollars; CL= Confidence limit, SD=Standard deviation
85
FIGURE XII: DISTRIBUTION OF COI VALUES FROM MONTE-CARLO SIMULATION
The distribution of COI values is negatively skewed. Using the formula for Pearson’s
Index of skewness yields a value greater than -1.0, suggesting that the data set is not
significantly skewed and that the mean and standard deviation are valid measures of
central tendency and variability. [195]
3*(mean-median)/standard deviation = 3* ($31,213-$33,855)/$9,100 = -0.87
BREAKDOWN COST ANALYSIS
The cost of absenteeism accounted for 96% of the mean COI. Pharmacotherapy
contributed 1.8% to the lifetime COI, while physician fees contributed 1.5%. Hospital
costs and dispensing fees contributed a total of 0.5% to the expected COI.
Monte Carlo Simulation atPatient Seeking Treatment
Pro
babi
lity
$1K $4K $7K $10K $13K $16K $19K $22K $25K $28K $31K $34K $37K $40K $43K $46K $49K
0.0800.0750.0700.0650.0600.0550.0500.0450.0400.0350.0300.0250.0200.0150.0100.0050.000
86
EPIDEMIOLOGIC PARAMETERS
The mean age at which patients entered the model was 31 years. Eighty-nine percent of
the patients who entered the model died from all causes, thereby exiting the model.
Approximately 11% of the cohort remained in the model at some point during their
expectant lifespan upon completion of the simulation. Just over 85% of patients stopped
treatment by the 4th cycle (2nd year).9 Over the course of the model, a mean of 53% of
patients relapsed; with an average rate of 0.79 relapses per patient. Approximately 19%
of patients did not respond to any pharmaco-therapeutic agent. The mean length of time
during which patients were not managed for GAD was 14(SD=9) years. The mean rates
of epidemiologic parameters are presented in Table 11.
DEBUGGING
The ‘debugging’ process provides internal validation of a decision model.[42] Extreme
values were input into the model parameters to determine whether they would produce
expected results. For clinical variables, values of 0 to 1 were input for each parameter.
As expected, response and remission, and spontaneous remission rates varied inversely
with resulting COI values. Furthermore, intolerance, relapse and discontinued treatment
parameters varied in the same direction as the resulting COI. For economic variables,
value inputs from 0 to twice the upper limit were input into the model. As expected, an
increase in cost for each economic parameter resulted in an increase in COI.
9 SEE FIGURE VI
87
FIGURE XIII: HEALTH STATE CUMULATIVE TRANSITION PROBABILITIES GRAPH
Markov Probability Analysis
Stage
Pro
babi
lity
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64
1.000.950.900.850.800.750.700.650.600.550.500.450.400.350.300.250.200.150.100.050.00
Family Physician Initial AssessmenSpecial ist Assessment 2nd LineSpecial ist Assessment 3rd l ineMaintenance 1st l ine Maintenance Benzo onlyMaintenance 1st Line + BenzoMaintenance 3rd LineTreatment stoppedDeath
88
TABLE 12:EPIDEMIOLOGIC PARAMETERS OF STUDY COHORT
Statistic Hospitalization
rate
Relapse per
patient Relapse
rate
Treatment discontinuation
rate
Treatment resistant
rate
Years of discontinued
treatment Mean 0.01 0.79 0.53 0.98 0.19 14.41 SD 0.10 0.92 0.50 0.13 0.39 8.74 Minimum 0.00 0.00 0.00 0.00 0.00 0.00 2.5%CL 0.00 0.00 0.00 1.00 0.00 1.00 10.0%CL 0.00 0.00 0.00 1.00 0.00 3.00 Median 0.00 1.00 1.00 1.00 0.00 14.00 90.0%CL 0.00 2.00 1.00 1.00 1.00 27.00 97.5%CL 0.00 3.00 1.00 1.00 1.00 30.50 Maximum 1.00 7.00 1.00 1.00 1.00 31.50
CL= Confidence limit, SD=Standard deviation
SENSITIVITY ANALYSIS One way sensitivity results for cost variables are presented in Figure VII. The COI was
most sensitive to the cost of absenteeism. The range of uncertainty for the cost of
absenteeism was $1,465-$2,720 and exerted the highest impact compared to all other
variables, varying the COI between $20,500-37,500. The mean COI result was robust
against reasonable variations in all other cost variables, with mean COI values
maintained at approximately $31,000. The range of uncertainty for clinical variables was
narrower overall ($30,000-$32,000) compared to economic variables ($20,500-37,500).
Spontaneous remission exerted the most uncertainty, varying the mean COI by
approximately $2,000, with the remaining variables varying the mean COI by $200 or
less. One way sensitivity results for clinical variables are presented in Figure VIII.
89
FIGURE XIV: Sensitivity analysis of cost variables
Tornado Diagram atPatient Seeking Treatment
Expected Value
$20K $21K $22K $23K $24K $25K $26K $27K $28K $29K $30K $31K $32K $33K $34K $35K $36K $37K $38K
Cost to society: 6 month cost =2*6*8*21.08: 1365 to 25c_Maintenance_dailycost_1st_Line: 1.12 to 2.07Daily hospital cost: 1305 to 2424c_FP_Follow_up: 21.25 to 39.46A196: 57.72 to 107.19c_maintenance_dailycost_3rdLine: 2.04 to 3.79A195: 113.74 to 211.22c_initial_FP_assessment: 42.53 to 78.98c_Maintenance_dailycost_1stL ine_benzo: 0.96 to 1.7c_dispensing_fee: 4.9 to 9.1c_ti trate_max_1st_l ine: 1.19 to 2.20c_Initial_dailycost_1stLine_Benzo: 0.80 to 1.49c_Initial_dailycost_1st_Line: 0.7 to 1.30c_max_3rd_line: 4.04 to 7.50c_Maintenance_dailycost_benzo: 0.16 to 0.30c_initial_dailycost_3rdline: 0.83 to 1.55c_Initial_dailycost_benzo: 0.10 to 0.19c_ti trate_1st_line_benzo: 1.41 to 2.63c_ti trate_benzo: 0.16 to 0.30
90
FIGURE XV: SENSITIVITY ANALYSIS OF CLINICAL VARIABLES
Tornado Diagram atPatient Seeking Treatment
Expected Value
$30.5K $31.0K $31.5K $32.0K
p_spontaneous_remission: 0.2 to 0.25p_benzo monotherapy:0 to 1
p_remission 1st Line: 0.352 to 0.441p_treatmentdisc_benzo: .363 to .6613p_treatmentdisc_3rd_line: 0.363 to 0.66134p_treatmentdisc_1stl ine_Benzo: 0.363 to 0.6613p_response_3rd_line: 0.397 to 0.702p_intolerance_1stl ine: 0.047 to 0.143p_remission_3rd_line: 0.231 to 0.482p_treatmentdisc_1stl ine: 0.7022 to 0.7266Response 1st l ine: 0.64086014 to 0.71357141p_response_benzo: 0.464 to 0.672p_response_1stl ine_benzo: 0.602 to 0.684
91
SUB ANALYSIS The addition of 10 session of CBT therapy resulted in minimal impact on the COI.
Adding 10 sessions of CBT to the 1st line assessment health state added 0.7% to the
lifetime COI. If CBT were to be added during the 2nd line or 3rd assessment health state,
the additional cost increase by 0.3% and 0.01% respectively. Results of the sub-
analysis are presented in Table 13.
TABLE 13: SUB-ANALYSIS OF COST OF ILLNESS WITH 10 SESSION OF CBT THERAPY
Statistic Reference Case CBT with 1st
Line CBT with 2nd
Line CBT with 3rd
Line
Expected value $31,199 $31,429 $31,285 $31,225 Percent difference with Reference Case + 0.7% + 0.3% +0.01%
Mean $31,213 $31,383 $31,318 $31,352
SD $9,100 $9,119 $9,114 $8,991
2.5% CL $5,250 $5,334 $5,356 $5,344
Median $33,855 $34,004 $33,894 $33,862
97.5%CL $42,095 $42,284 $42,373 $42,218 * Values are in 2008 Canadian dollars CBT=Cognitive behavioral therapy (10 sessions during assessment health states),CL=Confidence limit, SD= Standard deviation
92
META-ANALYSIS
CPA LITERATURE INCLUDED IN META-ANALYSIS
A total of 50 articles were cited as evidence for managing GAD by the CPA and
identified for this study; 17 articles were cited as evidence supporting 1st line agents
directly from the CPA guidelines. In addition, nineteen studies were identified from meta-
analytic evidence (i.e., original studies found in reference list) of 1st line treatment;
bringing the number of articles considered for 1st line agents to 36. Subsequently, 22
studies were excluded; 4 were pooled analyses,[138-141] 14 were duplicate
studies(cited in 1 or more meta-analysis and/or the guidelines), 3 were non-retrievable or
had non-extractable data,[142-144] and 1 study was not published. As a result, 14
studies were included for 1st line agents.[145-158]
Twenty-six studies supporting 2nd line agents were cited in the CPA guidelines, a further
41 studies were extracted from meta-analytic evidence. From the 67 studies
considered, 40 were excluded; 2 were pooled analysis,[139,140] 31 were duplicate
studies, and 7 were either non-retrievable or had non-extractable data[159-165], a
further 7 did not report evidence for benzodiazepine agents.
[150,157,175,181,182,189,190] Subsequently, 20 studies were included for evidence in
support of benzodiazepine. [166-174,176-180,183-188]
All 6 studies cited by the CPA for 3rd line agents were included in the meta-analysis.
[177,178.191-194] The literature search tree showing the disposition of all articles
93
identified is presented in Figure IX. A list of studies included in the meta-analysis for
each treatment line is presented in Tables 14-16.
REMISSION RATES
The meta-analytic probability of a remission in patients with GAD treated with 1st line
therapy was 39.7% [CI95% 35.2-44.1%]. There was no pooled remission result with
benzodiazepine therapy due to lack of data reported in the included studies; however,
the pooled analyses of 1st line treatment remission rates with the bezodiazepine arm
resulted in a meta-analytic probability of a remission rate of 41.0% [CI95% 36.3-45.6%] for
the adjunctive treatment line. The meta-analytic probability of remission for 3rd line
therapy was 35.7% [CI95% 23.1-48.2%]. The meta-analytic probability for 1st line therapy
was based on 12 study arms and 1502 patients. Only one study arm with 116 subjects
reported a remission rate for benzodiazepine agents. Thirteen study arms and 1618
patients were considered for pooled analysis of the adjunctive treatment line. The meta-
analytic probability of remission with 3rd line treatment was based on 2 study arms with a
total of 56 patients. Confidence intervals of meta-analytic rates of remission, response
and reduction in HAM-A scores for SNRI, benzodiazepine and 3rd line treatment options
overlap. A summary of response and remission rates by drug and treatment line is
presented in Table 16.
RESPONSE RATES
The meta-analytic probability of a response in patients with GAD treated with 1st line
therapy was 67.7% [CI95% 64.1-71.4%], 56.8% [CI95% 46.4-67.2%] for the
benozodiazepine monotherapy group, and 64.3%[CI95% 60.2-68.4%] 1st line adjunctive
94
* Does not include evidence for NOT recommending propranolol.
CPA evidence supporting recommendations of pharmacotherapy management for Generalized Anxiety Disorder N= 50*
CPA evidence supporting 1st line agents N= 17
CPA evidence supporting 2nd line agents N= 26
CPA evidence supporting 3rd line agents N= 6
Studies cited in meta-analysis or pooled analysis of evidence for 1st line agents
N= 19
Studies cited in meta-analysis or pooled analysis of evidence for 2nd line agents
N= 41
Studies cited in meta-analysis or pooled analysis of evidence for 3rd line agents
N= 0
Number of studies included for 1st line agents N = 36
Number of studies included for 2nd line agents N=67
Number of studies included for 3rd line agents N= 6
Studies excluded: Pooled analysis (4) duplicate (14), not
published (1), non-extractable/non-retrievable (3)
N= 22
Studies excluded: Pooled analysis (3) duplicate (30)
non-extractable/non-retrievable (7) agent other than benzodiazepine (7)
N=40
Studies excluded: N= 0
Total number of articles included in meta-analysis for 1st line outcomes:
N= 14
Total number of articles included in meta-analysis for benzodiazepine outcomes:
N=20
Total number of articles included in meta-analysis for 3rd line outcomes:
N= 6
FIGURE XVI: Literature search results for a meta-analysis of clinical outcomes for GAD.
95
TABLE 14: STUDIES INCLUDED AS 1ST LINE EVIDENCE AND RELATED HAM-A SCORES
Drug Drug First author, year HAM-A Baseline HAM-A Endpoint Mean change in scores [95% CI] class (Fixed dose mg ) Mean SD N Mean SD N
SNRI Venlafaxine Allgulander, 2001 (150) 26.3 4.2 131 9.9 1.6 106 -16.4 [-17.2, -15.6] Allgulander, 2001 (37.5) 26.6 4.3 138 12.8 2.1 102 -13.8 [-14.6, -13.0] Allgulander, 2001 (75) 26.3 4.2 130 10.8 1.7 101 -15.5 [-16.3, -14.7] Davidson, 1999 (150) 23.0 4.0 87 13.8 2.2 73 -9.2 [-10.2, -8.2] Davidson, 1999 (75) 23.7 4.1 87 13.0 2.1 80 -10.7 [-11.7, -9.7] Gelenberg, 2000 25.0 5.0 115 11.6 1.9 60 -13.4 [-14.4, -12.4] Nimatoudis, 2004 27.1 4.8 24 7.9 1.3 24 -19.2 [-21.2, -17.2] Rickels, 2000 (150) 24.5 4.1 81 12.1 2.0 52 -12.46 [-13.4, -11.3] Rickels, 2000 (225) 23.6 3.7 86 12.1 1.9 58 -11.5 [-12.4, -10.6] Rickels, 2000 (75) 24.7 4.4 86 13.5 2.2 60 -11.2 [-12.3, -10.1] Silverstone, 2001 13.2 2.1 32 25.7 4.1 32 -13.2[-14.7, -11.7] OVERALL SNRI 997 748 -13.3 [-14.9, -11.7] SSRI Escitalopram Baldwin, 2006 (5) 27.1 4.5 134 11.6 1.9 116 -15.5 [-16.3, -14.6] Baldwin, 2006 (10) 26.0 4.1 136 9.2 1.5 118 -16.8 [-17.5, -16.0] Baldwin, 2006 (20) 27.7 4.9 133 11.4 1.8 111 -16.4 [-17.2, -15.4] Bielski, 2005 23.7 3.9 60 8.4 1.4 39 -15.3 [-16.4, -14.2] Davidson, 2004 23.6 4.6 154 12.3 2.0 116 -11.3 [-12.1, -10.4] Paroxetine Baldwin, 2006 20.8 2.3 25 8.88 1.43 20 -11.9 [-13.0, -10.8] Ball, 2005 27.3 4.2 139 12.59 2.02 113 -14.7 [-15.5, -13.9] Bielski, 2005 23.4 3.1 61 10.1 1.62 33 -13.3 [-14.2, -12.4] Pollack, 2001 24.2 3.8 161 12.2 1.96 127 -12.0 [-12.6, -11.3] Rickels, 2003 (20) 24.1 3.6 189 11.6 1.86 143 -12.5 [-13.1, -11.9] Rickels, 2003 (40) 23.8 3.4 197 11.6 1.86 143 -12.2 [-12.8, -11.6] Rocca, 1997 26.7 2.6 30 11.1 3.3 25 -15.6 [-17.2, -14.0] Sertraline Allgulander, 2004 24.6 4.6 182 12.9 2.07 147 -11.7 [-12.4, -10.9] Ball, 2005 21.4 3.4 28 9.44 1.52 23 -12.0 [-13.5, -10.6] OVERALL SSRI 1629 1274 -13.6 [-14.6, -12.6] OVERALL 1st Line 2626 2022 -13.5 [-14.3, -12.6] Heterogeneity: Tau² = 4.24; Chi² = 526.48, df = 24 (P < 0.00001); I² = 95%; Test for overall effect: Z = 31.65 (P < 0.00001) SNRI= serotonin norepinephrine reuptake inhibitors SSRI= selective serotonin reuptake inhibitors
96
TABLE 15:STUDIES INCLUDED AS SECOND LINE EVIDENCE (BENZODIAZEPINE MONOTHERAPY) AND RELATED HAM-A SCORES HAM-A scores reported in studies First author, year
(Fixed dose mg ) Baseline Endpoint Mean Reduction
[95% CI] Drug Mean SD N Mean SD N Alprazolam Castillo,1987 26.1 4.3 31 15.8 2.6 32 -10.3 [-12.1, -8.6] Cohn,1984 28.5 4.7 80 18.9 3.1 46 -9.6 [-11.0, -8.2] Enkelmann,1991 28.0 4.6 32 12.5 2.1 28 -15.5 [-17.3, -13.7] Lydiard,1997 24.1 4.0 63 11.1 1.8 45 -13.0[-14.1, -11.9] Moller, 2001 27.9 7.6 102 12.6 7.5 98 -15.3 [-17.4, -13.2] Rickels, 2005 24.9 3.8 88 14.0 2.3 68 -10.9 [-11.9, -9.9] Bromazepam Fontaine,1983 24.1 4.0 16 8.8 1.5 15 -15.3[-17.4, -13.2] Fontaine,1986 28.8 4.8 20 14.8 2.5 17 -14.0 [-16.4, -11.6] Llorca, 2002 25.3 3.4 116 12.3 2.0 99 -13.0 [-13.7, -12.3] Diazepam Boyer,1993 24.4 4.0 55 12.3 2.0 47 -12.1 [-13.3, -10.9] Fontaine,1983 27.4 4.5 16 15.3 2.5 16 -12.1 [-14.7, -9.6] Golberg,1982 25.2 4.2 18 9.2 1.5 18 -16.0 [-18.0, -13.9] Pourmotabbed,1986 25.4 3.1 10 9.8 4.4 10 -15.6 [-18.9, -12.3] Power,1989 18.7 3.7 22 10.5 7.0 17 -8.20 [-11.87, -4.53] Rickels,1993 26.0 4.3 56 13.5 2.2 49 -12.5 [-13.8, -11.2] Rickels,1997 28.5 5.5 67 12.0 2.0 44 -16.5 [-17.9, -15.1] Rickels,2000 24.0 4.0 104 13.0 2.2 80 -11.0 [-11.9, -10.1] Lorazepam Cohn,1984 28.5 4.7 80 20.5 3.4 33 -8.0 [-9.5, -6.4] Cutler,1993 28.3 4.7 20 15.8 2.6 19 -12.5 [-14.9, -10.1] Feltner, 2003 24.9 4.1 104 13.2 2.2 86 -11.7 [-12.6, -10.8] Fontaine,1986 24.7 3.7 68 13.1 2.2 36 -11.6 [-12.7, -10.5] Fresquet, 2000 21.5 3.2 30 15.3 2.5 29 -6.2 [-7.7, -4.7] Laakman,1998 27.5 4.6 57 12.0 2.0 47 -15.5 [-16.8, -14.2] Overall Benzodiazepine 1255 979 -12.2 [-12.4, -11.9]
Heterogeneity: Chi² = 245.24, df = 22 (P < 0.00001); I² = 92%; Test for overall effect: Z = 84.76 (P < 0.00001) CI=Confidence Interval; HAM-A= Hamilton Anxiety Scale; N=Number of subjects; SD=Standard deviation; TCA=Tricyclic Antidepressant
97
TABLE 16:STUDIES INCLUDED AS THIRD LINE EVIDENCE AND RELATED HAM-A SCORES. HAM-A-A scores reported in studies Baseline Endpoint Mean reduction Author, year Mean SD N Mean SD N [CI95%] Adjunctive therapy
Olanzapine + Fluoxetine Pollack, 2006 17.4 6.5 12 10.4 6.4 9 -7.0 [-12.6, -1.4]
Risperidone + Anxiolytics Brawman, 2005 22.1 3.8 19 12.3 2.6 19 -9.8 [-11.9, -7.7]
Gambi, 2005 26.4 6.0 44 9.6 5.8 39 -16.8 [-19.3, -14.3] SSRI Citalopram Varia, 2002 22.2 4.7 13 6.2 3.2 13 -16.0 [-19.1, -12.9] Antihistamine Hydroxyzine Lader,1998 26.6 4.3 81 14.5 3.1 70 -12.1 [-13.3, -10.9] Llorca, 2002 25.5 3.6 105 13.3 2.9 88 -12.2 [-13.1, -11.3] Overall 3rd line 274 238 -12.3 [-13.0, -11.7]
Heterogeneity: Chi² = 26.74, df = 5 (P <0.0001); I² = 81% Test for overall effect: Z = 37.64 (P < 0.00001) CI=Confidence Interval; HAM-A= Hamilton Anxiety Scale; n=number of subjects; SD=Standard deviation; SSRI= selective serotonin reuptake inhibitors
98
TABLE 17: META-ANALYTIC RESPONSE AND REMISSION RATES BY TREATMENT LINE
RESPONSE REMISSION ARMS N PMA CI95%LL CI95%UL ARMS N PMA CI95%LL CI95%UL
1st Line Paroxetine 6 772 0.645 0.611 0.679 5 711 0.349 0.308 0.390 Escitalopram 5 617 0.718 0.643 0.793 4 557 0.437 0.377 0.497 Sertraline 2 210 0.643 0.578 0.708 2 210 0.388 0.215 0.561 Venlafaxine 7 712 0.684 0.601 0.767 1 24 0.625 0.340 0.910 Total 20 2311 0.677 0.641 0.714 12 1502 0.397 0.352 0.441
2nd Line (Monotherapy) Alprazolam 3 253 0.573 0.434 0.712 NA Bromazepam 1 116 0.578 NA NA NA Lorazepam 4 259 0.465 0.407 0.522 1 116 0.535 N/A NA Diazepam 2 147 0.786 0.720 0.852 NA Total 10 775 0.568 0.464 0.672 NA 2nd Line (adjunctive) 1st Line & Benzodiazepine
30 3086 0.643 0.602 0.684 13 1618 0.410 0.363 0.456
Total 3rd line 4 242 0.549 0.397 0.702 2 56 0.357 0.231 0.482
N= number of subjects,na=not applicable, PMA= Meta-analytic probability, SE= Standard error, SNRI= serotonin norepinephrine reuptake inhibitors, SSRI= selective serotonin reuptake inhibitors
99
with benzodiazepine and 54.9% [CI95% 39.7-70.2%] for 3rd line treatment. Response rates
were based on 20 study arms and 2311 patients treated with 1st line treatment, 10 study
arms and 775 patients treated with benzodiazepine monotherapy therapy, 30 arms and
3086 patients with 1st line adjunctive therapy, and 4 study arms and 242 patients treated
with 3rd line therapy. Details appear in Table 16.
MEAN CHANGES IN HAM-A SCORES
The weighted average HAM-A score at baseline for 1st line treatment study arms was
24.4 (SD=0.8), 25.2 (SD=0.9) for the benzodiazepine study arms and 22.1 (SD=1.8) for
the 3rd line study arms. The mean difference from baseline HAM-A scores were: 1st line
-13.5 (CI95%-14.3,-12.6), benzodiazepine, -12.16 (CI95%-12.44, -11.87), 3rd line: -12.3
(CI95% -13.0, -11.7). The mean difference in HAM-A scores were similar for the SSRI
drug class [-13.64 (CI95% -14.65, -12.62)], and venlafaxine [-13.28 (CI95%-14.9, - 11.7].
A summary of HAM-A scores by treatment line, drug class and drugs is presented in
Table 18.
ASSESSMENT OF HETEROGENEITY
Between study heterogeneity of outcomes is likely for each of the three treatment lines
given the highly significant p-values for chi-squares, and tau-squared and I-squared
values (1st line evidence: Tau² = 4.24; Chi² = 526.48, df = 24 (P < 0.00001); I² = 95%;
benzodiazepines: Chi² = 245.24, df = 22 (P < 0.00001); I² = 92%; 3rd line evidence Chi²
= 26.74, df = 5 (P < 0.0001); I² = 81%). There is also evidence of heterogeneity within
drug classes and individual drugs with the exception of sertraline [Tau² = 0.00; Chi² =
0.10, df = 1 (P = 0.75); I² = 0%] and bromazepam [Tau² = 0.92; Chi² = 4.47, df = 2 (P =
0.11); I² = 55
100
TABLE 18:SUMMARY HAM-A SCORES BY TREATMENT LINE, DRUG CLASS AND DRUG Study N Mean HAM-A Scores Drug Class Drug Arms Baseline Endpoint Baseline (SD) Difference [CI95%] p-value SSRI Paroxetine 7 802 604 23.9 (1.17) -13.0 [-13.9, -12.21] p< 0.00001 Escitalopram 5 617 500 25.5 (1.95) -15.0 [-17.07, -13.01] p< 0.00001 Sertraline 2 210 170 24.3 (1.40) -11.7 [-12.42, -11.10] p=0.75 SSRI Total 14 1629 1274 24.1 (0.94) -13.6 [-14.65, -12.62] p< 0.00001 SNRI Venlafaxine 11 997 748 24.3 (1.60) -13.28 [-14.89, -11.66] p< 0.00001 Total 1st Line 25 2626 2022 24.4 (0.77) -13.48 [-14.33, -12.63] p< 0.00001
Benzodiazepine Alprazolam 6 396 317 26.2 (1.84) -12.36 [-14.14, -10.58] p< 0.00001 Bromazepam 3 152 131 25.7 (2.29) -13.83 [-15.28, -12.39] p=0.11 Lorazepam 6 359 250 25.2 (1.65) -10.91 [-13.41, -8.42] p< 0.00001 Diazepam 8 348 281 24.2 (1.61) -13.09 [-14.80, -11.38] p< 0.00001 Benzodiazepine TOTAL 23 1255 979 25.2 (0.86) -12.16 [-12.44, -11.87] p< 0.00001 Total 3rd Line 6 274 238 22.1 (1.84) -12.31 [-12.96, -11.67] p< 0.0001
CI= Confidence Interval; N= number of subjects, SNRI= serotonin norepinephrine reuptake inhibitors SSRI= selective serotonin reuptake inhibitors TCA= Tricyclic anti-depressant
101
SECTION 4: DISCUSSION, CONCLUSIONS, RECOMMENDATIONS
SUMMARY OF FINDINGS
The estimated mean lifetime COI per patient with GAD was (2008 CAD) $31,213 (SD=$9,100).
Most (96%) of this estimate was attributed to the cost of absenteeism. The remaining 4% of
costs was attributed to pharmacotherapy, physician fees, dispensing fees and hospital costs.
The average age at which patients were diagnosed with GAD was 31 years. A large percentage
of patients (85%) discontinued treatment by 2nd year of treatment. Approximately 53% of
patients relapsed by the end of the model simulation. The model structure also resulted in
approximately 19% of patients not responding to any pharmaco-therapeutic agent at which point
patients where either hospitalized for 10 days or simply had their treatment stopped. The mean
length of time during which patients discontinued treatment was 14 (SD=9) years. The results
were most sensitive to the cost of absenteeism, which exerted the highest impact compared to
all other variables, varying the COI between $20,500 - $37,500.
STRUCTURE
ANALYTIC FRAMEWORK
This incidence based model reflects the clinical pathway of a lifetime of waxing and waning of
symptoms due to GAD in patients 18-80. Prior to the development of this study, models
reflecting lifetime course of illness related to GAD have not been published. Published COI
studies were cross-sectional or prevalence based in design and estimated the health care
charges over periods of ≤18 months. Full economic evaluations related to GAD have been
designed using summary population data or conventional decision-tree models and were ≤12
months in length. One COI limited the cost to medical expenses, estimating that GAD incurred a
102
marginal mean cost of USD $2,138 (95% CI, $1,641–$2,632) over a one year period in cohort
taken from a health care charges database.[69] Another COI study estimated significantly
higher median medical care charges in patients with GAD than those with no GAD ($2,375 vs.
$1448). The study also predicted health care charges of pain interference with daily activities in
GAD patients were highest for patients with GAD and high pain interference ($42,620) followed
by no GAD high pain interference ($9,601 P<0.0001).[73] However, there is evidence that
anxiety is under-treated and under-diagnosed, limiting the usefulness of either population based
studies or COI studies using healthcare charges. [141-143] In a cross-sectional survey of New
York city’s non-institutionalized adults, 62% of respondents with either depression or anxiety
were not treated for their condition.[141] Reports based in the National Comorbidity Survey
Replication suggest that less than 3% of patients with GAD make use of any health care
services.[143] Furthermore, neither productivity nor absenteeism was incorporated into the
estimation of the cost of GAD in either of the above mentioned COI studies. The use of a
Markov decision analytic framework to estimate the COI of newly diagnosed patients with GAD
allows for the consideration of chronicity as well as for the waxing and waning of symptoms. An
incidence based approach avoids dealing with the issue of under-diagnosis or under-treatment
of patients. Furthermore, the model could easily be adapted to estimate cost-effectiveness by
adding a second branch comparing a novel therapeutic (e.g. duloxetine) to existing treatment
options.
PATIENTS SEEKING TREATMENT
In this study, costs accrued once patients were diagnosed with GAD by a physician in family
practice which may result in a conservative bias to the result. Patients with GAD are not likely
to initially present with anxiety complaints.[8,91] GAD patients are more likely to present
somatic symptoms such as pain, insomnia, cardiac and gastrointestinal symptoms, that may
103
dominate the focus of their initial visit.[19, 20] A reasonable assumption therefore, is that prior
to a primary diagnosis of GAD, patients are referred to one or more specialists for further
assessment of their somatic symptoms.10 This assumption is supported by several studies. A
study by Logue reports that GAD was the primary diagnosis in 50% of patients seeking a
cardiac evaluation;[140] while studies by Walker and Lydiard, who respectively reported that
GAD was the primary diagnosis in 13 and 25% of patients with irritable bowel
syndrome(IBS).[138,139] It should be noted that the Walker and Lydiard studies did not report
on the same type of patient population as did Logue. While Walker and Lydiard reported on
patients who have been diagnosed with IBS, we do not know how many GAD patients in the
Logue studies were eventually diagnosed with a cardiac condition. Furthermore, consistent
rates of GAD patients presenting with pain or insomnia were not found. Given the missing or
inconsistent nature of the data, this study does not account for the resource utilization accrued,
prior to the primary diagnosis of GAD. The exclusion of these costs biases the mean COI
presented in this thesis towards a more conservative estimate.
BASELINE HAM-A SCORES
Patients with GAD proceeded through the model with similar baseline HAM-A scores, derived
from the meta-analysis of CPA evidence.11 Baseline HAM-A scores were similar for 1st and
benzodiazepine monotherapy options while slightly lower mean for the 3rd line [24.4(SD=0.8),
25.2(SD=0.6) and 22.1(SD=1.8) respectively]. Mean differences from baseline to endpoint
HAM-A scores were greater for patients treated with 1st line therapy compared to those treated
with benzodiazepine monotherapy or 3rd line therapy; 1st line: -13.5 (CI95%-14.3,-12.6),
benzodiazepine, -12.16 (CI95%-12.44, -11.87), 3rd line: -12.3 (CI95% -13.0, -11.7).
10 See Table 5 on page 37 11 See Table 17 on page 100
104
TREATMENT ALGORITHM
The model assumed physicians would manage GAD patients as per the pharmacotherapy
algorithm recommended by the CPA guidelines (with deleted 2nd line agents as listed in the
Methods section). Since no studies have reported the treatment algorithm usually carried out for
patients with GAD in Canada, it is unknown which treatment pattern family physicians prescribe
for GAD patients or what access they have to therapists in a Canadian setting. Therefore, a
comparison of the estimated mean COI to one that would reflect ‘real-life’ management of GAD
is not possible. Furthermore, while the CPA provides no recommendation regarding the
combination of psycho- and pharmacotherapy due to lack of evidence, a US survey of over
1800 non-institutionalized adults reported that 23% of patients with current anxiety were treated
with a combination of medication and counseling.[141] A sub-analysis was performed in this
study of CBT counseling in addition to pharmacotherapy in either 1st, 2nd or 3rd line assessments
and determined that combination therapy would add between 0.01-0.07% in Candian dollars to
the base case mean COI.
PROCESS
CLINICAL VARIABLES
Remission and response of treatment options in this study
Meta-analytic remission rates for 1st, 1st line adjunctive with benzodiazepine (2nd line option),
and 3rd line therapy were 39.7% [CI95% 35.2-44.1%], 41.0% [CI95% 36.3-45.6%] and 35.7% [CI95%
23.1-48.2%] respectively. While 2nd line therapy of 1st line adjunctive with benzodiazepine
produced the highest meta-analytic remission rate, all confidence interval rates overlapped
between treatment lines; implying that one line of therapy was not clinically superior to another.
There was no pooled remission result for benzodiazepine as monotherapy (2nd line option).
105
The meta-analytic probability of a response in patients with GAD treated with 1st line therapy
was 67.7% [CI95% 64.1-71.4%], 56.8% [CI95% 46.4-67.2%] for the benozodiazepine monotherapy
group, and 64.3%[CI95% 60.2-68.4%] 1st line adjunctive with benzodiazepine and 54.9% [CI95%
39.7-70.2%] for 3rd line treatment. Response rates for 1st line were clearly higher, and
uncertainty ranges narrower than those for benzodiazepine monotherapy. The 3rd line meta-
analytic effect yielded a wider range of uncertainty, overlapping both the 1st and 2nd line
response rates confidence intervals. Treatment with 1st line recommended agents yields a
higher response and greater reduction in mean HAM-A scores than benzodiazepine
monotherapy or 3rd line treatment.
Treatment discontinued
Over 85% of the patients in this model discontinued treatment by the 2nd year. Treatment may
have been discontinued by patients for any reason including remission or lack of efficacy. As a
reminder, intolerance was accounted for earlier in the decision tree. Increased
persistence/adherence to medication may reduce the likelihood of a return of symptoms,
thereby reducing absenteeism and increasing productivity.
Relapse rate
Relapse rates for this study were derived from a study by Yonkers et al., who reported
cumulative relapse rates over an 8 year follow up period.[110] For this study, the cumulative
probability for both genders was computed by weighting the cumulative probability of each
gender by their respective sample size. Individual yearly rates were then computed. The
relapse rate used in this study was elected assuming that a weighted mean relapse rate from an
8 year follow up of GAD patients could be extrapolated on to a longer time horizon. At the 8-
year follow up by Yonkers, cumulative relapse rates at year 8 were 43% for men and 36% for
106
women.[37] Assuming that the mean age onset in this study was 31 years, and patients exited
at 80, a conservative relapse rate for a patient in this model over a 49 year time horizon is 53%.
In a sensitivity analysis, the COI was not sensitive to the relapse rate; and therefore does not
introduce a substantial bias either way. In an 8-year follow up study, cumulative relapse rates at
year 1 and year 8 were 19% and 43% for men; and 6% and 36% for women.[37]
Spontaneous remission
Spontaneous remission exerted the greatest impact of all clinical variables; varying the COI
from approximately $30,000 to $32,000. The spontaneous remission rate of “around 20-25%”
was taken directly from a consensus statement on GAD from the International Consensus
Group on Depression and Anxiety, from a meeting held in March 2002 in South Africa.[111] The
statement does not define nor describe ‘spontaneous remission’. In this study, the spontaneous
remission rate was used solely in the TREATMENT DISCONTINUED health state. Individuals
enter this state upon discontinuing treatment for any reason. Within this health state, individuals
are then differentiated between those who ‘remit spontaneously’ and those who do not. From
that point, patients either feel the need to seek treatment anew and re-enter the Family
physician health state (1st health state) or they do not feel the need to seek treatment; even
those patients have not ‘spontaneously remitted’. This model assumes that patients
experiencing ‘spontaneous remission’ no longer feel the need to seek treatment and no longer
exhibit any GAD related symptoms. This model also assumes that patients who have not
experienced ‘spontaneous remission’ have experienced a relapse of symptoms and may or may
not seek treatment anew.
Attrition rate
107
Although several studies compare suicidal ideation in a GAD cohort to a co-morbid GAD cohort
and other anxiety disorder, none report raw data of suicide rates attributable to GAD. The age-
specific attrition rate used in this study would be reasonable infer that these rates include
suicide rates from a population that includes all mental disorders. Therefore the net increase of
suicide as a direct result of GAD is hypothesized to be negligible for the purposes of this study.
RESOURCE UTILIZATION AND COSTS
Absenteeism
The estimated COI in this study was sensitive to the cost of absenteeism. The mean number of
impaired days per month upon which the cost of absenteeism was calculated in this study was
2, with an implicit range of uncertainty between 1.4 and 2.6 days per month. The mean number
of impaired days was based on Kessler’s study of two large surveys conducted in the United
States: the National Co-morbidity Survey and the Midlife Development in the United States
Survey. The surveys distinguished between productivity loss while at work and work days lost.
The number of work days lost was assessed by asking respondents how many days in the past
month they were unable to work or carry out their normal daily activities due to their emotions,
nerves, or mental health. Productivity loss was assessed by asking respondents how many
days in the past month, exclusive of work loss days; they cut back on the amount of work
performed due to problems with their emotions, nerves, or mental health. Responses were
aggregated, and the percent of responders were then categorized into 0, 1 to 2, 3 to 5, and 6 or
more work impairment days.[67] The distribution of the percentage of respondents in each
category was bi-modal for data associated with GAD and MDD as seen in Table 18.
108
TABLE 19: PAST MONTH’S WORK IMPAIRMENT ASSOCIATED WITH GAD AND MDD REPORTED BY KESSLER[67] Survey NCS MDUSS Days MDD&GAD GAD MDD&GAD GAD 0 58.2% 78.2% 49.9% 79.7% 1or 2 8.1% 7.9% 9.2% 9.5% 3-5 11.8% 2.6% 8.8% 2.5% 6 or more 21.9% 11.3% 32.1% 8.3%
GAD= Generalized Anxiety Disorder, NCS=National Co-morbidity Survey, MDD=Major Depressive Disorder, MDUSS=Midlife Development in the United States Survey
By using Kessler’s data, this study implicitly incorporated both productivity loss as well as
absenteeism. The cost of absenteeism is considered to be a conservative estimate. First, the
cost related to absenteeism was not accrued for patients over the age of 65. Furthermore, cost
related to absenteeism was only accrued while patients were assessed for treatment and not
while patients were on maintenance regimens or in any other health states. This implies that by
the time patients were in the maintenance therapy states, anxiety symptoms were managed and
did not interfere with daily life or work activities.
The COI was not sensitive to physician or pharmacist fees, or drug costs. The distributions for
physician and pharmacist fees incorporated a wider range of uncertainty (± 30% of the likeliest
value) than those published in the Ontario Schedule of Benefits. Although most CPA
recommended agents were considered in this model, some were excluded. For 2nd line therapy,
buspirone, imipramine, pregabalin, and bupropion were not considered in this study because of
the agents were either; not listed as a benefit in the Ontario drug formulary, had limited clinical
use in Ontario, were cited as having a risk of death from overdose or required limited use
authorization. From a cost perspective, given that the estimated daily cost for these agents is
largely within the range of costs for 2nd line therapy (Table 19) used in this model, it is unlikely
that consideration of these agents would have a significant impact on the mean COI.
109
TABLE 20: DAILY COST OF AGENTS EXCLUDED FROM 2ND LINE THERAPY Agent Daily Dose Daily Cost* Bupropion 150-450mg .40-1.20 Buspirone 20-30mg 1.30-1.96 Imipramine 25-150mg .12-.77 Pregabalin 50-150mg 1.84-2.53†
*MOHLTC:Ontario Drug Formulary†Loblaws Pharmacy 18 Musgrove Ave Toronto, Ontario.
LIMITATIONS
Lack of available sources reporting both clinical and resource utilization data was a significant
limitation to this study. Of note, remission rates for benzodiazepines were scarce primarily due
to the quality of reporting. Discontinuation rates for benzodiazepines were derived from a
persistence rate and none was available for 3rd line treatment. Furthermore, there are no
available data regarding the Canadian utilization patterns of drugs by GAD patients. The
inherent selection bias for the meta-analysis as well as the heterogeneity of pooled results for
remission and response rates were also limitations to the study. The limitations for both clinical
and economic variables contributed to the uncertainty of the mean estimate.
INTOLERANCE
No peer reviewed studies reporting intolerance rates for each treatment line used in this study
have been published. Meta-analysis of intolerance rates to SSRI and SNRI agents in patients
with MDD by Machado et al [101] was used as proxy for intolerance to 1st line treatment in this
model. Therefore, intolerance to SSRI and SNRI agents in patients with MDD was assumed to
be the same for patients with GAD. Despite a meta-analysis reporting the effect size of
110
withdrawal for any reason and withdrawal for lack of efficacy in GAD patients treated with
benzodiazepines, Martin et al [102] did not report a meta-analytic rate of withdrawal due to
intolerance. Therefore, a pooled analysis of drop-out rates which included 14 study arms
(n=928) from CPA evidence for bromzepam (1 arm), lorazepam (5 arms) and diazepam (8
arms) was used as a proxy for intolerance rate for benzodiazepines. Furthermore, no
publication reported intolerance rates to SSRI or SNRI treatment adjunctive with
benzodiazepine agents. Since the intolerance rates for 1st line were higher than those for
benzodiazepine monotherapy, the rates by Machado et at were once again used as proxy for
intolerance to 1st line adjunctive with benzodiazepine treatment. No data regarding the
aggregate intolerance to 3rd line therapy was available. However, given the structure of the
model, GAD patients would need to be intolerant to the 1st and 2nd line treatment options before
entering the 3rd line option. Therefore, the probability of intolerance to the 3rd line as well was
assumed to be too low to be considered significant to the outcome of this model. The thesis
committee concurred with this assumption. The model therefore infers that GAD patients cannot
be intolerant to agents in all three treatment options. In a one way sensitivity analysis, COI was
not sensitive to intolerance rates.
REMISSION AND RESPONSE
Compared to the data available for 1st line treatment, sample sizes used to pool remission rates
for the remaining treatment options were low. Twelve study arms (n=1502) were included for 1st
line pooled remission rates, 1 study arm (n=116) for benzodiazepine monotherapy (therefore
resulting in no pooling of remission results), and 2 study arms (n=56) for 3rd line remission rates.
Given the lack of data available to derive the meta-analytic probability of remission for 2nd
treatment and the small sample size of 3rd line studies, a comparison to the 1st line meta-analytic
remission rate would not be informative. Given that the meta-analytic probability of remission for
venlafaxine was based on 1 study (n=24); additional studies reporting remission rates on GAD
111
patients treated with venlafaxine would have reduced the degree of uncertainty for that outcome
and potentially change the pooled estimate found here. Studies reporting outcomes from 2nd line
therapy for GAD tended not to report remission rates. Furthermore outcomes were generally
presented in figure format, increasing the possibility of incurring data extraction errors.
However, the meta-analytic mean of response rates for all three treatment lines were based on
larger sample sizes. Response rates for 1st line treatment were based on 20 study arms (n=
2311), 10 study arms (n=775) for the benzodiazepine monotherapy therapy, and 4 study arms
(n=242) for 3rd line therapy. In a one way sensitivity analysis, COI was not sensitive to remission
or response rates.
TREATMENT DISCONTINUATION
Discontinuation rate for 1st line therapy was based on a study of over 14,000 patients with a
variety of anxiety and mood disorders; with no disaggregated data for GAD. Furthermore,
results of discontinuation rates were limited to sertraline, paroxetine and citalopram.
Discontinuation of benzodiazepine was based on a survey conducted on the general population
in the United Kingdom. One minus the persistent use of benzodiazepines used for more than 1
year was used as a proxy for discontinuation rate for this treatment line. This produced a
discontinuation rate of 52%. No treatment discontinuation rates were available for either 1st line
adjunctive therapy with benzodiazepine or for 3rd line therapy. The discontinuation rate for
benzodiazepine monotherapy was used as a proxy and a more conservative available estimate.
RESOURCE USE AND COSTS
There were no Canadian data on utilization patterns of drugs included in the treatment lines.
Therefore, a comparison of mean daily costs for medication could not be validated. However,
112
given that COI was not sensitive to drug costs, this lack of data was did not contribute to the
uncertainty of the estimate.
META-ANALYSIS
The studies included in this meta-analysis were sampled from evidence cited by the CPA as
well as original research from meta-analytic evidence guideline for the management of GAD.
The selection of studies was therefore guided by the CPA guideline committee imposing a large
limitation to the extrapolation of the results found here due to the inherent selection bias.
Heterogeneity of effects was a limitation in the meta-analysis. However, response and
remission rates as well as HAM-A scores were combined despite the presence of heterogeneity.
The decision to proceed with the combination of results was based on the context of the study,
and the absence of guidelines related to the presence of heterogeneity. [65,66] One method of
reducing the effect of heterogeneity is to remove studies that differ in patient characteristics or
underlying condition. However, given that this study set out to determine the effect size of each
treatment line recommended by the CPA, within the parameters set out by the thesis committee;
excluding studies with extractable data to achieve statistical homogeneity would not address the
study question. Another method used in an attempt to reduce heterogeneity was to stratify the
results by individual drugs as opposed to treatment line. Heterogeneity was still present and
was possibly due to the combination of individual study arms of each drug; which in turn
combined different patient and disease characteristics to the effect size by the dissolution of the
randomization component (where available) of studies included in the analysis. Furthermore,
the use of the random effects model would account for between study variance and serve to
minimize but does not completely eliminate heterogeneity in pooled estimates.
CONCLUSIONS
113
A model reflecting the course of illness of GAD has been developed. From a societal
perspective, absenteeism exerts a significant impact to the cost of illness of GAD. Lack of
prospective clinical data contributes to the uncertainty of the COI estimate.
Meta-analysis
Since confidence intervals of meta-analytic rates of remission, response and reduction in HAM-
A scores overlap; there is no statistical evidence to support CPA guideline recommendations of
pharmacotherapeutic management for GAD.
ESTIMATED BURDEN OF ILLNESS The estimated burden of illness of GAD in Canada based on this COI ranges between
$397,488,000 and $944,034,000. This estimated range is based on a lifetime prevalence rate
of 2.4%-5.7% and 26 million people over the age of 15 in Canada in 2006.[14-16,145]
Assuming that a patient entering our model with an onset of GAD at 31 years of age and lives
until the age of 80, the mean cost of GAD per year per patient would be $637 [$31,213/(80-31)].
This is a conservative estimate since 11% of the patients in this study outlive the model.
Nevertheless, given these parameters, the burden of illness in Canada would therefore be
between $397,488,000 and $944,034,000 (2008). Comparison of the burden of illness between
studies should be undertaken with extreme caution; given the variability of methods and
variables included in the analysis. However, the economic burden of illness for 1998 published
by Health Canada reports total costs (direct and indirect) for mental disorders at $7.9 billion
dollars (1998). Discounting the 2008 burden of illness at 5% per annum, yields a discounted
estimate between $244,022,000 and $579,553,070. With caution, one may infer that GAD
contributes between 3.1% and 7.3% to the burden of mental health in Canada.
114
RECOMMENDATIONS
Research determining which therapies are most efficacious given patient and disease
characteristics; such as severity of GAD, age of patient or for treatment resistant patients may
improve the estimate of the COI of GAD. Future research could also investigate the aspects
around the appropriateness of setting lines of therapy in GAD using levels of evidence based on
published and unpublished research. Resource utilization of Canadian treatment patterns for
GAD would be useful to estimate COI and to assess whether guidelines are followed.
115
REFERENCES
1. García-Altés A, Ondategui-Parra S, Neumann PJ. Cross national comparison of technology assessment processes. Int J Technol Assess Health Care 2004; 20:300-10.
2. Diagnostic and Statistical Manual of Mental Disorders (4th ed). Washington, DC: American Psychiatric Association; 1994.
3. World Health Organization. The ICD-10 classification of mental and behavioral disorders: Clinical description and diagnostic guidelines. Geneva: WHO; 1993.
4. Jacobi F, Wittchen H-U, Holting C, et al. Prevalence, co-morbidity and correlates of mental disorders in the general population: Results from the German Health Interview and Examination Survey (GHS). Psychol Med 2004;34:597–611.
5. Grant BF, Hasin DS, Stinson FS, et al. Prevalence, correlates, co-morbidity, and comparative disability of DSM-IV generalized anxiety disorder in the USA: Results from the National Epidemiologic Survey on Alcohol and Related conditions. Psychol Med 2005;35:1747–59.
6. Bereza BG, Machado M, Einarson TR. Systematic review and quality assessment of economic evaluations and quality of life studies related to generalized anxiety disorder. Clin Ther 2009;31:1279-1308.
7. Sadock BJ, Sadock VA, eds. Kaplan & Sadock’s comprehensive textbook of psychiatry, 8th ed. New York: Lippincott Williams & Wilkins; 2005.
8. Wittchen H-U, Hoyer J. Generalized anxiety disorder: Nature and course. J Clin Psychiatry 2001;62(Suppl 11):15-9.
9. Rickels K, Rynn MA. What is generalized anxiety disorder? J Clin Psychiatry 2001;62 (Suppl 11): 4-12.
10. Antony MM, Orsillo SM, Roemer L, editors. Practitioner’s guide to empirically-based measures of anxiety. New York: Kluwer Academic Publishers; 2001.
11. Campbell L, Brown T. Generalized anxiety disorder. In: Antony MM, Barlow DH, editors. Handbook of assessment and treatment planning psychological disorders. New York: Guilford Press; 2002.
12. Hales RE, Yudofsky SC, eds. The American psychiatric publishing textbook of clinical psychiatry. 4th ed. Arlington, VA: American Psychiatric Publishing Inc.; 2003.
13. Le Roux H, Gatz M, Wetherall JL. Age at onset of generalized anxiety disorder in older adults. Am J Geriatr Psychiatry 2005; 13:23-30.
14. Kessler RC, Berglund P, Demler O, and others. Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry 2005;62:593–602.
15. Kessler RC, Chiu WT, Demler O, et al. Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry 2005;62:617–27.
16. Offord DR, Boyle MH, Campbell D, et al. One-year prevalence of psychiatric disorder in Ontarians 15 to 64 years of age. Can J Psychiatry 1996;41:559–63.
17. Carter RM, Wittchen H-U, Pfister H, Kessler RC. One year prevalence of subthreshold and threshold DSM-IV generalized anxiety disorder in a nationally representative sample. Depress Anxiety 2001;13:78-88.
18. Noyes R. Jr, Comorbidity in generalized anxiety disorder. Psychiatr Clin North Am 2001; 24:41-55.
116
19. Wittchen H-U. Generalized anxiety disorder: Prevalence, burden, and cost to society. Depress Anxiety 2002; 16:1622-1710.
20. Roy-Byrne P, Wagner A. Primary care perspectives on generalized anxiety disorder. J Clin Psychiatry 2004;65 (Suppl 13) :20-6.
21. Canadian Psychiatric Association. Clinical practice guidelines: Management of anxiety disorders. Can J Psychiatry 2006;51:S1-S92.
22. Baldwin DS, Anderson IM, Nutt DJ, et al. Evidence-based guidelines for the pharmacological treatment of anxiety disorders: recommendations form the British Association for Psychopharmacology. J Psychopharmacol 2005;19:567-96.
23. Anxiety Review Panel. Evans M, Bradwejn J, Dunn L (Eds.). Guidelines for the treatment of anxiety disorders in primary care. Toronto, ON: Queen’s Printer of Ontario, 2000.
24. National Institute for Health and Clinical Excellence Clinical Guideline 22 - Anxiety, December 2004. Available at: http://www.nice.org.uk/CG022quickrefguide [Accessed March 12, 2008].
25. Bandelow B, Zohar J, Hollander E, Kasper S, Möller H-J, WFSBP Task Force on Treatment Guidelines for Anxiety, Obsessive-compulsive and posttraumatic stress disorders. World Federation of Societies of Biological Psychiatry (WFSBP) Guidelines for the pharmacological treatment of anxiety, obsessive-compulsive and posttraumatic stress disorders. World J Biol Psychiatry 2002; 3: 171-99.
26. Chambless DL, Gillis MM. Cognitive therapy of anxiety disorders. J Consult Clin Psychol 1993;61:248–60.
27. Borkovec T, Whisman M. Psychosocial treatment for generalised anxiety disorder. In: Mavissakalian M, Prien R, editors. Long-term treatment for the anxiety disorders. Washington, DC: American Psychiatric Press; 1996: 171–99.
28. Borkovec T, Ruscio A. Psychotherapy for generalized anxiety disorder. J Clin Psychiatry 2001;62 (Suppl 11):37–42.
29. Gould R, Otto M, Pollack M, et al. Cognitive behavioral and pharmacological treatment of generalized anxiety disorder: a preliminary meta-analysis. Behav Ther 1997;28:285–305.
30. Gould R, Safren S, Washington D, Otto M. A meta-analytic review of cognitive-behavioral treatments. In: Heimberg R, Turk C, Mennin D, editors. Generalized anxiety disorder: Advances in research and practice. New York: Guilford Press; 2004.
31. Goodman WK, Bose A, Wang Q. Treatment of generalized anxiety disorder with escitalopram: pooled results from double-blind, placebo-controlled trials. J Affect Disord 2005;87:161–7.
32. Katz I, Reynolds C, Alexopoulos G, Hackett D. Venlafaxine ER as a treatment for generalized anxiety disorder in older adults: pooled analysis of five randomized placebo-controlled clinical trials. J Am Geriatr Soc 2002;50:18–22.
33. Mitte K, Noack P, Steil R, Hautzinger M. A meta-analytic review of the efficacy of drug treatment in generalized anxiety disorder. J Clin Psychopharmacol 2005;25:141–50.
34. Kapczinski F, Lima MS, Souza JS, Schmitt R. Antidepressants for generalized anxiety disorder. Cochrane Database Syst Rev 2003;(2):CD003592.
35. Stocchi F, Nordera G, Jokinen R, and others. Efficacy and tolerability of paroxetine for the long-term treatment of generalized anxiety disorder. J Clin Psychiatry 2003;64:250–8.
36. Yonkers K, Dyck I,WarshawM, KellerM. Factors predicting the clinical course of generalised anxiety disorder. Br J Psychiatry 2000;176:544–9.
117
37. Yonkers K, Bruce S, Dyck I, Keller M. Chronicity, relapse and illness-course of panic disorder, social phobia and generalized anxiety disorder: finding in men and women from 8 years of follow up. Depress Anxiety 2003;17:173-9.
38. Montgomery S, Sheehan D, Meoni P, and others. Characterization of the longitudinal course of improvement in generalized anxiety disorder during long-term treatment with venlafaxine XR. J Psychiatr Res 2002;36:209–17.
39. Florea I, Trap Huusom A, Allgulander C. Escitalopram for relapse prevention in GAD. [Abstract NR513] In: American Psychiatric Association. New Research Abstracts, Annual Meeting of the American Psychiatric Association. Washington, DC: American Psychiatric Association, 2005.
40. Drummond MF, Sculpher MJ, Torrance GW, O’Brien BJ, Stoddart GL. Methods for the evaluation of health care programmes. 3rd ed. Oxford, UK: Oxford University Press; 2005.
41. Philips Z, Bojke L, Sculpher M, Claxton M, Golder S. Good practice guidelines for decision analytic modeling in health technology assessment. Pharmacoeconomics 2006;24:355-71.
42. Weinstein M, O’Brien B, Hornberger J, et al. Principles of good practice for decision analytic modeling in health-care evaluation: Report of the ISPOR Task Force on Good Research Practices—Modeling Studies. Value Health 2003;6:9-17
43. von Neumann J, Morgenstern O. Theory of games and economic behavior. Princeton, NJ: Princeton University Press, 1947.
44. Sackett DL, Bias in analytic research. J Chronic Dis 1979;32:51-63. 45. Elwyn G, Edwards A, Eccles M, Rovner D. Decision analysis in patient care. Lancet
2001;358:571-4. 46. Detsky AS, Naglie IG, Krahn MD. Clinical decision analysis. Ann R Coll Physicians
Surg Can 1994; 3:157-9. 47. Plume S. Choices, chances, values. Ann Thorac Surg 1992;53:373. 48. Naimark D, Krahn K, Naglie G, Redelmeier D, Detsky A. Primer on medical
decision analysis: Part 5 Working with Markov processes. Med Decis Making 1997;17:152-9.
49. Siegel J, Weinstein M, Russell LB, Gold MR. Recommendations for reporting cost-effectiveness analyses. JAMA 1996;276:1339-4.
50. Guest JF, Russ J, Lenox-Smith A. Cost effectiveness of venlafaxine XL compared with diazepam in the treatment of generalized anxiety disorder in the United Kingdom. Eur J Health Econ 2005; 6:136-45.
51. Heuzenroeder L, Donnelly M, Haby MH, et al. Cost effectiveness of psychological and pharmacological interventions for generalized anxiety disorder and panic disorder. Aust N Z J Psychiatry 2004; 38:602-12.
52. Iskedjian M, Walker JH, Bereza BG, LeMelledo JM, Einarson TR. Cost effectiveness of escitalopram for generalized anxiety disorder in Canada. Curr Med Res Opin 2008; 24:1539-48.
53. Issakidis C, Sanderson K, Corry J, Andrews G, Lapsley H. Modeling the population cost-effectiveness of current and evidence-based optimal treatment for anxiety disorders. Psychol Med 2004; 34:19-35.
54. Jorgensen TR, Stein DJ, Despiegel N, Drost PB, Hemels MEH, Baldwin DS. Cost-effectiveness analysis of escitalopram compared with paroxetine in treatment of generalized anxiety disorder in the United Kingdom. Ann Pharmacother 2006; 40:1752-8.
118
55. Ansseau M, Fischler B, Dierick M, Mignon A, Leyman S. Prevalence and impact of generalized anxiety disorder and major depression in primary care in Belgium and Luxemburg: the GADIS study. Eur Psychiatry 2005;20:229-35.
56. Belanger L, Ladouceur R, Morin CM. Generalized anxiety disorder and healthcare use. Can Fam Physician 2005; 51:1362-3.
57. Bland RC, Newman SC, Orn H. Help seeking for psychiatric disorders. Can J Psychiatry 1998; 48:857-8.
58. Boerner RJ, Sommer H, Berger W, Kuhn U, Schmidt U, Mannel M. Kava-Kava extract LI 150 is as effective as opipramol and buspirone in generalized anxiety disorder- An 8-week randomized, double-blind multi-centre clinical trial in 129 out-patients. Phytomedicine 2003; 10 (Suppl 4):38-49.
59. Cramer V, Torgersen S, Kringlen E. Quality of life and anxiety disorders: A population study. J Nerv Ment Dis 2005; 193:196-202.
60. Dahl AA, Ravindran A, Allgulander C, Kutcher SP, Austin S, Burt T. Sertraline in generalized anxiety disorder: efficacy in treating the psychic and somatic anxiety factors. Acta Psychiatr Scand 2005; 111:429-35.
61. Essau CA, Conradt J, Petermann F. Frequency, co-morbidity and psychosocial impairment of anxiety disorders in German adolescents. J Anxiety Disord 2000; 14:263-79.
62. Fournier L, Lesage AD, Toupin J, Cyr M. Telephone surveys as an alternative for estimating prevalence of mental disorders and service utilization: a Montreal catchment area study. Can J Psychiatry 1997; 42:737-43.
63. Grant BF, Hasin DS, Stinson FS, Dawson DA, Ruan WJ, Huang B. Co-occurrence of 12-month mood and anxiety disorders and personality disorders in the US: results from the National Epidemiologic Survey on Alcohol and Related Conditions. J Psychiatr Res 2005; 39:1-9.
64. Henning ER, Turk CL, Mennin DS, Fresco DM, Heimberg RG. Impairment and quality of life in individuals with generalized anxiety disorder. Depress Anxiety 2007; 24:342-9.
65. Katon W, Von Korff M, Lin E, Lipscomb P, Russo J, Wagner E, Polk E. Distressed high utilizers of medical care DSM-III-R diagnoses and treatment needs. Gen Hosp Psychiatry 1990; 12:355-62.
66. Kennedy BL, Schwab JJ. Utilization of medical specialists by anxiety disorder patients. Psychosomatics 1997; 38:109-12.
67. Kessler RC, DuPont RL, Berglund P, Wittchen H-U. Impairment in pure and co-morbid generalized anxiety disorder and major depression at 12 months in two national surveys. Am J Psychiatry 1999; 156:1915-23.
68. Maier W, Gansicke M, Freyberger HJ, Linz M, Heun R, Lecrubier Y. Generalized anxiety disorder (ICD-10) in primary care from a cross-cultural perspective: a valid diagnostic entity? Acta Psychiatr Scand 2000; 101:29-36.
69. Marciniak MD, Lage ML, Dunayevich ED, et al. The cost of treating anxiety: the medical and demographic correlates that impact total medical costs. Depress Anxiety 2005; 21:178-84.
70. Massion AO, Warchaw MG, Keller MB. Quality of life and psychiatric morbidity in panic disorder and generalized anxiety disorder. Am J Psychiatry 1993; 150: 600-7.
71. McCraken SG, deWit H, Uhlenhuth EH, Johanson CE. Preference for diazepam in anxious adults. J Clin Psychopharmacol 1990; 10:190-6.
72. Mennin DS, Heimberg RG, Jack MS. Co-morbid generalized anxiety disorder in primary social phobia: Symptom severity, functional impairment and treatment response. J Anxiety Disord 2000; 14;325-43.
119
73. Olfson M, Gameroff MJ. Generalized anxiety disorder, somatic pain and health care costs. Gen Hosp Psychiatry 2007; 29:310-16.
74. Ormel J, VonKorff M, Bedirhan Ustan T, Pini S, Korten A, Oldehinkel T. Common mental disorders and disability across cultures: Results from the WHO collaborative study on psychological problems in general health care. JAMA 1994; 272:1741-8.
75. Saarni SI, Suvisaari J, Sintonen H, et al. Impact of psychiatric disorders on health-related quality of life: general population survey. Br J Psychiatry 1997; 190:326-32.
76. Schonfeld WH, Verboncoeur CJ, Fifer SK, Lipschultz RC, Lubeck DP, Buesching DP. The functioning and well-being of patients with unrecognized anxiety disorders and major depressive disorder. J Affect Disord 1997; 43:105-19.
77. Souetre E, Lozet H, Cimarosti I, et al. Cost of anxiety disorders; impact of co-morbidity. J Psychosom Res 1994; 38(Suppl 1)151-60.
78. Stanley MA, Beck JG, Novy DM, et al. Cognitive-behavioral treatment of late-life generalized anxiety disorder. J Consult Clin Psychol 2003; 71:309-19.
79. Steffans DC, McQuoid DR. Impact of symptoms of generalized anxiety disorder on the course of late life depression. Am J Geriatr Psychiatry 2005; 13:40-7.
80. Stein MB, Heimberg RG. Well-being and life satisfaction in generalized anxiety disorder: comparison to major depressive disorder in a community sample. J Affect Disord 2004; 79:161-6.
81. Sullivan MD, Edlund MJ, Zhang L, Unutzer J, Wells KB. Association between mental health disorders, problem drug use and regular prescription opioid use. Arch Intern Med 2006;166:2087-93.
82. Wetherall JL, Thorp SR, Patterson TL, Golshan S, Jeste DV, Gatz M. Quality of life in geriatric generalized anxiety disorder: a preliminary investigation. J Psychiatr Res 2004;38:305-12.
83. Wittchen H-U, Carter RM, Pfister H, Montgomery SA, Kessler RC. Disabilities and quality of life in pure and co-morbid generalized anxiety disorder and major depression in a national survey. Int Clin Psychopharmacol 2000; 15:319-28.
84. Wittchen H-U, Zhao S, Kessler RC, Eaton WW. DSM-III-R generalized anxiety disorder in the national co-morbidity survey. Arch Gen Psychiatry 1994; 51:355-64.
85. Sintonen H. The 15D Measure of health related quality of life: Reliability, validity and sensitivity of its health state descriptive system. Melbourne, Australia: Centre for Health Program Evaluation, Monash University; 1994.
86. EuroQol Group. EuroQol- a new facility for the measurement of health related quality of life. Health Policy 1990; 16: 199-208.
87. Fisher P. Psychopathology of generalized anxiety disorder. Psychiatry 2007;6: 171-5.
88. Drummond MF, Iglesias CP, Cooper NJ. Systematic reviews and economic evaluations conducted for the National Institute for Health and Clinical Excellence in the United Kingdom: a game of two halves? Int J Technol Assess Health Care 2008; 24:146-50.
89. Konnopka A, Leichsenring F, Leibing E, König HH. Cost-of-illness studies and cost-effectiveness analyses in anxiety disorders: a systematic review. J Affect Disord 2009;114:14-31.
90. Hoffman DL, Dukes EM, Wittchen HU. Human and economic burden of generalized anxiety disorder. Depress Anxiety 2008;25:72-90.
91. Ustun TB, Sartorius N (editors). Mental illness in general health care: an international study. Chichester: John Wiley 1995;173:18–23.
92. Schulberg HC, McClelland M, Coulehan JL, Block M, Werner G. Psychiatric decisionmaking in family practice. Gen Hosp Psychiatry 1986;8:1-6.
120
93. Konijn P. Pharmaceuticals products comparative price levels in 33 European countries in 2005. Available at: http://epp.eurostat.ec.europa.eu/cache/ITY_OFFPUB/KS-SF-07-045/EN/KS-SF-07-045-EN.pdf. [Accessed May 20, 2009].
94. Ruger PJ, Emmons KM. Economic evaluations of smoking cessation and relapse prevention programs for pregnant women: A systematic review. Value Health 2008;11:180-90.
95. Government of Canada. Human resources and skills development Canada web site accessed (Life expectancy). Available at: http://www4.hrsdc.gc.ca/[email protected]?iid=3#M_1 [Accessed: date March 17, 2009]
96. Statistics Canada. Complete life table, Canada, 2000 to 2002: (All cause mortality) males; http://www.statcan.gc.ca/pub/84-537-x/t/txttables/cam.txt; females; http://www.statcan.gc.ca/cgi-bin/af-fdr.cgi?l=eng&loc=/pub/84-537-x/t/txttables/caf.txt [Accessed: date March 17, 2009]
97. Statistics Canada. (Gender proportion) http://www12.statcan.ca/english/census06/data/highlights/agesex/pages/Page.cfm?Lang=E&Geo=PR&Code=01&Table=1&Data=Count&Sex=1&StartRec=1&Sort=2&Display=Page [Accessed March 17, 2009].
98. Borkovec T, Ruscio A. Psychotherapy for generalized anxiety disorder. J Clin Psychiatry 2001;62 (Suppl 11):37–42; discussion 3–5.
99. Gould R, Otto M, Pollack M, et al. Cognitive behavioral and pharmacological treatment of generalized anxiety disorder: a preliminary meta-analysis. Behav Ther 1997;28:285–305.
100. Chambless DL, Gillis MM. Cognitive therapy of anxiety disorders. J Consult Clin Psychol 1993;61:248–60.
101. Machado M, Iskedjian M, Ruiz I, Einarson TR. Remission, dropouts, and adverse drug reaction rates in major depressive disorder: a meta-analysis of head-to-head trials. Curr Med Res Opin 2006;22:1825–37.
102. Martin JLR, Sainz-Pardo M, Furukawa TA, Martin-Sanchez E, Seone T, Galan C. Benzodiazepines in generalized anxiety disorder: heterogeneity of outcomes based on a systematic review and meta-analysis of clinical trials. Psychopharm 2007;21:774-82.
103. Einarson TR. Pharmacoeconomic applications of meta-analysis for single groups using antifungal onychomycosis lacquers as an example. Clin Ther 1997;19:559-69.
104. Bereza BG, Machado M, Einarson TR. Meta-analysis of clinical outcomes from guideline recommended pharmacotherapies for generalized anxiety disorder. Presented at the 2nd Latin American Conference of the International Society for Pharmacoeconomic Outcomes and Research (ISPOR), Rio de Janeiro, Brazil, Sept 10-12, 2009.
105. Methods for meta-analysis in medical research. Sutton AJ, Abrams KR, Jones DR, Sheldon TA, Song F, editors. Hoboken, NJ: John Wiley & Sons Ltd Publishers 2004; 3: 53.
106. Mullins CD, Shaya FT, Fanlun M, Wang J, Harrison D. Persistence, switching, and discontinuation rates among patients receiving eertraline, paroxetine, and citalopram. Pharmacother 2005;25:660-7.
107. Ohayon MM, Caulet M, Priest RG, Guilleminault C. Psychotropic medication consumption patterns in the UK general population. J Clin Epimdemiol 1998;51:273-83.
121
108. Wilson EB. Probable inference, the law of succession, and statistical inference. J Am Stat Assoc 1927: 22: 209-12.
109. Vasile RG, Bruce SE, Goisman RM, Pagano M, Keller MB. Results of a naturalistic longitudinal study of benzodiazepine and SSRI use in the treatment of generalized anxiety disorder and social phobia. Depress Anxiety 2005;22:59-67.
110. Yonkers KA, Bruce SE, Dyck IR, Keller MB. Chronicity, relapse and illness course of panic disorder, social phobia, and generalized anxiety disorder: Findings in men and women form 8 years of follow up. Depress Anxiety 2003;17:173–9.
111. Ballenger J, Davidson J, Lecrubier Y, Nutt D, Borkovec T, Rickels K, Stein D, Wittchen H-U. Consensus statement on generalized anxiety disorder from the international consensus group on depression and anxiety. J Clin Psychiatry 2001;62:53-8.
112. Wang P, Lane M, Olfson M, Pincus H, Wells K, Kessler R. Twelve month use of mental health services in the United States. Arch Gen Psychiatry 2005;62:629-40.
113. Health Canada. A Report on Mental Illnesses in Canada. Ottawa, Canada 2002 Available at: http://www.hc-sc.gc.ca/pphb-dgspsp/publicat/miic-mmac/index.html http://www.phac-aspc.gc.ca/publicat/miic-mmac/pdf/chap_4_e.pdf [Accessed April 18th, 2009].
114. Product Monograph: Alprazolam, ProDoc Lté, Laval, Quebec, Prepared December 16th, 2003.
115. Product Monograph: Diazepam, Hoffmann-La Roche Ltd. Mississauga, Ontario. October 10th, 2007.
116. Prescribing Information: Lorazepam, Wyeth Canada. St Laurent Quebec, March 24, 2003.
117. Product Monograph: Apo-Paroxetine, Apotex Inc. Weston, Ontario. June 6th, 2007.
118. Product Monograph: Co-Sertraline, Cobalt Pharmaceuticals Inc. Mississauga, Ontario. October 23rd, 2006.
119. Product Monograph: Escitalopram. Lundbeck Canada Inc. Montreal, Quebec. August 28th, 2008.
120. Product Monograph: Venlafaxine. Wyeth Canada. Montreal, Quebec. March 11th, 2009.
121. Product Monograph: Bromazepam, Hoffmann-La Roche Ltd. Mississauga, Ontario. August 24th, 2007.
122. Varia I, Rauscher F. Treatment of generalized anxiety disorder with citalopram. Int Clin Psychopharmacol 2002;17:103–7.
123. Pollack MH, Simon NM, Zalta AK, and others. Olanzapine augmentation of fluoxetine for refractory generalized anxiety disorder: a placebo controlled study. Biol Psychiatry 2006;59:211–5.
124. Gambi F, De Berardis D, Campanella D, and others. Mirtazapine treatment of generalized anxiety disorder: a fixed dose, open label study. J Psychopharmacol 2005;19:483–7.
125. Brawman-Mintzer O. Adjunctive risperidone in generalized anxiety disorder: A double-blind, placebo-controlled study. J Clin Psychiatry 2005;66:1321–5.
126. Rickels K, Downing R, Schweizer E, et al. Antidepressants for the treatment of generalized anxiety disorder: a placebo-controlled comparison of imipramine, trazodone, and diazepam. Arch Gen Psychiatry 1993;50:884–95.
122
127. Government of Ontario. Minister of health and long term care web site. Available at HTTP://WWW.HEALTH.GOV.ON.CA/ENGLISH/PROVIDERS/PROGRAM/OHIP/SOB/PHYSSERV/PHYSSERV_MN.HTML [Accessed April 18th, 2009].
128. Government of Ontario. Minister of health and long term care web site. Available at https://www.healthinfo.moh.gov.on.ca/formulary/ [Accessed April 18th, 2009].
129. Government of Ontario. Minister of health and long term care web site. Available at HTTP://WWW.HEALTH.GOV.ON.CA/ENGLISH/PUBLIC/PUB/DRUGS/DISPENSE.HTML [Accessed April 18th, 2009].
130. Estimation of Total Acute Care Inpatient Costs, With Average Cost per Stay (Unit Cost , 2004-2005 Run Date:09-04-12 Ref: HS3.1 © 1996-2007, Canadian Institute for Health Information (CIHI). Available at: HTTP://DSP-PSD.PWGSC.GC.CA/COLLECTION_2008/CIHI-ICIS/H115-39-2005E.PDF [Accessed March 13th, 2009] HTTP://APPS.CIHI.CA/MICROSTRATEGY/ASP/MAIN.ASPX?EVT=5005&SRC=MAIN.ASPX.RWD.CIHI Patient cost estimator – ontario 2007-8
131. DeMartinis N, Rynn M, Rickels K, Mandos L. Prior benzodiazepine use and buspirone response in the treatment of generalized anxiety disorder. J Clin Psychiatry 2000;61:91–4.
132. Statistics Canada. Wage rate: Available at: HTTP://WWW40.STATCAN.GC.CA/L01/CST01/LABR69A-ENG.HTM [Accessed April 18th, 2009].
133. Giacobbe P, McIntyre R, Goldstein B, Kennedy S, Flint A. Rates and patterns of treatment seeking by individuals with mood and anxiety disorders. Psychiatr Serv 2008;59:1099.
134. Krahn M, Naglie G, Naimark D, Redelmeier D, Detsky A. Primer on medical decision analysis: Part 4-Analysing the model and interpreting the results. Med Decis Making 1997;17:142-51.
135. López-Ibor R, Generalized anxiety disorder: A 40 year follow up study. Acta Psychiatr Scand 2007;115:372-9.
136. Cochran W. The combination of estimates from different experiments. Biometrics. 1954;10:101–124.
137. Statistics Canada. Consumer Price Index: Available at: HTTP://CANSIM2.STATCAN.GC.CA/CGIWIN/CNSMCGI.PGM?LANG=E&SP_ACTION=RESULT&SP_ID=3956&SP_TYP=5&SP_SORT=1&SP_PORTAL=2.
138. Walker EA, Roy-Byrne PP, Katon WJ, et al. Psychiatric illness and irritable bowel syndrome: a comparison with inflammatory bowel disease. Am J Psychiatry 1990;147:1656–6
139. Lydiard RB, Fossey MD, Marsh W, et al. Prevalence of psychiatric disorders in patients with irritable bowel syndrome. Psychosomatics 1993;34:229–234
140. Logue MB, Thomas AM, Barbee JG, et al. Generalized anxiety disorder patients seek evaluation for cardiological symptoms at the same frequency as patients with panic disorder. J Psychiatr Res 1993;27:55–59
141. Gwynn RC, McQuistion HL, McVeigh KH, Garg RK, Frieden TR, Thorpe LE. Prevalence, diagnosis, and treatment of depression and generalized anxiety disorder in a diverse urban community. Psychiatr Serv 2008;6:641-7
142. Schulberg HC, McClelland M, Coulehan JL, Block M, Werner G. Psychiatric decision making in family practice future research directions Gen Hosp Psychiatry 1986;8:1-6
123
143. Wang PS, Lane M, Olfson M, Pincus HA, Wells KB, Kessler RC. Twelve-month use of mental health services in the United States: Results from the National Comorbidity Survey Replication. Arch Gen Psychiatry. 2005;62:629-40
144. 2006 census Available at Statistics Canada Website:http://www12.statcan.ca/census-recensement/2006/dp-pd/hlt/97 51/pages Accessed March 21st, 2010.
145. Economic burden of illness 1998. Available at Health Canada website: HTTP://WWW.PHAC-ASPC.GC.CA/PUBLICAT/EBIC-FEMC98/PDF/EBIC1998.PDF Accessed March 21st, 2010.
146. HTAGuidelines for the economic evaluation of health technologies: Canada [3rd Edition]. Ottawa:Canadian Agency for Drugs and Technologies in Health; 2006.
124
LIST OF PUBLICATIONS AND ABSTRACTS In preparation or submitted Bereza BG, Machado M, Papadimitropoulos M, Sproule B, Ravindran AV, Einarson TR. The cost of illness of generalized anxiety disorder (GAD) in Canada: A Markov approach. (In preparation) Bereza BG, Machado M, Ravindran AV, Einarson TR. Evidence based review of clinical outcomes of guideline recommended pharmacotherapies for generalized anxiety disorder: a meta-analysis (In preparation) Longo CJ, Bereza BG. An analysis of monthly “out-of-pocket” costs for patients with breast cancer compared to common cancers in Ontario, Canada (Submitted) Accepted or published Bereza BG, Machado M, Einarson TR. Systematic review and quality assessment of economic evaluations and quality of life studies related to generalized anxiety disorder. Clinical Therapeutics 2009; 31:1279-1308 Iskedjian M, Desjardins O, Piwko C, Bereza B, Jaszewski B, Einarson TR. Willingness-to-pay for a treatment for pain in multiple sclerosis. PharmacoEconomics 2009; 27:149-58
Bereza BG, Machado M, Einarson TR. Assessing the reporting and scientific quality of meta-analyses of randomized controlled trials of treatments for anxiety disorders. Annals of Pharmacotherapy 2008; 42:1402-9
Iskedjian M, Walker JH, Bereza BG, Danchenko N, Le Melledo J-M, Einarson TR. Cost effectiveness of escitalopram in the treatment of generalized anxiety disorder. Current Medical Research & Opinion 2008; 24: 1539-48
Piwko C, Desjardins OB, Bereza BG, Machado M, Jaszewski B, Freedman MS, Einarson TR, Iskedjian M. Pain due to multiple sclerosis: analysis of the prevalence and economic burden in Canada. Pain Research & Management. 2007; 12:259-65 Iskedjian M, Bereza B, Gordon A, Piwko C. Einarson TR. Meta-analysis of cannabis based treatments for neuropathic and multiple sclerosis-related pain. Current Medical Research & Opinion. 2007; 23:17-24 Peer-reviewed – Abstracts (Poster presentations) Bereza BG, Machado M, Papadimitropoulos M, Sproule B, Ravindran AV, Einarson TR.
125
A dynamic Markov approach assessing the cost of illness of generalized anxiety disorder (GAD) in Canada. International Society for Pharmacoeconomic Outcomes and Research (ISPOR), 12th Annual European Congress Paris 2009 Bereza BG, Machado M, Einarson TR. Meta-analysis of clinical outcomes from guideline recommended pharmacotherapies for generalized anxiety disorder. International Society for Pharmacoeconomic Outcomes and Research (ISPOR), 2nd Latin American Conference Brazil 2009 Machado M, Bereza BG, Gonzalez C, Einarson TR. Pharmacoeconomic contribution of Latin America to ISPOR conferences. International Society for Pharmacoeconomic Outcomes and Research (ISPOR), 2nd Latin American Conference Brazil 2009 Piwko C, Vicente C, Bereza B, Ventin A. Economic burden of severe chronic hand eczema/dermatitis in Canadian adults. International Society for Pharmacoeconomic Outcomes and Research (ISPOR) 14th Annual International Meeting, Orlando, USA 2009 Bereza BG, Machado M, Einarson TR. Economic evaluations of generalized anxiety disorder: A systematic review. The Society for Medical Decision Making. Philadelphia, USA 2008 Bereza BG, Machado M, Einarson TR. Assessing the reporting and scientific quality of meta-analyses of randomized controlled trials of treatments for anxiety disorders. International Society for Pharmacoeconomic Outcomes and Research, 13th Annual International Meeting, Toronto, Canada 2008 Desjardins O, Bereza B, Jaszewski B, Iskjedjian M, Einarson TR, Piwko C. Economic Burden of Pain to Multiple Sclerosis in Canada. Canadian Association of Physical Medicine & Rehabilitation Annual Scientific Meeting, Vancouver, Canada, 2006 Desjardins O, Bereza B, Jaszewski B, Malmberg C, Iskedjian M, Einarson TR, Piwko C. The prevalence and management of pain due to Multiple Sclerosis in Canada. Consortium of Multiple Sclerosis Centres Vancouver, Canada, 2006. Piwko C, Desjardins O, Bereza B, Jaszewski B, Malmberg C, Iskedjian M, Einarson TR. Pain due to Multiple Sclerosis: Analysis of the economic burden in Canada. International Society for Pharmacoeconomic Outcomes and Research (ISPOR), 11th Annual International Meeting, Philadelphia, PA United States, 2006. Iskedjian M, Bereza B, Desjardins O, Jaszewski B, Piwko C, Einarson TR. Converting the scores of a clinical instrument for measuring pain to a preference based one: Report Value in Health. 9(3):A6, May/June 2006. Walker J, Iskedjian M, Bereza BG, Hemels M, Einarson TR, Cost Effectiveness of Escitalopram in the Treatment of Generalized Anxiety Disorder; International Society for Pharmacoeconomic Outcomes and Research(ISPOR), Florence, Italy Fall 2005. Desjardins O, Piwko C, Bereza B, Gordon A, Devonshire V, Freedman M, Ursell M, Vicente C, Iskedjian M, Jaszewski B, Einarson TR, Gafni A. Development and pre-testing of a willingness to pay instrument for a treatment for pain in Multiple Sclerosis. Presented at the Canadian
126
Association for Health Services and Policy Research Conference, Montreal, QC, Canada, September 2005.