adverse cardiac events among older … · ii 1 abstract adverse cardiac events among older patients...
TRANSCRIPT
ADVERSE CARDIAC EVENTS AMONG OLDER ONTARIO PATIENTS ON VENLAFAXINE: A POPULATION-BASED STUDY
By
Joanne Man-Wai Ho
A thesis submitted in conformity with the requirements for the degree of Masters of
Science, the Institute for Health Policy, Management and Evaluation, University of
Toronto
©Copyright by Joanne Man-Wai Ho (2015)
ii
1 ABSTRACT
Adverse Cardiac Events Among Older Patients on Venlafaxine: A Population-Based Study
Masters of Science 2015 Joanne Man-Wai Ho
Institute of Health Policy, Management and Evaluation University of Toronto
Background: It is unknown whether the common antidepressant, venlafaxine, is
associated with increased cardiovascular risk.
Objective: To examine the cardiac safety of venlafaxine in older individuals.
Methods: A population-based retrospective cohort study (Ontario, Canada) compared
patients aged ≥66 years who commenced treatment with venlafaxine, sertraline or
citalopram between 2000 and 2009. The primary outcome was a composite of death or
hospitalization for myocardial infarction or heart failure within the first year of therapy.
Results: Following inverse probability of treatment weighting with the propensity score,
we found no difference in the primary outcome with venlafaxine relative to sertraline
(hazard ratio (HR) 0.97; 95% confidence interval (CI) 0.93 to 1.02). When compared to
citalopram, however, an increased risk was observed (HR 1.39; 95% CI 1.19 to 1.62).
Conclusions: As compared with sertraline, venlafaxine was not associated with an
increased risk of adverse cardiac events. The increased risk observed in the venlafaxine
–citalopram comparison warrants further study.
iii
2 ACKNOWLEDGEMENTS
I would like to thank and acknowledge the following people:
I would like to thank my supervisor, Dr. David Juurlink, for all his guidance,
support and patience. In particular, I would like to thank him for allowing me to explore
my own research questions (to find that “fire in the belly”) and to pitch them during
“Research Dragon’s Den.” Those countless hours spent appraising my research
questions taught me the importance of developing meaningful research projects with
rigorous methodology. I also thank him for his thoughtful feedback during the writing
process and for challenging me to communicate clearly and concisely.
I would like to thank my committee. I thank Dr. Peter Austin for always being
available to discuss any methodological detail, Dr. Muhammad Mamdani for helping me
with project design and troubleshooting; and Dr. Sharon Straus for her tremendous
support and guidance in all aspects of research life.
Many thanks to Ms. Tara Gomes. I sincerely appreciate her abundance of
teaching, patience and kindness during this entire process.
I would like to thank Mr. Ashif Kachra, Ms. Karen Hood and Ms. Karen Arbour for
administrative support and Ms. Chelsea Hellings for project coordination.
Thank you to the ICES Systems department, particularly Mr. Jesse Stamplecoski
and Mr. Jackson Wong for their patience.
Thank you to Minnie for being the most supportive Sister. You were always
available to answer questions about epidemiology, or to provide me with precious desk
space in your office or nourishment.
iv
Finally, thank you to my Husband, Michael, for his endless patience, support and
encouragement, and to my Daughters, Emily and Matilda, both born during this thesis.
v
Table of Contents
1 ABSTRACT ........................................................................................................................... ii
2 ACKNOWLEDGEMENTS ................................................................................................ iii
3 LIST OF TABLES AND FIGURES ................................................................................. vii 3.1 List of Tables .............................................................................................................................. vii 3.2 List of Figures ............................................................................................................................. vii 3.3 List of Appendices ................................................................................................................... viii
4 BACKGROUND AND RATIONALE ................................................................................. 1 4.1. Background .................................................................................................................................. 1
4.1.1 Depression: Definition and Epidemiology ................................................................................ 1 4.1.2 Depression: Management ................................................................................................................ 2 4.1.3 Generalizability of Clinical Trial Results to the Real-World Patient .............................. 9 4.1.4 Observational Studies .................................................................................................................... 11
4.2 Venlafaxine ................................................................................................................................. 12 4.2.1 Clinical Use .......................................................................................................................................... 12 4.2.2 Pharmacokinetics and Pharmacodynamics .......................................................................... 13 4.2.3 Venlafaxine and the Cardiovascular System ......................................................................... 16
4.3 Research Question ................................................................................................................... 21 4.4 Hypothesis .................................................................................................................................. 21
5 METHODS .............................................................................................................................. 22 5.1 Study Design and Setting ....................................................................................................... 22
5.1.1 Data Sources ....................................................................................................................................... 22 5.2 Cohort Definition ...................................................................................................................... 26
5.2.1 Study and Control Groups ............................................................................................................ 26 5.2.2. Inclusion Criteria ............................................................................................................................. 26 5.2.3 Exclusion Criteria ............................................................................................................................. 26
5.3 Outcome Definition .................................................................................................................. 27 5.3.1 Outcome Validation Studies......................................................................................................... 27
5.4 Covariates .................................................................................................................................... 28 5.4.1 Variables Associated with Drug Exposure or Outcome ................................................... 29
5.5 Statistical Analysis ................................................................................................................... 32 5.5.1 Baseline Characteristics ................................................................................................................ 32 5.5.2 Propensity Score .............................................................................................................................. 32 5.5.3 Survival Analysis Time-to-Event Analysis ............................................................................. 35 5.5.4 Regression-based analysis ........................................................................................................... 36
5.6 Ethics ............................................................................................................................................. 36
6 RESULTS ................................................................................................................................ 38 6.1 Description of study and control cohorts ........................................................................ 38
6.1.1 Venlafaxine-Sertraline Comparison ......................................................................................... 38 6.1.2 Venlafaxine-Citalopram Comparison ....................................................................................... 39
6.2 Primary Analysis ...................................................................................................................... 40 6.2.1 Venlafaxine-Sertraline Comparison ......................................................................................... 40 6.2.2 Venlafaxine-Citalopram Comparison ....................................................................................... 41
vi
6.3 Secondary Analysis .................................................................................................................. 41 6.3.1 Stratified Analysis ............................................................................................................................ 41 6.3.2 Time-Varying Analysis of Dose: Venlafaxine-Sertraline Comparison ........................ 42 6.3.3 Time-Varying Analysis of Dose: Venlafaxine-Citalopram Comparison ..................... 42
7 DISCUSSION .......................................................................................................................... 43 7.1 Major Findings ........................................................................................................................... 43
7.1.1 Consistency with Existing Literature ....................................................................................... 44 7.1.2 Limitations .......................................................................................................................................... 46 7.1.3 Implications for Clinical Practice ............................................................................................... 48
7.2 Contributions ............................................................................................................................. 49 7.3 Future Directions ..................................................................................................................... 49
8 REFERENCES......................................................................................................................... 51 9 TABLES AND FIGURES ................................................................................................................. 64 9.1 TABLES ......................................................................................................................................... 64
9.1.1 Table 1a: Baseline Characteristics of Older Patients on Venlafaxine Compared to Sertraline ........................................................................................................................................................ 64 9.1.2 Table 1b: Baseline Characteristics of Older Patients on Venlafaxine Compared to Citalopram ...................................................................................................................................................... 66 9.1.3 Table 2a: Adverse Cardiac Events in Older Patients on Venlafaxine Compared to Sertraline (N=90,114) ............................................................................................................................... 68 9.1.4 Table 2b: Adverse Cardiac Events in Older Patients on Venlafaxine Compared to Citalopram (N=52,785)............................................................................................................................. 68 9.1.5 Table 3: Exploratory Time-Dependent Dose Analysis of Older Patients on Venlafaxine Compared to Sertraline ................................................................................................... 69
9.2 FIGURES........................................................................................................................................ 70 9.2.1 Figure 1: Boxplot Distribution of Inverse Probability of Treatment Weights for Venlafaxine to Sertraline and Venlafaxine to Citalopram Comparisons .............................. 70 9.2.2 Figure 2a: Crude and Weighted Survival Curves for Adverse Cardiac Events among Individuals on Venlafaxine Compared to Sertraline ...................................................... 71 9.2.3 Figure 2b: Crude and Weighted Survival Curves for Adverse Cardiac Events among Individuals on Venlafaxine Compared to Citalopram ................................................... 72 9.2.4 Figure 3a: Adverse Cardiac Events in Older Patients without Cardiovascular Disease on Venlafaxine Compared to Sertraline (N=43,451) ................................................... 73 9.2.5 Figure 3b: Adverse Cardiac Events in Older Patients with Cardiovascular Disease on Venlafaxine Compared to Sertraline (N=46,663) .................................................................... 73 9.2.6 Figure 4a: Adverse Cardiac Events in Older Patients without Cardiovascular Disease on Venlafaxine Compared to Citalopram (N=26,786)................................................. 74 9.2.7 Figure 4b: Adverse Cardiac Events of Older Patients with Cardiovascular Disease on Venlafaxine Compared to Citalopram (N=25,999) ................................................................. 74
10 APPENDICES ...................................................................................................................... 75 10.1.1 APPENDIX 1a: Aggregated Diagnosis Groups of Older Individuals on Venlafaxine Compared to Sertraline ............................................................................................................................ 75 10.1.2 APPENDIX 1b: Aggregated Diagnosis Groups of Older Individuals on Venlafaxine Compared to Citalopram .......................................................................................................................... 78
vii
3 LIST OF TABLES AND FIGURES
3.1 List of Tables Table 1a: Baseline Characteristics of Older Patients on Venlafaxine Compared to Sertraline Table 1b: Baseline Characteristics of Older Patients on Venlafaxine Compared to Citalopram Table 2a: Adverse Cardiac Events of Older Patients on Venlafaxine Compared to Sertraline Table 2b: Adverse Cardiac Events of Older Patients on Venlafaxine Compared to Citalopram Table 3: Exploratory Time-Dependent Dose Analysis of Older Patients on Venlafaxine Compared to Sertraline
3.2 List of Figures Figure 1: Boxplot Distribution of Inverse Probability of Treatment Weights for Venlafaxine to Sertraline and Venlafaxine to Citalopram Comparisons Figure 2a: Crude and Weighted Survival Curves for Adverse Cardiac Events among Individuals on Venlafaxine Compared to Sertraline Figure 2b: Crude and Weighted Survival Curves for Adverse Cardiac Events among Individuals on Venlafaxine Compared to Citalopram Figure 3a: Adverse Cardiac Events of Older Patients Without Cardiovascular Disease on Venlafaxine Compared to Sertraline Figure 3b: Adverse Cardiac Events of Older Patients With Cardiovascular Disease on Venlafaxine Compared to Sertraline Figure 4a: Adverse Cardiac Events of Older Patients Without Cardiovascular Disease on Venlafaxine Compared to Citalopram Figure 4b: Adverse Cardiac Events of Older Patients With Cardiovascular Disease on Venlafaxine Compared to Citalopram
viii
3.3 List of Appendices Appendix 1a: Aggregated Diagnosis Groups of Older Patients on Venlafaxine Compared to Sertraline Appendix 1b: Aggregated Diagnosis Groups of Older Patients on Venlafaxine Compared to Citalopram
1
4 BACKGROUND AND RATIONALE
4.1. Background
4.1.1 Depression: Definition and Epidemiology
Depression, or Major Depressive Disorder, is defined by the Diagnostic and
Statistical Manual of Mental Disorders, Fifth Edition(1) as at least two weeks of
depressed mood or markedly diminished pleasure that results in a change from previous
functioning. This is also accompanied by at least four of the following: changes in
appetite or sleep, decreased energy, interest or concentration, feelings of guilt, or
psychomotor agitation or retardation.(1) Depression is common, with a lifetime
prevalence of 12-16% in North America.(2) Among patients with cardiovascular disease,
the prevalence may be as high as 30%.(3-5) Independently associated with morbidity
and mortality,(6,7) impaired quality of life and occupational function, depression is one
of the main causes of disability worldwide.(2,8-10) By affecting adherence to medication
or a healthy lifestyle, depression also impairs a patient’s ability to manage their other
chronic diseases.(11,12) A cross-sectional study of diabetic patients with depressive
symptoms from 2 primary care clinics in Washington state, United States, found an
association between depressive symptom severity and health care utilization.(11)
Health care costs of patients with moderate and severe depression were 51% to 85%
higher compared to those with low severity depression.(11) The National Health Service
in the United Kingdom found that the direct costs of treating depression (£887 million)
2
exceeded the combined cost of treating hypertension (£439 million) and diabetes (£300
million).(13) Furthermore, the indirect costs of missing work due to depression are
substantial to the patient and general economy.(13) As a result, depression impacts
millions of people worldwide on individual and societal levels.
4.1.2 Depression: Management Depression can be treated with nonpharmacological or pharmacological
therapy.(14)
4.1.2.1 Depression: Nonpharmacological Therapy
Nonpharmacological therapy includes neurostimulation and psychotherapy and
can be administered alone or in combination with pharmacologic therapy.
Neurostimulation involves the delivery of either electric or magnetic
interventions to the brain. It includes electroconvulsive therapy (ECT), repetitive
transcranial magnetic stimulation, vagus nerve stimulation and deep brain
stimulation.(15) The Canadian Network for Mood and Anxiety Treatment recommends
ECT as first line therapy for patients with depression and concomitant psychosis or acute
suicidal ideation; or treatment-resistant depression.(15) ECT is effective with response
rates of 50-90% among adults(16,17) and is generally regarded as safe. However, the
risk of adverse cardiac or cognitive events is higher among patients with pre-existing
cardiovascular disease and advanced age (≥65 years) respectively.(18-20)
Psychotherapy requires the establishment of a patient-therapist relationship and
can be administered in individual and group settings, alone and in combination with
3
medication.(14) Psychotherapy (specifically, interpersonal therapy and cognitive
behavioural therapy) for major depressive disorder, has been shown to have equivalent
efficacy to pharmacotherapy and is not associated with adverse drug effects.(14,21) Its
ability to prevent recurrence alone compared to antidepressant therapy however, is
unclear due to heterogeneous results from randomized clinical trials (RCTs).(22-25)
These RCTs often compared psychotherapy with the selective serotonin reuptake
inhibitors (SSRIs), fluoxetine or paroxetine.(22-25) Psychotherapy was equivalent to
pharmacotherapy in preventing recurrence of depression in RCTs that included younger
adults and had experienced practitioners administering psychotherapy.(22-25) Because
psychotherapy alone has a longer time lag effect,(21) it is not recommended in
depressed patients with severe suicidality or psychotic symptoms.(14) Furthermore,
patients often do not seek psychotherapy due to financial cost and depression-
associated “psychological barriers.”(26)
4.1.2.2 Depression: Pharmacological Therapy
Although nonpharmacological treatments are available, medication remains a
common element of therapy for many patients. With a prescription rate exceeding one
prescription for every 10 persons in the United States per year between 2005 and 2008,
antidepressants are used by tens of millions of people every day.(27,28) A population-
based time series analysis of older Ontario adults found an increased prevalence of
antidepressant use from 5.5% in 1993 to 10.9% in 2002.(29) Antidepressants for older
Ontario individuals accounted for 42% of mental health disease medication costs which
totaled $149.4 million in 2002.(29)
4
Second Generation Antidepressants
Among the various drug treatment options for depression, second generation
antidepressants are the most widely prescribed in North America.(27) These include
selective serotonin reuptake inhibitors (SSRIs), serotonin-noradrenaline reuptake
inhibitors (SNRI), and other drugs such as bupropion, mirtazapine and trazodone. Over
the past decades, antidepressant prescriptions for older individuals shifted away from
first generation antidepressants, such as tricyclic antidepressants and monoamine
oxidase inhibitors, towards second generation antidepressants.(29) Potential benefits
of this shift in practice among older adults include a lower risk of anticholinergic-
mediated adverse events, such as cognitive impairment and falls.(30) This would be
achieved through the avoidance of tricyclic antidepressants which possess
anticholinergic activity. In addition, monoamine oxidase inhibitors further heighten the
increased risk of drug-drug interactions faced by older adults..(29,30) The risks of this
shift include a significant cost increase of at least 61% in drug expenditures.(29)
Efficacy Antidepressant efficacy is conventionally measured with depression rating
scales. Depression rating scales can be used to diagnose depression, quantify severity,
or assess changes over time, including the response to treatment.(31) There are many
validated scales used in research, but the two most commonly used are the observer-
rated Hamilton Depression Rating Scale (HDRS) (31,32) and Montgomery Åsberg
Depression Rating Scale (MADRS).(31,33) The original 17-item HDRS is widely used in
clinical research and practice.(2,31,34) Scores of 0-7, 8-12, 13-17 and ≥18 indicated
5
normal, mild depression, less than major depression and major depression,
respectively.(32) Limitations of this scale include its bias towards somatic, behavioural
and anxiety symptoms, and its lack of generalizability across different populations,
particularly the older population.(31) The Montgomery-Åsberg Depression Rating Scale
(MADRS) is also observer-rated and consists of 10-items.(33) Scores of 0-6, 7-19, 20-34,
and ≥35 indicate no symptoms, and mild, moderate and severe depression respectively.
It was derived from a heterogeneous pool of depressed patients who were enrolled in
RCTs of amitriptyline, clomipramine, mianserin and maprotiline.(31) Compared to the
HDRS, the MADRS may be more sensitive to change and does not measure somatic
symptoms.(33) Conventionally, response to therapy is defined as a score reduction of
≥50%, while remission is achieved when scores fall within the pre-defined normal
ranges. (2,31)
Comparative effectiveness studies have found no difference in efficacy among
medication classes in the treatment of depression although up to 30% of patients do not
respond to these medications.(14) Patients enrolled in explanatory randomized
controlled trials are often different from real-world patients(35-38) and suffer from
more severe depression (HDRS score>20).(37) Therefore, the effectiveness, or benefit
observed in the real-world, of antidepressant therapy among the general population is
unclear.
Risks of Second Generation Antidepressants
Although second generation antidepressants as a class do not possess
anticholinergic effects, which are undesirable for the older population,(39) they may
6
cause other adverse effects.(40-42) With the prevalent use of these medications,(25-
26) it is important for health care professionals to understand the adverse effect profile
and properly inform patients.
Second-generation antidepressants are associated with adverse gastrointestinal,
central nervous system, metabolic and cardiovascular effects, and an increased risk of
fragility fractures.(43,44,45,46) Gastrointestinal treatment-emergent adverse effects
include nausea, vomiting and diarrhea. A range of central nervous system adverse
effects associated with second-generation antidepressants include sedation, headaches,
nervousness and agitation.(43) Unlike other SSRIs, paroxetine possesses anticholinergic
properties, which may manifest as cognitive impairment among older individuals.(43)
Metabolic effects of SSRIs include hyponatremia, and although this generation of
antidepressants is generally ‘weight neutral,” paroxetine, an SSRI, is associated with
weight gain.(43) An increase in upper gastrointestinal bleeding due to the inhibition of
serotonergic-mediated platelet adhesion has been observed among patients on non-
steroidal anti-inflammatory drugs and concomitant antidepressants that affect
serotonin.(43) Sexual dysfunction is common, affecting up to 50% of patients on second
generation antidepressants but may occur more frequently given that it is often
underreported.(43) Patients on SSRIs, particularly fluoxetine and paroxetine, are more
likely than those on SNRIs to experience sexual dysfunction.(43) The risk of fragility
fractures has been found to be increased among patients on second generation
antidepressants.(45) Mechanisms for this adverse event include decreased bone
density, and impaired sleep quality thereby increasing fall risk.(45) Finally, the safety
7
profiles of individual second generation antidepressants may differ, particularly with
regard to cardiovascular safety.(44)
Cardiac Safety of Second Generation Antidepressants
Depression is common among patients with cardiovascular disease.(2-4) Left
untreated, it can affect an individual’s ability to manage their chronic cardiovascular
disease thereby increasing morbidity and mortality.(6,11,12) With the popularity of
pharmacologic therapy,(25,26) it is important to determine whether antidepressants
might exacerbate cardiovascular disease.
Our literature review of the risk of cardiovascular events among adults using
second generation antidepressants found 5 randomized controlled trials, 6
observational studies, 5 case reports and 1 case series of 248 patients with toxic
ingestions of venlafaxine.
The cardiac safety of second generation antidepressants has been the focus of
two clinical trials.(47,48) A randomized controlled trial of 44 outpatients sought to
investigate venlafaxine’s effect on heart rate variability. The authors measured ex vivo
drug levels and noradrenaline receptor activity, and symptoms of depression and
anxiety and demonstrated that patients on venlafaxine exhibited decreased variation in
respiratory sinus arrhythmia during paced breathing compared to those on
paroxetine.(47) The SADHEART study was a randomized, double-blind, placebo-
controlled trial of sertraline in patients with depression who were recently hospitalized
for acute coronary syndrome.(48) The authors did not find a difference in adverse
cardiac events among the sertraline-treated group.(48) A small nonrandomized
8
controlled study comparing paroxetine, sertraline and venlafaxine among patients
suffering from post-traumatic stress disorder found a higher rate of side effects such as
palpitations and subsequent drop-outs among those on venlafaxine.(49) A telephone
survey-based case-control study investigated the risk of myocardial infarction among
individuals on high serotonin transporter affinity antidepressants compared to those on
antidepressants with low and moderate serotonin transporter affinities or no
antidepressants.(50) They found a decreased risk of myocardial infarction among
patients on antidepressants with high serotonin transporter affinity. This attenuated
risk may be due to SSRI-mediated depletion of serotonin stores which may cause
decreased platelet adhesion and aggregation, a component of coronary artery
thrombosis, and increased release of endothelial nitric oxide.(39) This latter effect
causes vasodilation, which is cardioprotective.(39) It should be noted, however, that
this study’s participants were relatively young (mean age 51 years with a standard
deviation of 9 years), had no history of cardiovascular disease and were not necessarily
new users of their antidepressants.(50) Although an observational study by Johnson et
al. found an increase in adverse cardiac effects such as increased heart rate and blood
pressure among older community-dwelling patients with depression on venlafaxine,
there were no cases of myocardial infarction or heart failure. This study, however, had
a small sample size thereby limiting its statistical power to detect differences in such
cardiac events.(51) To date, there have been no population-based studies specifically
comparing the cardiac toxicity of the commonly prescribed antidepressants,
9
venlafaxine, sertraline or citalopram, and providing further insight into the safety of
these drugs among older patients with cardiovascular disease.(52)
With the higher background prevalence of cardiac disease and depression
among older patients, a causal association with a medication would be difficult to
delineate. Nevertheless, given the millions of antidepressants dispensed annually, any
safety differences would be important to establish.(27,28)
4.1.3 Generalizability of Clinical Trial Results to the Real-World Patient
Generalizability, or external validity, is the extent to which study results can be
applied to real-world patients. Clinical trials, particularly randomized controlled trials,
are necessary to establish drug efficacy.(38) While pragmatic randomized controlled
trials attempt to study a drug’s efficacy in the real-world setting, only 10% of
randomized controlled trials are of this nature.(53) The majority are explanatory
randomized controlled trials which maximize internal validity through the careful
selection of trial participants.(35,38,53) In explanatory trials, individuals are subjected
to stringent inclusion and exclusion criteria to minimize confounding variables or
adverse events; or to maximize the effect size given a limited sample size.(37)
Consequently, these explanatory trial patients differ from real-world patients in the
setting where they receive treatment, overall medical profile, and degree of monitoring
from health care professionals.(38,54) Real-world patients tend to be managed in
nonacademic settings, to have advanced age and multi-morbidity; and are often not
monitored as closely.(35)
10
Patients with more comorbidities and increased age are often excluded from
explanatory RCTs but not in real-world practice, and these patients are at increased risk
of adverse drug events.(38,55) Comorbidities may result in altered pharmacodynamics,
pharmacokinetics and increase the risk for drug-drug and drug-disease interactions.(56)
For example, patients with impaired kidney function have a reduced ability to clear
renally-excreted drugs, leading to increased serum drug levels and the risk of adverse
drug events.(56) Furthermore, normal aging is accompanied by a decrease in renal
function thereby predisposing older patients to adverse drug events.(56) Clinical trials
that exclude patients with comorbidities and advanced age cannot provide health care
professionals with comprehensive drug safety information.
The issue of generalizability has been investigated in a number of areas of
medicine(35,54,57) including the treatment of depression. Zimmerman and his
coauthors applied commonly used trial eligibility criteria to psychiatric outpatients
diagnosed with unipolar depression without psychosis.(37) The majority of
antidepressant trials excluded patients with less severe depression on the HDRS (score
≤20; score ≥18 indicates severe/major depression), significant suicidality, recent
substance or ethanol abuse, and a comorbid psychiatric illness. Of the 599 real-world
patients, only 123 (20.5%) would have been included in an antidepressant explanatory
clinical trial.(37)
Finally, the cost associated with randomized controlled trials may hinder the
ability to have large sample sizes or longer observation times. These trials may be
adequately powered to investigate drug efficacy, yet underpowered to detect
11
uncommon but clinically significant adverse drug events, particularly those that develop
after prolonged exposure. Clinicians should be aware of the limitations of available
clinical trials when considering pharmacotherapy in the treatment of depression.
4.1.4 Observational Studies
In observational studies, scientists observe the effect of an exposure on a study
population. Population-based observational studies include real-world patients and
frequently have the added advantage of a large sample size and the potential for a
longer observation window.(58) This provides adequate statistical power to
characterize uncommon, but clinically significant events, and to detect adverse events
which develop with accumulated exposure.(58) This methodology allows one to
investigate the risk of adverse events associated with drugs administered in the real-
world setting. By complementing randomized controlled trials, population-based
studies are powerful tools in post-marketing surveillance of medications.
12
4.2 Venlafaxine
Venlafaxine was first introduced in 1994 by Wyeth® Pharmaceuticals, and since
then has been used by tens of thousands of Ontario residents every year.(29) It is
thought to alleviate depression by inhibiting serotonin and noradrenaline reuptake and
is in the same SNRI class as duloxetine.(59-62)
4.2.1 Clinical Use
Since its introduction, venlafaxine has been widely used in the treatment of
psychiatric and pain disorders, and hot flashes.(43,63-66) Venlafaxine is effective in the
treatment of major depressive disorder and a recent systematic review found it to have
similar benefits to SSRIs.(44) The Canadian Network for Mood and Anxiety Treatment
(CANMAT) and American Psychiatric Association guidelines both identify venlafaxine as
a first- line agent for the treatment of depression.(66,67) Although the benefit of
venlafaxine in the treatment of bipolar disease, attention deficit hyperactive disorder
and concomitant depression is less clear, it is still used to treat these disorders.
Venlafaxine’s serotonergic and noradrenergic effects have been hypothesized to provide
relief to patients suffering from neuropathic pain, particularly diabetic neuropathy, a
population with a high prevalence of depression.(65,68,69) Finally, venlafaxine has
been used in the symptom management of hot flashes, particularly in women
undergoing breast cancer treatment.(63) Patients with psychiatric disease and
neuropathy are at increased risk for heart disease; and concomitant psychiatric disease
affects chronic disease management, thereby increasing cardiovascular morbidity and
mortality.(4,6,7,70-72) As a result, it would be important to understand venlafaxine’s
13
cardiovascular safety profile since it is frequently prescribed to patients predisposed to
adverse cardiac events.
4.2.2 Pharmacokinetics and Pharmacodynamics
4.2.2.1 Pharmacokinetics
Venlafaxine is available in immediate release (IR) and extended release (ER)
formulations and is administered orally at doses of 37.5 to 375 mg per day. Although
pharmacokinetic studies describe a slower rate of absorption of the ER formulation
compared to the IR formulation, the amount of drug absorbed is equivalent.(40)
Venlafaxine IR and ER preparations are well absorbed with peak plasma drug levels
achieved within 2 and 6 hours respectively, (40,73) but due to significant first pass
metabolism, its bioavailability is moderate at approximately 45%.(74)
Venlafaxine is metabolized in the human liver into 3 metabolites-O-
desmethylvenlafaxine (ODV), N-desmethylvenlafaxine and N,O-desmethylvenlafaxine.
The main metabolite, ODV, is formed by the hepatic enzyme cytochrome p450 (CYP)
2D6 and has similar in vitro serotonergic, noradrenergic and dopaminergic activity as the
parent compound.(40) Patients may have one of four phenotypes of CYP 2D6
metabolizing phenotypes: poor (little or no 2D6 activity), intermediate (between poor
and extensive), extensive (normal activity) and ultrarapid (more than normal). Therefore
patients with poor, intermediate, extensive and ultrarapid CYP2D6 phenotypes have
low, low-normal, normal and high ratios of ODV/venlafaxine respectively.(75) It is
unclear, however, whether these different ratios of ODV/venlafaxine translate to
14
significant clinical outcomes. A pooled analysis of four RCTs found that extensive
metabolizers had greater improvements in their depressive symptoms while on
venlafaxine compared to poor metabolizers.(75) They did not find any difference in the
rate of adverse effects. Whyte et al. examined the effect of 2D6 phenotype among 46
older adults with depression while on venlafaxine.(76) They did not find a difference in
depressive symptoms, or adverse effects (total or cardiac), among those with poor or
intermediate compared to extensive 2D6 phenotypes. They did, however, find a
nonsignificant trend towards adverse events in those with poor 2D6 phenotypes. This
study’s main limitation was insufficient statistical power, particularly since it included
few patients with poor metabolizer phenotype.(76) Venlafaxine’s minor metabolites, N-
desmethylvenlafaxine via cytochrome p450 3A3/4 metabolism and N,O-
desmethylvenlafaxine are inactive.(39,40) CYP3A4 is a minor pathway for venlafaxine’s
metabolism, but it produces an inactive N-desmethylvenlafaxine metabolite.(40)
Pharmacokinetic studies with ketoconazole, a known CYP 3A4 inhibitor, have found
increased serum drug concentrations of venlafaxine among extensive and particularly
poor metabolizers (CYP 2D6).(77)
The principal route of excretion of venlafaxine and its metabolites is through the
kidney.(40) Given the decrease in renal function with normal aging, drug clearance may
be reduced in the geriatric population.(40,78) This may predispose older patients to
increased drug levels rendering them vulnerable to adverse drug effects.
4.2.2.2 Pharmacodynamics
15
Venlafaxine, a bicyclic phenylethylamine derivative, is a racemic mixture of (R)-
and (S)-enantiomers.(40) Both enantiomers inhibit the synaptosomal uptake of
serotonin, but only the (R)-enantiomer inhibits the reuptake of noradrenaline.(39,79)
This noradrenergic effect is more pronounced at higher doses, such as those exceeding
200mg/day,(59,64,65) therefore venlafaxine acts like an SSRI at lower doses.(41,60,62)
Patients who ingest toxic doses of venlafaxine present with signs and symptoms
consistent with excessive serotonin and noradrenergic activity.(80,81) Venlafaxine also
weakly inhibits the uptake of dopamine.(79) In vitro studies have demonstrated
venlafaxine’s ability to block sodium channels, which may increase the risk of cardiac
dysrhythmias associated with QRS prolongation.(82) Venlafaxine does not interact with
muscarinic anticholinergic, alpha-1 adrenergic or histaminergic receptors in vitro.(73)
This lack of anticholinergic activity(39) may make it a favourable agent for older
patients;(40-42) however, these patients may also be vulnerable to its noradrenergic
effects, which are more pronounced at higher doses.(40,41,83-85)
16
4.2.3 Venlafaxine and the Cardiovascular System
In theory, venlafaxine could increase the risk of adverse cardiac events through
two mechanisms. First, it may potentiate cardiac dysrhythmias and, second, through its
noradrenergic effects, it may cause myocardial ischemia or heart failure.
4.2.3.1 Arrhythmia
Medications that block cardiac ion channels may affect conduction and increase
the risk of fatal dysrhythmias. Blockade of sodium and potassium channels may result in
QRS and QT interval prolongation respectively, thereby increasing the risk of fatal
ventricular dysrhythmias.(86) Although venlafaxine has mild sodium channel blocking
properties,(82) QRS widening is rare, even at higher therapeutic doses or following
overdose.(87,88) QT-interval prolongation is also uncommon in the absence of other
risk factors such as hypokalemia, hypomagnesemia or other QT prolonging medications
(e.g. fluoroquinolones).(80,87-89)
Previous Observational Studies
To investigate the risk of dysrhythmias in the real-world setting, Martinez et al.
conducted a nested case-control study investigating the risk of sudden cardiac death or
near death among older patients in those on venlafaxine compared to other
antidepressants.(90) This population-based study used the United Kingdom General
Practice Research Database, a database of electronic medical records of primary care
practices, and did not find an increased risk compared to fluoxetine (odds ratio (OR)
0.66; 95% confidence interval (CI) 0.38 to 1.14), citalopram (OR 0.89; 95% CI 0.50 to
17
1.60) or dosulepin (OR 0.83; 95% CI 0.46 to 1.52).(90) This study matched for age and
adjusted for cardiovascular disease and other factors associated with ventricular
arrhythmia. This study’s primary outcome, however, focused on venlafaxine’s potential
to increase the risk of ventricular arrhythmia through its effect on cardiac sodium
channels or the QT interval. Although ventricular arrhythmia may be associated with
acute myocardial ischemia, this study’s outcome would not identify cases of myocardial
ischemia occurring in the absence of a ventricular arrhythmia, a more common
event.(90) A cohort study using Medicare claims from 1999 to 2003 in five different
states in the United States compared the risk of sudden cardiac death or ventricular
arrhythmia among new users of 21 antidepressants.(91) They found no difference in
sudden cardiac death or ventricular arrhythmia among patients on venlafaxine
compared to paroxetine, their selected reference drug based on its lack of effect on
cardiac conduction.(91) This large observational study accounted for covariates
associated with cardiovascular disease. However, the majority of subjects were young
adults(<10% age ≥75 years of age).(91) Based on this existing literature, venlafaxine’s
independent risk of sudden cardiac death and ventricular arrhythmia does not seem to
be a significant concern. Clinicians, however, should still consider arrhythmias in the
setting of concomitant drugs or diseases that also cause QT or QRS interval
prolongation.
4.2.3.2 Noradrenergic Effects
18
Venlafaxine’s noradrenergic effects have been demonstrated with in vitro
studies at the transporter level.(79) Studies conducted among healthy volunteers have
illustrated venlafaxine’s ability to potentiate noradrenaline through an attenuated blood
pressure response to tyramine and enhanced dorsal vein vasoconstriction in response to
noradrenaline.(41,60,62) Patients who overdose on venlafaxine manifest these
noradrenergic effects with tachycardia, hypertension and in some cases, heart
failure.(80,92)
These noradrenergic effects can also lead to tachycardia and hypertension at
therapeutic doses,(51,83,89,93) and some case reports implicate venlafaxine as a
possible contributor to acute myocardial infarction and heart failure.(94,95) A small
randomized controlled trial (n=52) of long term care patients demonstrated an increase
in adverse events necessitating drug cessation of venlafaxine compared to
sertraline.(96) Although not sufficiently powered to explore cardiac toxicity,
venlafaxine-treated patients exhibited a small increase in heart rate and more cases of
heart failure.(96) Among patients enrolled in clinical trials of venlafaxine, 5.5% on doses
exceeding 200mg/day experienced clinically significant increases in blood pressure
(diastolic blood pressure of ≥15 mm Hg from baseline).(83) Smajkic et al. compared
patients with post-traumatic stress disorder who received paroxetine, sertraline or
venlafaxine and found a higher rate of side effects and subsequent drop-outs among
those on venlafaxine.(49) While this study did not specifically assess cardiac outcomes
nor include older patients, more venlafaxine patients suffered from palpitations, which
may be noradrenergically mediated.(49)
19
Previous Observational Studies
Since exploratory RCT patients are typically younger, healthier and more closely
monitored, real-world patients are likely at higher risk of adverse effects.(35) Results
from observational studies have suggested higher risks of adverse events among older
depressed patients on venlafaxine. In an observational study of older patients
diagnosed with major depressive disorder without psychosis, 28.8% of patients
experienced new-onset cardiovascular symptoms, with 24% of normotensive and 54%
of hypertensive patients at baseline experiencing clinically significant increases in blood
pressure.(51) Coupland et al. conducted an observational study investigating various
adverse events associated with numerous classes of antidepressants. Although
cardiovascular events were included among the many outcomes, they did not examine
this outcome by controlling for specific confounding factors. Finally, rather than
studying venlafaxine separately, it was grouped with several other unrelated
antidepressants, therefore, venlafaxine’s risk of heart failure and ischemic heart disease
is still undefined.(52) A case-control study using telephone surveys investigated the risk
of myocardial infarction among users of high serotonin transporter affinity
antidepressants compared to nonusers or to users of antidepressants with low and
moderate serotonin transporter affinities. They found a decreased risk of myocardial
infarction among patients on antidepressants with high serotonin transporter affinity
which included sertraline; however, their study participants were younger, had no
history of cardiovascular disease and were not necessarily new users of their
antidepressants.(50) Furthermore, instead of performing a head-to-head comparison of
20
sertraline and venlafaxine, this study compared groups of antidepressants with different
serotonin transporter affinities.(50) Venlafaxine was assessed along with other non-
SSRI antidepressants, along with tricyclic antidepressants, trazodone, bupropion and
mirtazapine, which aside from their serotonin transporter affinity properties, are
pharmacologically different.
To date, no study has specifically studied the risk of heart failure and ischemic
heart disease among real-world patients on venlafaxine. In order to characterize this
risk, we conducted a population-based study using the Province of Ontario’s health
administrative databases. We hypothesized that by virtue of its noradrenergic effects,
venlafaxine may be associated with an increased risk of adverse cardiac events among
older patients relative to other antidepressants that do not prevent noradrenaline
reuptake (i.e. SSRIs).
21
4.3 Research Question
Are older patients (aged 66 years or older) started on the serotonin-
noradrenaline reuptake inhibitor venlafaxine at different risk of adverse cardiac events,
defined as hospitalization for acute myocardial infarction or heart failure, or death,
compared to those on the selective serotonin reuptake inhibitors sertraline or
citalopram in the first year of use?
4.4 Hypothesis
We hypothesized that older patients who initiated venlafaxine would be at
increased risk of adverse cardiac events in the first year of use compared to those who
initiated serotonin or citalopram.
22
5 METHODS
5.1 Study Design and Setting
We conducted a population-based retrospective cohort study of Ontario residents aged
66 years or older who commenced treatment with either venlafaxine, sertraline or citalopram
between April 1st 2000 and March 31st 2009.(97) These patients have universal coverage for
hospital care, physician services and prescription medications.
5.1.1 Data Sources
At the Institute for Clinical Evaluative Sciences, we linked health administrative
databases using an encrypted version of each subject’s unique health insurance number.(98,99)
We used the following databases: Ontario Drug Benefit, Canadian Institute for Health
Information’s Discharge Abstract Database and National Ambulatory Care Reporting System,
Ontario Health Insurance Plan and the Registered Persons Database. These administrative
databases are routinely linked to study drug safety.(100,101)
5.1.1.1 Ontario Drug Benefit Program The province of Ontario covers the price of 3,800 prescription medications, and some
natural products and supplies for diabetic testing for individuals eligible for the Ontario Drug
Benefit (ODB) program. Individuals covered by this program fulfill at least one of the following
criteria: all community dwelling individuals at least 65 years of age, individuals residing in a long
term care facility or Home for Special Care, individuals receiving Home Care or on social
assistance; or individuals with medication costs that far exceed their income (Trillium
23
program).(102) The ODB database provides the following information for each covered drug
claim since 1997: date the prescription was filled, cost, quantity, the patient’s long term care
status, and identifiers for the drug, patient, physician and outpatient pharmacy where the
prescription was filled. We used this database’s prescription drug information to define the
exposures, and to identify covariates (medications and as surrogate markers of comorbidities).
5.1.1.2 Canadian Institute for Health Information databases The Canadian Institute for Health Information (CIHI) is an independent nonprofit
organization which maintains data from the Canadian health care system by working with
provincial ministries of health, Statistics Canada and Health Canada. CIHI used diagnosis codes
from the International Classification of Diseases-9 (ICD-9) and ICD-10 prior to 2002, and
following 2002 respectively.(103) Procedures are identified with codes from the Canadian
Classification of Health Interventions (CCI) and Canadian Classification of Diagnostic,
Therapeutic, and Surgical Procedures (CCP) codes. CIHI maintains its various databases by
systematically assessing, documenting and improving data quality. Data abstracted from
patient charts or medical records, namely the Discharge Abstract Database, National
Ambulatory Care Reporting System and the Canadian Organ Replacement System, undergo re-
abstraction studies to ensure data completeness and consistency resulting in increasing
accuracy over time.(103,104)
CIHI Discharge Abstract Database The CIHI Discharge Abstract Database (CIHI-DAD) contains information about every
hospital stay at most acute care hospitals in Canada. It includes data on patient demographics
24
(date of birth, gender, and postal, county and residence codes), clinical data (ICD-9 and ICD-10
diagnostic codes, CCI and CCP procedure codes, physician identifiers), and administrative data
(hospital or institution identifier, admission category, length of stay, disposition). The CIHI-DAD
is regularly validated with re-abstraction studies, most recently for ICD-10 codes in 2009-2010,
and diagnosis code accuracy for all heart diseases exceeded 80%.(103)
CIHI National Ambulatory Care Reporting System The CIHI National Ambulatory Care Reporting System (CIHI-NACRS) provides information
about patient care administered in Same Day Surgery, outpatient clinics such as renal dialysis or
cancer care, and emergency departments. It provides data on patient demographics (date of
birth, gender, postal, county and residence codes), clinical data (ICD-9 and ICD-10 diagnostic
and CCI and CCP procedural codes, physician identifiers), and administrative data (hospital or
institution identifiers, disposition following emergency department visit).
We used both CIHI DAD and NACRS to define our primary outcome, and comorbidities.
Although the Ontario Mental Health Reporting System (OMHRS) and Continuing Care Reporting
System (CCRS) are part of CIHI, these databases were not used. OMHRS contains clinical and
social information about adult patients admitted to mental health beds in the province. Our
study period started in 2000, the earliest year when venlafaxine, sertraline and citalopram were
available through ODB, whereas OMHRS became available through CIHI only in 2005.(103)
CCRS contains clinical information, which includes medical illness, cognition and behaviour,
medication use, nutrition and special procedures, about patients residing in long term care or
complex continuing care facilities. Although this database was available since 1997, it does not
25
provide information about all patients included in our study cohort, specifically those not
receiving care from such facilities.
5.1.1.3 Ontario Health Insurance Plan Database
The Ontario Health Insurance Plan (OHIP) database contains information regarding
claims by health care providers (including physicians and certain laboratories) who provide
services to inpatients and outpatients. It provides data on patient demographics (date of birth,
gender), clinical data (diagnostic code, health care provider identifiers), administrative data
(referring physician, hospital identifier if applicable, service date). Over 95% of outpatient
physician fee-for-service billing by primary care physicians are included in OHIP. The remaining
5% of physician claims are provided through alternate funding plans provided through
academic centres, family health teams, health service organizations, and some psychiatric
hospitals. Unlike CIHI-DAD and CIHI-NACRS, this database is not routinely validated for data
accuracy. We used this database in combination with CIHI databases to identify comorbidities.
5.1.1.4 Registered Persons Database The Registered Persons Database provides demographic information on any individual
who possesses an Ontario health card. This demographic information includes date of birth,
postal code. Although it provides information regarding death, this database is not updated as
regularly as other databases thereby limiting its accuracy especially regarding timing of death.
At the time of this study, the Vital Statistics database, which contains the medical cause and
date of death, was not available. As a result, we used the Registered Persons Database to
identify the outcome, death.
26
5.2 Cohort Definition
5.2.1 Study and Control Groups
We identified all patients aged 66 years or older who were new users of venlafaxine
(study group), sertraline (control) or citalopram (control) during the study period. We defined
new users as patients who had not filled a prescription for any antidepressant in the preceding
year. Sertraline- and citalopram-treated patients served as control groups because these drugs
are similar to venlafaxine in serotonergic activity,(79) efficacy(44) and popularity in Canada(43)
but do not potentiate noradrenaline. Patients were followed until they experienced the
primary outcome (defined below), switched or discontinued antidepressant therapy, reached
the end of the study period (March 31st, 2010) or completed one year of therapy. Based on
previous literature, which included case series and observational studies,(51,94-96) this one
year observation window was judged to be sufficient time to capture drug-specific adverse
cardiac events, particularly because they are most likely to manifest within the first several
weeks to months of treatment.(51,94,95,105)
5.2.2. Inclusion Criteria
We included all patients aged ≥66 years of age with a valid Ontario Health Card who
started a new prescription for venlafaxine, sertraline or citalopram between April 1st 2000 and
March 31st 2009. Patients were only included once in the study accrual period.
5.2.3 Exclusion Criteria
27
We did not study patients during their first year of eligibility for prescription drug
coverage (age 65) by the ODB. These patients would have incomplete medication records for
the preceding one year, which would have prohibited us from measuring variables pertinent to
the exposure drugs and outcome. We excluded patients concomitantly taking other
antidepressants, as well as those who started any combination of venlafaxine, sertraline or
citalopram on the same day. Patients on tamoxifen, an estrogen receptor antagonist used in
the treatment of estrogen-receptor positive breast cancer, were excluded because SSRIs and
SNRIs may differentially modulate the response to tamoxifen and subsequently mortality.(106)
5.3 Outcome Definition
We defined the primary outcome a priori as a composite of death from any cause or
hospital admission for acute myocardial infarction or heart failure. In a secondary analysis, each
component was analyzed separately. We identified death using the RPDB, and hospitalization
for acute myocardial infarction or congestive heart failure with ICD-9 and ICD-10 codes from
the CIHI DAD. We defined acute myocardial infarction using ICD-9 code 410, or ICD-10 codes
I21 and I22, and I20 (unstable angina); and congestive heart failure (CHF) using ICD-9 code 428
or ICD-10 code I50. In order to exclude patients with nonischemic chest pain, we limited our
definition of myocardial infarction to patients hospitalized for at least three days.(107) The
date of death or hospital admission was used as the outcome date for all analyses. These
outcome definitions for congestive heart failure and myocardial infarction have been previously
validated, with positive predictive values of approximately 90%.(108-110)
5.3.1 Outcome Validation Studies
28
Validation studies of these ICD-9 and -10 codes have been performed in Canada,
specifically Ontario, Saskatchewan and Alberta.(108-111) Studies linking CIHI-DAD ICD-9 and -10
codes for AMI and CHF with Saskatchewan hospital records and the Ontario-based Fastrak II
Acute Coronary Syndrome registry had positive predictive values of 88-95%, however the
prevalence of these conditions was high (90-100%).(108,110) Furthermore, in the Ontario-
based Fastrak II study, the sensitivity of only the ICD-9 code 410 for myocardial infarction was
89%.(108) In studies that linked these codes with either electronic medical records from
primary care physicians, or a chart review of urban and rural hospitals in Ontario, the
prevalence of AMI and CHF was lower at 5-10%, yet the positive predictive value remained at
89-94% and sensitivity at 61%.(109,111) Although including outpatient OHIP diagnosis codes
for CHF might have increased the sensitivity of our outcome definition to 86% by including mild
cases not requiring hospitalization, we decided against this due to their poor positive predictive
values ranging from 38 to 46%.(112)
Finally, as a sensitivity analysis to assess the robustness of our findings to potentially
unmeasured confounding variables, we conducted a ‘tracer analysis.’ Here we replicated our
analyses using gastrointestinal hemorrhage as the outcome of interest, because we expected
no differential risk of hemorrhage among the three patient groups.(113,114)
5.4 Covariates
We recorded covariates that would be associated with the prescription of either
venlafaxine (study), citalopram (control) or sertraline (control), or with adverse cardiac events
(outcome). Broadly, these covariates included demographics, antidepressant dose category,
markers of burden of illness, comorbidities related to drug exposure or outcome, and
29
medications associated with cardiovascular disease, mortality and neuropathic pain.(Tables 1a
and b, Appendices 1A and B)
Demographic covariates included age and gender. As markers of burden of illness, we
measured a patient’s number of distinct medications(115) and hospitalization status in the past
year(116), and the Aggregated Diagnosis Groups within the past 3 years.(117) Aggregated
Diagnosis Groups are a validated measure of burden of illness that account for a patient’s
diagnoses in the ambulatory and hospitalized setting, chronicity and severity of illness,
diagnostic certainty, etiology of the disease and involvement of specialty care.(117) This range
of comorbidity markers allowed us to estimate the burden of illness of all patients regardless of
their hospitalization status and they have been validated in health services research for
mortality and health care utilization.(115,117)
5.4.1 Variables Associated with Drug Exposure or Outcome
5.4.1.1 Variables Associated with Drug Exposure
We measured variables that were associated with exposure to our drugs of interest-
venlafaxine, citalopram or sertraline. These included comorbidities such as depression, anxiety,
dementia, neuropathic pain and hot flashes associated with tamoxifen therapy among breast
cancer patients. We measured depression, anxiety and dementia with ICD-9 and ICD-10 codes
from CIHI DAD and NACRS databases. The latter condition is associated with an increased risk of
mortality and behavioural psychological symptoms of dementia, which include depression and
anxiety.(118,119) As mentioned, we excluded patients on tamoxifen since antidepressants may
differentially affect the effectiveness of this medication.(106)
30
Since venlafaxine’s noradrenergic effects are more prominent at higher doses (≥200 mg
per day)(41,60,62,120), we thought that it would be important to explore any association
between dose and the primary outcome. In clinical practice, new drugs are introduced to older
patients at the lowest possible dose and then slowly titrated upwards while monitoring for
adverse effects.(121) Patients on higher antidepressant doses are more likely to suffer from
severe depression, and less likely to have genetic polymorphisms that render them intolerant of
this medication at even lower doses. We defined low, moderate and high dose categories for
venlafaxine, sertraline and citalopram based on clinical and pharmacologic properties. The low
dose categories were defined as the usual initiation dose, i.e. the lowest available doses with
venlafaxine 37.5 mg/day, sertraline 25 mg/day and citalopram 10 mg/day. High dose categories
were defined as venlafaxine >200 mg/day, sertraline >150 mg/day and citalopram >40
mg/day.(84,85,120,122,123) Moderate dose categories were defined as venlafaxine 37.6-200
mg/day, sertraline 26-150 mg/day and citalopram 11-40 mg/day.
As surrogate markers of dementia and neuropathic pain, conditions for which
venlafaxine is sometimes used, we recorded the use of other medications often prescribed for
these conditions. Venlafaxine has been prescribed for patients with depression and
concomitant pain syndromes who may also be taking opioids, a risk factor for
mortality.(124,125) As mentioned earlier, venlafaxine has also been used to treat behavioural
psychological symptoms of dementia, a known risk factor for mortality.(119,126,127) Since
diagnosis codes for “pain syndromes” and “dementia” have not been validated or have variable
accuracy,(128-130) we attempted to identify these patients through their use of known pain
31
medications, opioids and gabapentin; or antipsychotics and cholinesterase inhibitors
respectively.
5.4.1.2 Variables Associated with Outcome
Our primary outcome was a composite of ischemic heart disease (acute myocardial
infarction and unstable angina), congestive heart failure and all-cause mortality. Therefore, we
recorded the proportion of patients with cardiovascular disease and its different variants
(included diagnosis codes for congestive heart failure, valvular disease, cardiovascular disease
unspecified, angina, myocardial infarction, ischemic heart disease, conduction disorder and
cerebrovascular disease) as comorbidities, and renal disease at baseline.(131) (Tables 1a and b)
The Aggregated Diagnosis Groups comorbidity tool allowed us to differentiate between stable
and unstable cardiovascular disease.(Appendices 1A and B) Medications associated with the
treatment or exacerbation of cardiovascular disease were also included. Medications
associated with the treatment of cardiovascular disease included antiplatelet agents such as
clopidogrel and acetylsalicylic acid, anticoagulants, loop diuretics, statins, negative
chronotropes (beta-blockers, non-dihydropyridine calcium channel blockers, digoxin),
antiarrhythmics, aldosterone antagonists and renin-angiotensin antihypertensives (angiotensin
converting enzyme inhibitors and angiotensin receptor blockers).(132,133) Medications
associated with exacerbation of heart disease include nonsteroidal anti-inflammatory drugs,
thiazolidinediones and systemic steroids. We were unable to measure other cardiovascular risk
factors such as smoking status, waist circumference, family history of cardiovascular disease
and the control of blood pressure and cholesterol from our available databases. There is no
reason, however, for these risk factors to be differently distributed among the three
32
antidepressant drug groups at baseline. Finally, antipsychotic and cholinesterase inhibitor use
are not only associated with drug exposure as discussed above; but also outcome as they have
been linked to increased mortality among older patients.(126,127)
5.5 Statistical Analysis
5.5.1 Baseline Characteristics
We compared the baseline characteristics of patients in the three groups using
standardized differences, which are not as sensitive to sample size as conventional P
values.(134) A meaningful difference was defined as a standardized difference exceeding
0.10.(134,135)
5.5.2 Propensity Score
5.5.2.1 Introduction to the Propensity Score Confounding When estimating the effect a drug treatment or exposure has on a particular outcome,
it is important to compare groups of individuals who are otherwise similar. This would allow
one to study the association between drug and outcome without interference from a
confounder. In our case, a confounder is a variable associated with both exposure drug and the
outcome. An example of a confounding variable in this study investigating venlafaxine (study
drug) and cardiovascular events (outcome) would be diabetes. Diabetes is associated with
neuropathy, for which venlafaxine is prescribed, and cardiovascular disease (outcome). If more
patients in the venlafaxine group had diabetes compared to the control groups then the
venlafaxine group would be more likely to suffer an adverse cardiac event, a source of bias.
33
Furthermore, diabetes is not an intermediate pathway between venlafaxine and the
development of cardiovascular disease.(56)
In randomized controlled trials, individuals are randomly allocated to either study or
control groups. This usually results in an equal distribution of variables, including confounders,
thereby allowing one to characterize the association between the study drug and outcome by
directly comparing the outcomes between groups. In observational studies, however,
investigators do not assign individuals to either study or control groups. Rather, subject
characteristics may influence whether they are exposed to the study or control groups. Unless
these characteristics, which include confounders, are accounted for, directly comparing
outcomes between study groups would be inaccurate. One tool that addresses this issue in
observational studies is the propensity score.
The Propensity Score
The propensity score is the probability of receiving a certain study treatment given the
individual’s observed characteristics.(136) When deriving the propensity score, one can include
variables that are imbalanced between treatment groups and/or that are associated with the
treatment and outcome. In general, propensity scores only account for observed covariates,
therefore one cannot adjust for unmeasured covariates, a potential source of bias.
Propensity scores can be used for matching, stratification, inverse probability of
treatment weighting and covariate adjustment. With matching and stratification, there is an
effort to compare subjects or strata from the treatment and comparison groups with similar
propensity scores, and therefore similar baseline covariates which may affect treatment
assignment or outcome.(136) During covariate adjustment with the propensity score, assuming
34
there is a linear relationship between propensity scores and the outcome, the outcome is
regressed on the propensity score as an indicator for treatment. Unlike the other methods, one
cannot easily assess the quality of the propensity score model (balance assessment) in
covariate adjustment.(137) Another disadvantage of propensity score covariate adjustment is
the slight bias towards the null hypothesis when the outcomes are binary or time-to-event, as
in our study. With weighting, each subject is weighted by the inverse probability of the
treatment they received (study vs. comparison) and the outcomes from the study and control
groups of this derived sample are directly compared.(136) Studies comparing the various
methods of propensity score implementation suggest that matching and weighting with the
propensity score were more successful than stratification and covariate adjustment at
removing systematic differences between study and control groups.(136) In our study, we used
the propensity score to weight.
5.5.2.2 Implementation of the Propensity Score
To control for baseline differences between the two treatment groups, we weighted the
cohort using inverse probability of treatment weights (IPTW) derived from the propensity
score.(136) We used IPTW in order to avoid losing study subjects, which could occur during the
matching process, to allow us to perform balance diagnostics after applying the propensity
score and to confirm the equalization of differences in baseline covariates between the
groups.(136,137) The propensity score was estimated using a logistic regression model that
included all potential confounders with the exception of drug dose listed in Tables 1a and b.
Separate propensity scores were generated for the venlafaxine/sertraline and
venlafaxine/citalopram comparisons. We then performed a balance assessment(136,137) of our
35
IPTW samples using standardized differences, and visually inspected the weights using
boxplots.(138)
5.5.3 Survival Analysis Time-to-Event Analysis
We performed a survival analysis to study the risk of adverse cardiovascular events
among patients on venlafaxine compared to those on sertraline or citalopram over time.
Using the weighted sample, we performed time-to-event analyses with sertraline and
citalopram as reference groups. We estimated hazard ratios with 95% confidence intervals in
the weighted sample using Cox proportional hazards regression and obtained a robust variance
estimate. From the fitted model, we derived survival curves for each treatment group.
Because of the observational nature of our study and subsequent lack of randomization, we did
not construct unadjusted Kaplan-Meier survival curves (1-probability of outcome). This analysis
does not allow one to account for confounding variables and other sources of bias. For
example, patients with cardiovascular disease are at increased risk for future cardiovascular
events compared to those without a history of cardiovascular disease. If patients with
cardiovascular disease had physicians who were concerned about venlafaxine’s noradrenergic
effects and therefore more likely to be prescribed sertraline or citalopram, then the unadjusted
Kaplan-Meier curve for these drugs may be lower than that of venlafaxine. Instead, we used the
method described by Cole et al. where we derived adjusted survival curves using inverse
probability weights.(139)
Supplementary analyses included stratification for pre-existing cardiovascular disease
and replication of all analyses with trimmed and stabilized weights.(138)
36
5.5.4 Regression-based analysis
We explored the relationship between drug dose and the primary outcome with
regression-based analyses. As previously described, we defined 3 drug dose categories: low,
moderate and high. Individuals would be classified into one of each drug dose category based
on the highest prescription dose filled during the observation window. Because it is customary
for patients to start with the lowest dose of medication and then increase according to
response and tolerance to treatment, patients tended to receive moderate or high doses later
on during the observation window. Therefore, by the time patients were administered these
higher doses, their propensity scores, which were derived from covariates at baseline and in the
preceding 3 years, may no longer be accurate. Because conventional IPTW using the propensity
score is designed for use with exposures that are fixed at baseline or in the preceding 3 years,
the exploratory secondary analyses that incorporated time-dependent covariates like dose
were conducted in the original, unweighted sample. A Cox proportional hazards model was fit
to estimate the effect of dose on the hazard of the composite outcome. In this set of analyses,
dose was treated as a time-varying covariate, and we adjusted for type of antidepressant
(venlafaxine vs. sertraline or citalopram) and all measured confounding variables that achieved
clinical and statistical significance (defined as a standardized difference >0.1)(Tables 1A and B,
Appendices 1A and B).
All analyses were conducted using SAS version 9.3 (SAS Institute, Cary, North Carolina).
5.6 Ethics
37
This study was approved by the Research Ethics Boards of Sunnybrook Health Sciences
Centre and the University of Toronto.
38
6 RESULTS
6.1 Description of study and control cohorts
During the 10-year study period, we identified 48,876 patients with depression who
commenced treatment with venlafaxine, 41,238 who commenced sertraline and 3,909 patients
who commenced citalopram. These patients were followed for a median of 105 (Interquartile
Range (IQR) 45 to 365), 90 (IQR 45 to 317) days and 287 (IQR 102 to 365) days respectively.
During the observation window, 65% of venlafaxine, 62% of sertraline and 76% of citalopram
patients switched doses. Only 2.9%, 2.3% and 12% of patients were in the high dose category
for venlafaxine, sertraline and citalopram respectively.(Tables 1a and 1b) During the one year
of follow-up, 8.5% of patients experienced the primary composite outcome, 1.4% switched
antidepressants, 63.7% discontinued their antidepressant and 26.3% completed the full year of
follow up without experiencing the primary outcome.
6.1.1 Venlafaxine-Sertraline Comparison
Subjects treated with venlafaxine and sertraline exhibited similar baseline demographics
(IPTW range 1.2 to 7.4, IQR 1.9 to 2.4)(Figure 1) and comorbidities, although minor differences
were found with regard to age and history of cardiovascular and psychiatric disease (Table 1a,
Appendix 1A). We successfully adjusted for these differences by weighting with the propensity
score (Table 1a, Appendix 1A). In the weighted samples, all variables had standardized
differences less than or equal to 0.007, indicating that all meaningful differences in means and
prevalence estimates of measured baseline covariates had been eliminated. Approximately
half of the study population on these two antidepressants had preexisting cardiovascular
disease at the outset of antidepressant therapy.
39
6.1.2 Venlafaxine-Citalopram Comparison
Patients on citalopram exhibited different characteristics from those on venlafaxine and
sertraline. This cohort of patients was much smaller, generally older (citalopram median age 81
years (IQR 74 to 87), venlafaxine median age 75 years (IQR 70 to 81)), received more
medications (citalopram median 11 (IQR 8-16) compared to venlafaxine 9 (IQR 5-13), less likely
to have anxiety and more likely to have a previous history of heart failure, stroke, renal disease,
dyslipidemia and dementia (Table 1b). Patients on citalopram had a longer duration of therapy
(median 287 days, IQR 102-365). They were also more likely to be on concomitant loop
diuretics, renin angiotensin agents, statins, warfarin, thiazolidinediones, and antipsychotics, but
less likely to be taking nonsteroidal anti-inflammatory drugs (NSAIDs) and opioids. From the
Aggregate Diagnosis Groups tool, the citalopram group was less likely than the venlafaxine and
sertraline groups to have a previous minor primary infection, allergies, asthma, stable chronic
medical disease, stable ophthalmologic disease, dermatologic conditions, and stable
psychosocial disease; and more likely to have progressive, recurrent diseases, unstable chronic
medical disease, major injury, and unstable psychosocial disease (Table 1b). Similar to the
venlafaxine-sertraline comparison, we generated propensity scores with all the variables in
Table 1b and Appendix 1b. These scores were then used to IPTW the samples and balance
diagnostics were performed (Table 1b). The citalopram group had several differences compared
to venlafaxine, which were balanced with IPTW (Table 1b) albeit with a greater range of weights
compared to the venlafaxine-sertraline comparison (IPTW range 1.1 to 372.8 (IQR 3.9 to
17.2))(Figure 1). In light of these inherent differences, we compared venlafaxine to sertraline
and citalopram separately rather than pooling the sertraline and citalopram groups as one
40
comparison group. These numerous differences also hindered us from performing the time-
varying dose analysis of venlafaxine compared to citalopram. With 23 covariates which would
be adjusted for in the analysis (18 observed covariates which differed between the drug groups
and an additional 5 clinically significant covariates)(k), 7.5% weighted proportion of citalopram
patients experiencing the outcome (p), and the 10 event per variable rule for proportional
hazards regression (Sample size N = 10 k/p),(140) we were underpowered for this analysis.
Furthermore, we were concerned that there were additional unmeasured confounders at
baseline or which may have arisen over time.
6.2 Primary Analysis
6.2.1 Venlafaxine-Sertraline Comparison
In the primary analysis, 3966 (8.1%) of venlafaxine- and 3707 (9.0%) of sertraline–
treated patients experienced the composite outcome of death or hospital admission for acute
myocardial infarction or heart failure (Table 2a). After weighting with the propensity score, we
found no significant difference in the risk of the primary outcome in patients started on
venlafaxine compared to sertraline (weighted hazard ratio 0.97; 95% confidence interval 0.93 to
1.02; Table 2a; Survival curves Figure 2a). We also found no significant difference in the
secondary outcomes of death or acute myocardial infarction (Table 2a). However, we
unexpectedly found that venlafaxine use was associated with a lower incidence of heart failure
(weighted hazard ratio 0.87; 95% CI 0.80 to 0.95). The propensity score-generated weights for
venlafaxine and sertraline groups were similar, and supplementary analyses using the stabilized
and trimmed inverse probability of treatment weights yielded similar results. As expected, we
41
found no significant difference in the risk of gastrointestinal hemorrhage (weighted hazard ratio
0.99; 95% CI 0.84 to 1.16).
6.2.2 Venlafaxine-Citalopram Comparison
In the venlafaxine-citalopram comparison, 3,966 (8.1%) of venlafaxine patients and 456
(11.7%) of citalopram patients experienced the primary outcome (Table 2b). After weighting
with the propensity score, there was an increased risk of the primary outcome in venlafaxine
patients with a weighted hazard ratio 1.39 (95% CI 1.19 to 1.62)( Table 2b; Survival curves
Figure 2b). We found an increased risk of death and acute myocardial infarction among users
on venlafaxine compared to citalopram (weighted hazard ratio 1.49 (95% CI 1.26 to 1.76) and
1.59 (95% CI 1.5 to 2.41) respectively) however, there was no significant difference in the
incidence in heart failure (weighted hazard ratio 1.22 (95% CI 0.87 to 1.72). Finally, we did not
find a difference in venlafaxine- and citalopram-users in the tracer outcome, gastrointestinal
bleed (weighted hazard ratio 1.27 (95% CI 0.82 to 1.96). Although there was a wide range of
weights in this comparison,(Figure 1) which could affect results by providing estimates with
high variance,(138) we acquired similar results when the analysis was repeated with trimmed
and stabilized weights.(138)
6.3 Secondary Analysis
6.3.1 Stratified Analysis
We performed a stratified analyses examining the risk of the primary outcome among
patients with or without pre-existing cardiovascular disease. The trends from the primary
analysis persisted in our secondary analysis in which we stratified for baseline cardiac disease
(Figures 3A and 3B, 4A and 4B).
42
6.3.2 Time-Varying Analysis of Dose: Venlafaxine-Sertraline Comparison
During the exploratory time-varying analysis of dose for the primary outcome, we could
not use the propensity scores for IPTW that were derived from baseline patient characteristics.
Instead, we adjusted for the variables that were statistically different between the venlafaxine
and sertraline cohorts (standardized difference >0.10)(Table 1a, Appendix 1A) or clinically
significant.(Table 3) These included age, sex, hospitalization status in the preceding 12 months,
stroke, congestive heart failure or cardiovascular disease, use of statin and antiplatelet
medications; and the Aggregated Diagnosis Group, “progressive symptoms” and
“stable/persistent psychosocial disease.”
Similar to the primary analysis, we found no difference in the primary outcome between
venlafaxine and sertraline. There was an increased risk for adverse cardiac events among
patients on high dose antidepressants compared to low dose antidepressants (Table 3). This
was not unexpected, since high dose antidepressants are often used in patients with more
severe depression, a known risk factor for adverse cardiac events.(7,70,141)
6.3.3 Time-Varying Analysis of Dose: Venlafaxine-Citalopram Comparison
Unlike the venlafaxine-sertraline scenario, patients in the venlafaxine and citalopram
cohorts were dissimilar (Table 1b). We identified 18 baseline covariates in which the groups
differed, and there remained the possibility of additional unmeasured confounders. There
could be differences over time, in addition to dose, that we would be unable to measure in this
study. Of note, citalopram patients continued their antidepressant for twice as long as
venlafaxine and sertraline patients (citalopram median 287 (IQR 102 to 365) days; venlafaxine
median of 105 days (IQR) 45 to 365; sertraline median 90 (IQR 45 to 317) days) and they were
43
more likely to be in the high dose category (citalopram 12% compared to venlafaxine 3%).
Therefore it is likely that additional differences between the study groups developed over time.
Furthermore, the propensity score was derived from baseline characteristics, therefore we
were unable to use it as covariate adjustment during the time-varying analysis.(Table 1b,
Appendix 1B). Due to these differences, and the inability to use the propensity score, the time
varying analysis of dose was not performed for the venlafaxine-citalopram drug comparison.
7 DISCUSSION
7.1 Major Findings
Using the health records of more than 94,000 older patients, we found no increased risk
of adverse cardiac events among patients treated with low to moderate-doses of venlafaxine as
compared with sertraline. However, when compared to citalopram, we found an increased risk
of adverse cardiac events, specifically acute myocardial infarction and death among venlafaxine
patients. These findings held regardless of baseline cardiovascular disease. This finding is
important in light of limited evidence that venlafaxine’s noradrenergic effects might confer
increased cardiovascular risk.
The lower risk of heart failure among venlafaxine patients compared to sertraline
patients was an unexpected finding. One hypothesis is that venlafaxine triggered
cardiovascular symptoms of palpitations, similar to the observational study by Johnson and
colleagues,(51) leading patients to seek medical attention. Patients treated in the community
for cardiovascular symptoms, such as hypertension or tachycardia, or mild cases of heart failure
or myocardial ischemia might avoid a hospital admission. Since our primary outcome only
44
looked at serious cardiac events resulting in death or hospitalization, these events may have
been missed and therefore resulted in a lower incidence of heart failure. Another hypothesis is
that venlafaxine might be more effective than sertraline or citalopram at treating
depression,(44) allowing patients to better manage their chronic medical conditions,
particularly cardiovascular disease. We observed no difference in the tracer outcome
(gastrointestinal bleeding) between groups, however this disease is less dependent on
adherence to medical therapy. Finally, there exists the possibility that residual confounding
could have influenced our findings. Future observational studies that include electronic medical
record databases in primary care practices or pragmatic randomized controlled trials could be
helpful to explore the basis of this association.
7.1.1 Consistency with Existing Literature
Our findings complement those of other recent population-based studies that found no
increased risk of death or cardiac arrhythmia among patients on venlafaxine;(90,91,142)
although these studies did not examine the risk of myocardial infarction or heart failure, which
might be particularly important given the drug’s noradrenergic effects.
7.1.1 Existing Literature
Clinical Trials
As mentioned above, a number of clinical trials of second generation antidepressants
found no increased risk in adverse cardiac events among patients on sertraline or
venlafaxine.(47,48) With small numbers of older patients, however, the results from these
clinical trials could not be generalized to this vulnerable population.(47,48) A recent meta-
analysis and meta-regression of randomized trials of the effect of antidepressants and late-life
45
depression found that few clinical trials included older individuals, however none included
safety as an outcome.(36) To date, the existing clinical trials do not adequately address cardiac
safety of venlafaxine among older patients with comorbidities.
Observational Studies
Several observational studies have tried to characterize the cardiac risk of
antidepressants in real-world patients. The case-control study by Sauer et al. using telephone
surveys found a decreased risk of myocardial infarction among patients on antidepressants with
high serotonin transporter affinity, which included sertraline, compared to those with low
serotonin transporter affinity; however, their study participants were younger, had no history
of cardiovascular disease and were not necessarily new users of their antidepressants.(50)
Furthermore, as previously mentioned, instead of performing a head-to-head comparison of
sertraline and venlafaxine, this study compared groups of antidepressants with different
serotonin transporter affinities.(50) Venlafaxine was grouped among other non-SSRI
antidepressants, along with tricyclic antidepressants, trazodone, bupropion and mirtazapine,
which aside from their serotonin transporter affinity properties are pharmacologically different.
Although, the observational study by Johnson et al. found an increase in adverse cardiac effects
such as increased heart rate and blood pressure among depressed older community-dwelling
patients on venlafaxine, there were no cases of myocardial infarction or heart failure. This
study’s small sample size, however, limited its statistical power to detect such cardiac
events.(51)
Our study is the first population-based study to specifically compare the cardiac safety
of the commonly prescribed antidepressants, venlafaxine, sertraline, and citalopram and it
46
provides further insight into the safety of these drugs among older patients with cardiovascular
disease.(52) Overall, our results provide a measure of reassurance to physicians treating
depression among older patients, particularly those with cardiovascular disease.
7.1.2 Limitations Some limitations of our study merit discussion. The results derive from older patients,
and the generalizability to younger patients is unknown. However, our study addresses
venlafaxine’s risk in an understudied population in a real-world setting; this is relevant since
most real-world patients treated for depression are excluded from randomized controlled
trials.(35) Although we utilized a propensity score incorporating variables that might relate to
both exposures and outcomes, we had no information on important factors such as body
weight, waist circumference, smoking status, cholesterol and blood pressure control, and
CYP2D6 phenotype. However, there is no apparent reason why these factors might differ
between venlafaxine, sertraline and citalopram groups. Nevertheless, the venlafaxine and
sertraline cohorts were generally similar even before weighting and we found no significant
difference in our tracer outcome. Patients on citalopram, however, were different from the
other antidepressant groups. In general, this smaller cohort consisted of older and frailer
patients with a heavier burden of disease, including cardiovascular disease. Therefore, despite
weighting with the propensity score and achieving similar results with trimmed and stabilized
weights and the tracer outcome, unmeasured confounders may have influenced the study. We
suggest a cautious interpretation of this study’s venlafaxine-citalopram comparison and feel
that further research with randomized controlled trial or propensity score matched cohort
study designs to explore this association is indicated.
47
The administrative nature of our data only allowed us to estimate dose, thereby limiting
our ability to study the risk of adverse cardiac events at higher venlafaxine doses where
noradrenergic effects are more pronounced. Although we used prescription day supply data
from the ODB database to characterize patient adherence to drug therapy, it is possible that
patients filled their prescription but did not take the medication. If venlafaxine patients were
less adherent, this would have biased the study results towards the null hypothesis. However,
there should be no reason for differences in adherence between venlafaxine and sertraline and
patients on those medications had similar durations of therapy. Conversely, citalopram
patients had longer courses of therapy and could have had greater adherence. These patients
were more likely to be older and have greater burdens of illness and dementia, which could
have resulted in placement at long term care facilities where medications are administered by a
second party. This, however, should not have affected the observed increased risk of adverse
cardiac events among patients on venlafaxine compared to citalopram. In addition, we could
not use the propensity score for this regression analysis with the time-dependent covariate,
dose, because this score was derived from baseline characteristics which may have changed
over time. We did, however, adjust for characteristics which were either clinically and/or
statistically significant in our venlafaxine-sertraline time-varying dose comparison. A potential
source of bias associated with this analysis is the nonrandom changes in drug dose. Increases in
drug dose may not have been random but rather preceded by an exacerbation of depression or
an indication of more severe disease.(143) By hindering one’s ability to manage his or her
chronic diseases, depression is associated with mortality and morbidity.(11,12) Although we
adjusted for covariates that were clinically and/or statistically significant, which may reduce this
48
bias, the nonrandom changes in drug dose should still be acknowledged. As mentioned
previously, due to the multiple differences between patients using citalopram compared to
venlafaxine or sertraline, we were unable to conduct a similar time-varying dose analysis
between venlafaxine and citalopram. Therefore, whether high-dose venlafaxine is associated
with increased cardiac risk remains unknown.
Because some clinicians concerned about the hemodynamic effects of venlafaxine might
have avoided this drug in those with cardiovascular or hypertensive disease, there is a risk of
channeling bias. Although this difference in baseline cardiovascular disease between groups
was successfully balanced with IPTW, this potential bias deserves mention.
Our validated primary outcome allowed us to identify adverse cardiac events severe
enough to result in death or hospitalization. It did not, however, allow us to capture less severe
cardiac events and effects such as mild heart failure, exacerbated hypertension or tachycardia
that might have been induced by venlafaxine.(51) If these mild conditions prompted timely
outpatient interventions, hospitalization could be prevented, therefore resulting in the
decreased risk of hospitalization for heart failure. As previously mentioned, although
outpatient OHIP diagnosis codes for hypertension, tachycardia, and heart failure exist, we did
not include them in our outcomes because they are not well validated in isolation.(112)
7.1.3 Implications for Clinical Practice
In summary, our population-based study of older Ontario adults found no significant
difference in the cardiovascular safety profiles of venlafaxine compared to sertraline. Our
results offer a measure of reassurance about the drug’s cardiac safety, however physicians
49
should still exercise caution in patients on higher doses or those who manifest overt
noradrenergic effects.
7.2 Contributions
Our study is interesting from methodological and clinical points of view.
Our study used the propensity score as IPTWs to weight our sample and therefore was
able to include all older individuals in Ontario who fulfilled our inclusion and exclusion criteria.
We recognized that using the propensity score in this fashion, however, limited our ability to
study the time-dependent covariate of dose. We found that IPTW was more successful when
comparing similar cohorts as in the venlafaxine/sertraline comparison.
Older Ontario patients on citalopram were more likely to be older, have cardiovascular
disease and to have a larger burden of illness compared to those on venlafaxine or sertraline.
Not only did we find that the majority of patients were on low to moderate doses of
antidepressants, but there was no significant difference in adverse cardiac events among those
patients on low to moderate doses of venlafaxine compared to sertraline.
7.3 Future Directions
A methodological study to investigate the risk of cardiac events among older patients on
venlafaxine compared to sertraline or citalopram may include a cohort study using propensity
score matching instead of IPTW. While this may result in the loss of venlafaxine patients who
were not successfully matched to citalopram patients, it would allow us to better characterize
the risk of venlafaxine compared to citalopram.
50
In light of the FDA’s recent alert regarding high-dose citalopram and the risk of cardiac
dysrhythmias, a population-based case-control study to compare citalopram to sertraline and
venlafaxine could be important to physicians, patients and health policy makers.
51
8 REFERENCES
(1) American Psychiatric Association, and American Psychiatric Association. Task Force on DSM-IV. Diagnostic and statistical manual of mental disorders: DSM-IV. : Amer Psychiatric Pub Inc; 1994.
(2) Patten SB, Kennedy SH, Lam RW, O'Donovan C, Filteau MJ, Parikh SV, et al. Canadian Network for Mood and Anxiety Treatments (CANMAT) clinical guidelines for the management of major depressive disorder in adults. I. Classification, burden and principles of management. J Affect Disord 2009 Oct;117 Suppl 1:S5-14.
(3) Patten SB, Li Wang J, Williams JVA, Currie S, Beck CA, Maxwell CJ, et al. Descriptive epidemiology of major depression in Canada. Canadian Journal of Psychiatry 2006;51(2):84.
(4) Thombs BD, Bass EB, Ford DE, Stewart KJ, Tsilidis KK, Patel U, et al. Prevalence of depression in survivors of acute myocardial infarction. J Gen Intern Med 2006 Jan;21(1):30-38.
(5) Kessler RC, Berglund P, Demler O, Jin R, Koretz D, Merikangas KR, et al. The epidemiology of major depressive disorder. JAMA: the journal of the American Medical Association 2003;289(23):3095.
(6) Hamer M, Bates CJ, Mishra GD. Depression, physical function, and risk of mortality: National Diet and Nutrition Survey in adults older than 65 years. Am J Geriatr Psychiatry 2011 Jan;19(1):72-78.
(7) Frasure-Smith N, Lesperance F, Talajic M. Depression following myocardial infarction. Impact on 6-month survival. JAMA 1993 Oct 20;270(15):1819-1825.
(8) Sanderson K, Tilse E, Nicholson J, Oldenburg B, Graves N. Which presenteeism measures are more sensitive to depression and anxiety? J Affect Disord 2007 Aug;101(1-3):65-74.
(9) Rapaport MH, Clary C, Fayyad R, Endicott J. Quality-of-life impairment in depressive and anxiety disorders. Am J Psychiatry 2005 Jun;162(6):1171-1178.
(10) Murray CJL LA. Alternative projections of mortality and disability by cause 1990–2020: Global Burden of Disease Study. The Lancet 1997;9064:1498-1504.
(11) Ciechanowski PS, Katon WJ, Russo JE. Depression and diabetes: impact of depressive symptoms on adherence, function, and costs. Arch Intern Med 2000 Nov 27;160(21):3278-3285.
52
(12) McIntyre RS, Konarski JZ, Wilkins K, Soczynska JK, Kennedy SH. Obesity in bipolar disorder and major depressive disorder: results from a national community health survey on mental health and well-being. Can J Psychiatry 2006 Apr;51(5):274-280.
(13) Scott J, Dickey B. Global burden of depression: the intersection of culture and medicine. Br J Psychiatry 2003 Aug;183:92-94.
(14) Parikh SV, Segal ZV, Grigoriadis S, Ravindran AV, Kennedy SH, Lam RW, et al. Canadian Network for Mood and Anxiety Treatments (CANMAT) clinical guidelines for the management of major depressive disorder in adults. II. Psychotherapy alone or in combination with antidepressant medication. J Affect Disord 2009 Oct;117 Suppl 1:S15-25.
(15) Kennedy SH, Milev R, Giacobbe P, Ramasubbu R, Lam RW, Parikh SV, et al. Canadian Network for Mood and Anxiety Treatments (CANMAT) Clinical guidelines for the management of major depressive disorder in adults. IV. Neurostimulation therapies. J Affect Disord 2009 Oct;117 Suppl 1:S44-53.
(16) Petrides G, Fink M, Husain MM, Knapp RG, Rush AJ, Mueller M, et al. ECT remission rates in psychotic versus nonpsychotic depressed patients: a report from CORE. J ECT 2001 Dec;17(4):244-253.
(17) Prudic J, Haskett RF, Mulsant B, Malone KM, Pettinati HM, Stephens S, et al. Resistance to antidepressant medications and short-term clinical response to ECT. Am J Psychiatry 1996 Aug;153(8):985-992.
(18) Tess AV, Smetana GW. Medical evaluation of patients undergoing electroconvulsive therapy. N Engl J Med 2009 Apr 2;360(14):1437-1444.
(19) Kramer BA. Use of ECT in California, revisited: 1984-1994. J ECT 1999 Dec;15(4):245-251.
(20) Shiwach RS, Reid WH, Carmody TJ. An analysis of reported deaths following electroconvulsive therapy in Texas, 1993-1998. Psychiatr Serv 2001 Aug;52(8):1095-1097.
(21) Thase ME, Friedman ES, Biggs MM, Wisniewski SR, Trivedi MH, Luther JF, et al. Cognitive therapy versus medication in augmentation and switch strategies as second-step treatments: a STAR*D report. Am J Psychiatry 2007 May;164(5):739-752.
(22) Hollon SD, DeRubeis RJ, Shelton RC, Amsterdam JD, Salomon RM, O'Reardon JP, et al. Prevention of relapse following cognitive therapy vs medications in moderate to severe depression. Arch Gen Psychiatry 2005 Apr;62(4):417-422.
(23) Reynolds CF,3rd, Dew MA, Pollock BG, Mulsant BH, Frank E, Miller MD, et al. Maintenance treatment of major depression in old age. N Engl J Med 2006 Mar 16;354(11):1130-1138.
53
(24) Dombrovski AY, Lenze EJ, Dew MA, Mulsant BH, Pollock BG, Houck PR, et al. Maintenance treatment for old-age depression preserves health-related quality of life: a randomized, controlled trial of paroxetine and interpersonal psychotherapy. J Am Geriatr Soc 2007 Sep;55(9):1325-1332.
(25) Vittengl JR, Clark LA, Dunn TW, Jarrett RB. Reducing relapse and recurrence in unipolar depression: a comparative meta-analysis of cognitive-behavioral therapy's effects. J Consult Clin Psychol 2007 Jun;75(3):475-488.
(26) Mohr DC, Hart SL, Howard I, Julian L, Vella L, Catledge C, et al. Barriers to psychotherapy among depressed and nondepressed primary care patients. Ann Behav Med 2006 Dec;32(3):254-258.
(27) Olfson M, Marcus SC. National patterns in antidepressant medication treatment. Arch Gen Psychiatry 2009;66(8):848.
(28) Pratt LA, Brody DJ, Gu Q. Antidepressant Use in Persons Aged 12 and Over: United States, 2005-2008. : US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Health Statistics; 2011.
(29) Mamdani MM, Parikh SV, Austin PC, Upshur RE. Use of antidepressants among elderly subjects: trends and contributing factors. Am J Psychiatry 2000 Mar;157(3):360-367.
(30) Rudolph JL, Salow MJ, Angelini MC, McGlinchey RE. The anticholinergic risk scale and anticholinergic adverse effects in older persons. Arch Intern Med 2008 Mar 10;168(5):508-513.
(31) Demyttenaere K, De Fruyt J. Getting what you ask for: on the selectivity of depression rating scales. Psychother Psychosom 2003 Mar-Apr;72(2):61-70.
(32) Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry 1960 Feb;23:56-62.
(33) Montgomery SA, Asberg M. A new depression scale designed to be sensitive to change. Br J Psychiatry 1979 Apr;134:382-389.
(34) McIntyre R, Kennedy S, Bagby RM, Bakish D. Assessing full remission. J Psychiatry Neurosci 2002 Jul;27(4):235-239.
(35) Rothwell PM. External validity of randomised controlled trials. The Lancet 2005;365(9453):82-93.
(36) Tedeschini E, Levkovitz Y, Iovieno N, Ameral VE, Nelson JC, Papakostas GI. Efficacy of antidepressants for late-life depression: a meta-analysis and meta-regression of placebo-controlled randomized trials. J Clin Psychiatry 2011 Dec;72(12):1660-1668.
54
(37) Zimmerman M, Chelminski I, Posternak MA. Generalizability of antidepressant efficacy trials: differences between depressed psychiatric outpatients who would or would not qualify for an efficacy trial. Am J Psychiatry 2005 Jul;162(7):1370-1372.
(38) Maclure M. Explaining pragmatic trials to pragmatic policy-makers. CMAJ 2009 May 12;180(10):1001-1003.
(39) Holliday SM, Benfield P. Venlafaxine. A review of its pharmacology and therapeutic potential in depression. Drugs 1995 Feb;49(2):280-294.
(40) ®T.M. Wyeth Pfizer Canada Inc. EFFEXOR® XR Capsules (Venlafaxine Hydrochloride) Extended Release Capsules. 2011:72.
(41) Debonnel G, Saint-André É, Hébert C, de Montigny C, Lavoie N, Blier P. Differential physiological effects of a low dose and high doses of venlafaxine in major depression. International Journal of Neuro-Psychopharmacology 2007;10(1):51-62.
(42) Feighner JP. The role of venlafaxine in rational antidepressant therapy. J Clin Psychiatry 1994 Sep;55 Suppl A:62-8; discussion 69-70, 98-100.
(43) Lam RW, Kennedy SH, Grigoriadis S, McIntyre RS, Milev R, Ramasubbu R, et al. Canadian Network for Mood and Anxiety Treatments (CANMAT) clinical guidelines for the management of major depressive disorder in adults. III. Pharmacotherapy. J Affect Disord 2009 Oct;117 Suppl 1:S26-43.
(44) Gartlehner G, Hansen RA, Morgan LC, Thaler K, Lux L, Van Noord M, et al. Comparative benefits and harms of second-generation antidepressants for treating major depressive disorder: an updated meta-analysis. Ann Intern Med 2011 Dec 6;155(11):772-785.
(45) Moura C, Bernatsky S, Abrahamowicz M, Papaioannou A, Bessette L, Adachi J, et al. Antidepressant use and 10-year incident fracture risk: the population-based Canadian Multicentre Osteoporosis Study (CaMoS). Osteoporos Int 2014 May;25(5):1473-1481.
(46) Sterke CS, Ziere G, van Beeck EF, Looman CW, van der Cammen TJ. Dose-response relationship between selective serotonin re-uptake inhibitors and injurious falls: a study in nursing home residents with dementia. Br J Clin Pharmacol 2012 May;73(5):812-820.
(47) Davidson J, Watkins L, Owens M, Krulewicz S, Connor K, Carpenter D, et al. Effects of paroxetine and venlafaxine XR on heart rate variability in depression. J Clin Psychopharmacol 2005 Oct;25(5):480-484.
(48) Glassman AH, O'Connor CM, Califf RM, Swedberg K, Schwartz P, Bigger JT,Jr, et al. Sertraline treatment of major depression in patients with acute MI or unstable angina. JAMA 2002 Aug 14;288(6):701-709.
55
(49) Smajkic A, Weine S, Djuric-Bijedic Z, Boskailo E, Lewis J, Pavkovic I. Sertraline, paroxetine, and venlafaxine in refugee posttraumatic stress disorder with depression symptoms. J Trauma Stress 2001 Jul;14(3):445-452.
(50) Sauer WH, Berlin JA, Kimmel SE. Effect of antidepressants and their relative affinity for the serotonin transporter on the risk of myocardial infarction. Circulation 2003 Jul 8;108(1):32-36.
(51) Johnson EM, Whyte E, Mulsant BH, Pollock BG, Weber E, Begley AE, et al. Cardiovascular changes associated with venlafaxine in the treatment of late-life depression. Am J Geriatr Psychiatry 2006 Sep;14(9):796-802.
(52) Coupland C, Dhiman P, Morriss R, Arthur A, Barton G, Hippisley-Cox J. Antidepressant use and risk of adverse outcomes in older people: population based cohort study. BMJ 2011 Aug 2;343:d4551.
(53) Zwarenstein M, Treweek S, Gagnier JJ, Altman DG, Tunis S, Haynes B, et al. Improving the reporting of pragmatic trials: an extension of the CONSORT statement. BMJ 2008 Nov 11;337:a2390.
(54) Amin A, Stokes M, Wu N, Gatt E, Makenbaeva D, Wiederkehr D, et al. Application of Randomized Clinical Trial Data to Actual Practice: Apixaban Therapy for Reduction of Stroke Risk in Non-valvular Atrial Fibrillation Patients. Curr Med Res Opin 2013 Jun 25.
(55) Zhang M, Holman CD, Price SD, Sanfilippo FM, Preen DB, Bulsara MK. Comorbidity and repeat admission to hospital for adverse drug reactions in older adults: retrospective cohort study. BMJ 2009 Jan 7;338:a2752.
(56) Brunton LB, Lazo JS, Parker KL editors. Goodman & Gilman's The Pharmacological Basis of Therapeutics. 11th ed. New York: McGraw-Hill; 2005.
(57) Lee S, Monz BU, Clemens A, Brueckmann M, Lip GY. Representativeness of the dabigatran, apixaban and rivaroxaban clinical trial populations to real-world atrial fibrillation patients in the United Kingdom: a cross-sectional analysis using the General Practice Research Database. BMJ Open 2012 Dec 14;2(6):10.1136/bmjopen-2012-001768. Print 2012.
(58) Rochon PA, Gurwitz JH, Sykora K, Mamdani M, Streiner DL, Garfinkel S, et al. Reader's guide to critical appraisal of cohort studies: 1. Role and design. BMJ 2005 Apr 16;330(7496):895-897.
(59) Melichar JK, Haida A, Rhodes C, Reynolds AH, Nutt DJ, Malizia AL. Venlafaxine occupation at the noradrenaline reuptake site: in-vivo determination in healthy volunteers. J Psychopharmacol 2001 Mar;15(1):9-12.
56
(60) Abdelmawla AH, Langley RW, Szabadi E, Bradshaw CM. Comparison of the effects of venlafaxine, desipramine, and paroxetine on noradrenaline- and methoxamine-evoked constriction of the dorsal hand vein. Br J Clin Pharmacol 1999 Sep;48(3):345-354.
(61) Siepmann T, Ziemssen T, Mueck-Weymann M, Kirch W, Siepmann M. The effects of venlafaxine on autonomic functions in healthy volunteers. J Clin Psychopharmacol 2007 Dec;27(6):687-691.
(62) Harvey AT, Rudolph RL, Preskorn SH. Evidence of the dual mechanisms of action of venlafaxine. Arch Gen Psychiatry 2000 May;57(5):503-509.
(63) Loprinzi CL, Kugler JW, Sloan JA, Mailliard JA, LaVasseur BI, Barton DL, et al. Venlafaxine in management of hot flashes in survivors of breast cancer: a randomised controlled trial. Lancet (London, England) 2000;356(9247):2059-2063.
(64) Ramasubbu R, Taylor VH, Saaman Z, Sockalingham S, Li M, Patten S, et al. The Canadian Network for Mood and Anxiety Treatments (CANMAT) task force recommendations for the management of patients with mood disorders and select comorbid medical conditions. Ann Clin Psychiatry 2012 Feb;24(1):91-109.
(65) Dworkin RH, O'Connor AB, Backonja M, Farrar JT, Finnerup NB, Jensen TS, et al. Pharmacologic management of neuropathic pain: evidence-based recommendations. Pain 2007;132(3):237-251.
(66) Bond DJ, Hadjipavlou G, Lam R, McIntyre R, Beaulieu S, Schaffer A, et al. The Canadian Network for Mood and Anxiety Treatments (CANMAT) task force recommendations for the management of patients with mood disorders and comorbid attention-deficit/hyperactivity disorder. Ann Clin Psychiatry 2012;24(1):23-37.
(67) Gelenberg AJ. Practice Guideline for the Treatment of Patients with Major Depressive Disorder: Approved in May 2010 and Published in October 2010. : American Psychiatric Association; 2010.
(68) Gormsen L, Rosenberg R, Bach FW, Jensen TS. Depression, anxiety, health‐related quality of life and pain in patients with chronic fibromyalgia and neuropathic pain. European Journal of Pain 2010;14(2):127. e1-127. e8.
(69) Rowbotham MC, Goli V, Kunz NR, Lei D. Venlafaxine extended release in the treatment of painful diabetic neuropathy: a double-blind, placebo-controlled study. Pain 2004;110(3):697.
(70) Frasure-Smith N, Lesperance F, Talajic M. Depression and 18-month prognosis after myocardial infarction. Circulation 1995 Feb 15;91(4):999-1005.
57
(71) Lichtman JH, Bigger Jr JT, Blumenthal JA, Frasure-Smith N, Kaufmann PG, Lespérance F, et al. Depression and Coronary Heart Disease Recommendations for Screening, Referral, and Treatment: A Science Advisory From the American Heart Association Prevention Committee of the Council on Cardiovascular Nursing, Council on Clinical Cardiology, Council on Epidemiology and Prevention, and Interdisciplinary Council on Quality of Care and Outcomes Research: Endorsed by the American Psychiatric Association. Circulation 2008;118(17):1768-1775.
(72) Grundy SM, Benjamin IJ, Burke GL, Chait A, Eckel RH, Howard BV, et al. Diabetes and cardiovascular disease: a statement for healthcare professionals from the American Heart Association. Circulation 1999;100(10):1134-1146.
(73) Muth EA, Moyer JA, Haskins JT, Andree TH, Husbands G. Biochemical, neurophysiological,
and behavioral effects of Wy‐45,233 and other identified metabolites of the antidepressant venlafaxine. Drug Dev Res 2004;23(2):191-199.
(74) Patat A, Troy S, Burke J, Trocherie S, Danjou P, Le Coz F, et al. Absolute bioavailability and electroencephalographic effects of conventional and extended-release formulations of venlafaxine in healthy subjects. The Journal of Clinical Pharmacology 1998;38(3):256-267.
(75) Lobello KW, Preskorn SH, Guico-Pabia CJ, Jiang Q, Paul J, Nichols AI, et al. Cytochrome P450 2D6 phenotype predicts antidepressant efficacy of venlafaxine: a secondary analysis of 4 studies in major depressive disorder. J Clin Psychiatry 2010 Nov;71(11):1482-1487.
(76) Whyte EM, Romkes M, Mulsant BH, Kirshne MA, Begley AE, Reynolds CF,3rd, et al. CYP2D6 genotype and venlafaxine-XR concentrations in depressed elderly. Int J Geriatr Psychiatry 2006 Jun;21(6):542-549.
(77) Lindh JD, Annas A, Meurling L, Dahl ML, AL-Shurbaji A. Effect of ketoconazole on venlafaxine plasma concentrations in extensive and poor metabolisers of debrisoquine. Eur J Clin Pharmacol 2003 Sep;59(5-6):401-406.
(78) Epstein M. Aging and the kidney. Journal of the American Society of Nephrology 1996;7(8):1106.
(79) Tatsumi M, Groshan K, Blakely RD, Richelson E. Pharmacological profile of antidepressants and related compounds at human monoamine transporters. Eur J Pharmacol 1997;340(2-3):249-258.
(80) Howell C, Wilson AD, Waring WS. Cardiovascular toxicity due to venlafaxine poisoning in adults: a review of 235 consecutive cases. Br J Clin Pharmacol 2007 Aug;64(2):192-197.
(81) Batista M, Dugernier T, Simon M, Haufroid V, Capron A, Fonseca S, et al. The spectrum of acute heart failure after venlafaxine overdose. Clin Toxicol (Phila) 2013 Feb;51(2):92-95.
58
(82) Khalifa M, Daleau P, Turgeon J. Mechanism of sodium channel block by venlafaxine in guinea pig ventricular myocytes. J Pharmacol Exp Ther 1999 Oct;291(1):280-284.
(83) Feighner JP. Cardiovascular safety in depressed patients: focus on venlafaxine. J Clin Psychiatry 1995 Dec;56(12):574-579.
(84) Preskorn S. Two in one: The venlafaxine story. Journal of Practical Psychiatry and Behavioral Health 1999;5(6):346-350.
(85) Blier P, Saint-Andre E, Hebert C, de Montigny C, Lavoie N, Debonnel G. Effects of different doses of venlafaxine on serotonin and norepinephrine reuptake in healthy volunteers. Int J Neuropsychopharmacol 2007 Feb;10(1):41-50.
(86) Sala M, Coppa F, Cappucciati C, Brambilla P, d'Allio G, Caverzasi E, et al. Antidepressants: their effects on cardiac channels, QT prolongation and Torsade de Pointes. Curr Opin Investig Drugs 2006 Mar;7(3):256-263.
(87) Mbaya P, Alam F, Ashim S, Bennett D. Cardiovascular effects of high dose venlafaxine XL in patients with major depressive disorder. Hum Psychopharmacol 2007 Apr;22(3):129-133.
(88) Isbister GK. Electrocardiogram changes and arrhythmias in venlafaxine overdose. Br J Clin Pharmacol 2009 May;67(5):572-576.
(89) Kok R, Nolen W, Heeren T. Cardiovascular changes associated with venlafaxine in the treatment of late-life depression. Am J Geriatr Psychiatry 2007 Aug;15(8):725; author reply 726.
(90) Martinez C, Assimes TL, Mines D, Dell'aniello S, Suissa S. Use of venlafaxine compared with other antidepressants and the risk of sudden cardiac death or near death: a nested case-control study. BMJ 2010 Feb 5;340:c249.
(91) Leonard CE, Bilker WB, Newcomb C, Kimmel SE, Hennessy S. Antidepressants and the risk of sudden cardiac death and ventricular arrhythmia. Pharmacoepidemiol Drug Saf 2011 Sep;20(9):903-913.
(92) Christoph M, Ebner B, Stolte D, Ibrahim K, Kolschmann S, Strasser RH, et al. Broken heart syndrome: Tako Tsubo cardiomyopathy associated with an overdose of the serotonin-norepinephrine reuptake inhibitor Venlafaxine. Eur Neuropsychopharmacol 2010 Aug;20(8):594-597.
(93) Danjou P, Hackett D. Safety and tolerance profile of venlafaxine. Int Clin Psychopharmacol 1995 Mar;10 Suppl 2:15-20.
59
(94) Charniot JC, Vignat N, Monsuez JJ, Kidouche R, Avramova B, Artigou JY, et al. Cardiogenic shock associated with reversible dilated cardiomyopathy during therapy with regular doses of venlafaxine. Am J Emerg Med 2010 Feb;28(2):256.e1-256.e5.
(95) Colucci VJ, Berry BD. Heart failure worsening and exacerbation after venlafaxine and duloxetine therapy. Ann Pharmacother 2008 Jun;42(6):882-887.
(96) Oslin DW, Ten Have TR, Streim JE, Datto CJ, Weintraub D, DiFilippo S, et al. Probing the safety of medications in the frail elderly: evidence from a randomized clinical trial of sertraline and venlafaxine in depressed nursing home residents. J Clin Psychiatry 2003 Aug;64(8):875-882.
(97) Ho JM, Gomes T, Straus SE, Austin PC, Mamdani M, Juurlink DN. Adverse cardiac events in older patients receiving venlafaxine: a population-based study. J Clin Psychiatry 2014 Jun;75(6):e552-8.
(98) Juurlink DN, Institute for Clinical Evaluative Sciences in Ontario. Canadian institute for health information discharge abstract database: a validation study. : Institute for Clinical Evaluative Sciences; 2006.
(99) Williams J, Young W. A summary of studies on the quality of health care administrative databases in Canada. Patterns of health care in Ontario: the ICES practice atlas.2nd ed.Ottawa: Canadian Medical Association 1996;339:45.
(100) Gill SS, Rochon PA, Herrmann N, Lee PE, Sykora K, Gunraj N, et al. Atypical antipsychotic drugs and risk of ischaemic stroke: population based retrospective cohort study. BMJ 2005;330(7489):445.
(101) Juurlink DN, Mamdani M, Kopp A, Laupacis A, Redelmeier DA. Drug-drug interactions among elderly patients hospitalized for drug toxicity. JAMA 2003 Apr 2;289(13):1652-1658.
(102) Hux JE, Fielding D. The Ontario Drug Benefit Program Copayment: Its Impact on Access for Ontario Seniors and Charges to the Program: a Background Document. : Institute for Clinical Evaluative Science in Ontario; 1999.
(103) CIHI. CIHI Data Quality Study of the 2009–2010 Discharge Abstract Database. 2012.
(104) The data quality study of the Canadian discharge abstract database. Proceedings of Statistics Canada Symposium: Citeseer; 2001.
(105) Ripple MG, Pestaner JP, Levine BS, Smialek JE. Lethal combination of tramadol and multiple drugs affecting serotonin. Am J Forensic Med Pathol 2000 Dec;21(4):370-374.
60
(106) Carver JR, Shapiro CL, Ng A, Jacobs L, Schwartz C, Virgo KS, et al. American Society of Clinical Oncology clinical evidence review on the ongoing care of adult cancer survivors: cardiac and pulmonary late effects. Journal of Clinical Oncology 2007;25(25):3991-4008.
(107) Tu JV, Austin PC, Filate WA, Johansen HL, Brien SE, Pilote L, et al. Outcomes of acute myocardial infarction in Canada. Can J Cardiol 2003;19(8):893-901.
(108) Austin PC, Daly PA, Tu JV. A multicenter study of the coding accuracy of hospital discharge administrative data for patients admitted to cardiac care units in Ontario. Am Heart J 2002 Aug;144(2):290-296.
(109) So L, Evans D, Quan H. ICD-10 coding algorithms for defining comorbidities of acute myocardial infarction. BMC Health Serv Res 2006 Dec 15;6:161.
(110) Varas-Lorenzo C, Castellsague J, Stang MR, Tomas L, Aguado J, Perez-Gutthann S. Positive predictive value of ICD-9 codes 410 and 411 in the identification of cases of acute coronary syndromes in the Saskatchewan Hospital automated database. Pharmacoepidemiol Drug Saf 2008 Aug;17(8):842-852.
(111) Tu K, Mitiku T, Guo H, Lee DS, Tu JV. Myocardial infarction and the validation of physician billing and hospitalization data using electronic medical records. Chronic Dis Can 2010 Sep;30(4):141-146.
(112) Schultz SE, Rothwell DM, Chen Z, Tu K. Identifying cases of congestive heart failure from administrative data: a validation study using primary care patient records. Chronic Dis Inj Can 2013 Jun;33(3):160-166.
(113) Van Walraven C, Mamdani M, Wells P, Williams J. Inhibition of serotonin reuptake by antidepressants and upper gastrointestinal bleeding in elderly patients: retrospective cohort study. BMJ 2001;323(7314):655.
(114) de Abajo FJ, Garcia-Rodriguez LA. Risk of upper gastrointestinal tract bleeding associated with selective serotonin reuptake inhibitors and venlafaxine therapy: interaction with nonsteroidal anti-inflammatory drugs and effect of acid-suppressing agents. Arch Gen Psychiatry 2008 Jul;65(7):795-803.
(115) Schneeweiss S, Seeger JD, Maclure M, Wang PS, Avorn J, Glynn RJ. Performance of comorbidity scores to control for confounding in epidemiologic studies using claims data. Am J Epidemiol 2001 Nov 1;154(9):854-864.
(116) Creditor MC. Hazards of hospitalization of the elderly. Ann Intern Med 1993 Feb 1;118(3):219-223.
61
(117) Austin PC, van Walraven C, Wodchis WP, Newman A, Anderson GM. Using the Johns Hopkins Aggregated Diagnosis Groups (ADGs) to predict mortality in a general adult population cohort in Ontario, Canada. Med Care 2011 Oct;49(10):932-939.
(118) Levy K, Lanctot KL, Farber SB, Li A, Herrmann N. Does pharmacological treatment of neuropsychiatric symptoms in Alzheimer's disease relieve caregiver burden? Drugs Aging 2012 Mar 1;29(3):167-179.
(119) Moschetti K, Cummings PL, Sorvillo F, Kuo T. Burden of Alzheimer's disease-related mortality in the United States, 1999-2008. J Am Geriatr Soc 2012 Aug;60(8):1509-1514.
(120) Khan A, Upton GV, Rudolph RL, Entsuah R, Leventer SM. The use of venlafaxine in the treatment of major depression and major depression associated with anxiety: a dose-response study. Venlafaxine Investigator Study Group. J Clin Psychopharmacol 1998 Feb;18(1):19-25.
(121) Pacala JT, Sullivan JT. Geriatrics Review Syllabus: A Core Curriculum in Geriatric Medicine. : American Geriatrics Society; 2010.
(122) FDA alert. FDA Drug Safety Communication: Revised recommendations for Celexa (citalopram hydrobromide) related to a potential risk of abnormal heart rhythms with high doses. March 28, 2012; Available at: http://www.fda.gov/Drugs/DrugSafety/ucm297391.htm. Accessed June/12, 2012.
(123) Lalit V, Appaya PM, Hegde RP, Mital AK, Mittal S, Nagpal R, et al. Escitalopram Versus Citalopram and Sertraline: A Double-Blind Controlled, Multi-centric Trial in Indian Patients with Unipolar Major Depression. Indian J Psychiatry 2004 Oct;46(4):333-341.
(124) Gomes T, Mamdani MM, Dhalla IA, Paterson JM, Juurlink DN. Opioid dose and drug-related mortality in patients with nonmalignant pain. Arch Intern Med 2011 Apr 11;171(7):686-691.
(125) Fischer B, Jones W, Rehm J. High correlations between levels of consumption and mortality related to strong prescription opioid analgesics in British Columbia and Ontario, 2005-2009. Pharmacoepidemiol Drug Saf 2013 Apr;22(4):438-442.
(126) Gill SS, Bronskill SE, Normand SL, Anderson GM, Sykora K, Lam K, et al. Antipsychotic drug use and mortality in older adults with dementia. Ann Intern Med 2007 Jun 5;146(11):775-786.
(127) Gill SS, Seitz D, Rochon PA. Atypical antipsychotic drugs, dementia, and risk of death. JAMA 2006 Feb 1;295(5):495-6; author reply 496-7.
(128) Quan H, Li B, Saunders LD, Parsons GA, Nilsson CI, Alibhai A, et al. Assessing validity of ICD-9-CM and ICD-10 administrative data in recording clinical conditions in a unique dually coded database. Health Serv Res 2008 Aug;43(4):1424-1441.
62
(129) Cummings E, Maher R, Showell CM, Croft T, Tolman J, Vickers J, et al. Hospital coding of dementia: is it accurate? HIM J 2011;40(3):5-11.
(130) St Germaine-Smith C, Metcalfe A, Pringsheim T, Roberts JI, Beck CA, Hemmelgarn BR, et al. Recommendations for optimal ICD codes to study neurologic conditions: a systematic review. Neurology 2012 Sep 4;79(10):1049-1055.
(131) van Deursen VM, Damman K, van der Meer P, Wijkstra PJ, Luijckx GJ, van Beek A, et al. Co-morbidities in heart failure. Heart Fail Rev 2012 Dec 25.
(132) Hunt SA, Abraham WT, Chin MH, Feldman AM, Francis GS, Ganiats TG, et al. 2009 Focused update incorporated into the ACC/AHA 2005 Guidelines for the Diagnosis and Management of Heart Failure in Adults A Report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines Developed in Collaboration With the International Society for Heart and Lung Transplantation. J Am Coll Cardiol 2009 Apr 14;53(15):e1-e90.
(133) Fihn SD, Gardin JM, Abrams J, Berra K, Blankenship JC, Dallas AP, et al. 2012 ACCF/AHA/ACP/AATS/PCNA/SCAI/STS Guideline for the Diagnosis and Management of Patients With Stable Ischemic Heart Disease: A Report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines, and the American College of Physicians, American Association for Thoracic Surgery, Preventive Cardiovascular Nurses Association, Society for Cardiovascular Angiography and Interventions, and Society of Thoracic Surgeons. J Am Coll Cardiol 2012 Dec 18;60(24):e44-e164.
(134) Mamdani M, Sykora K, Li P, Normand SLT, Streiner DL, Austin PC, et al. Reader's guide to critical appraisal of cohort studies: 2. Assessing potential for confounding. BMJ 2005;330(7497):960.
(135) Cohen J. Statistical power analysis for the behavioural sciences. Hillsdale, NJ: Academic Press; 1988.
(136) Austin PC. A Tutorial and Case Study in Propensity Score Analysis: An Application to Estimating the Effect of In-Hospital Smoking Cessation Counseling on Mortality. Multivariate Behav Res 2011;46(1):119-151.
(137) Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med 2009;28:3083-3107.
(138) Lee BK, Lessler J, Stuart EA. Weight trimming and propensity score weighting. PloS one 2011;6(3):e18174.
(139) Cole SR, Hernan MA. Constructing inverse probability weights for marginal structural models. Am J Epidemiol 2008 Sep 15;168(6):656-664.
63
(140) Peduzzi P, Concato J, Feinstein AR, Holford TR. Importance of events per independent variable in proportional hazards regression analysis II. Accuracy and precision of regression estimates. J Clin Epidemiol 1995;48(12):1503-1510.
(141) Lauzon C, Beck CA, Huynh T, Dion D, Racine N, Carignan S, et al. Depression and prognosis following hospital admission because of acute myocardial infarction. CMAJ 2003 Mar 4;168(5):547-552.
(142) Weeke P, Jensen A, Folke F, Gislason GH, Olesen JB, Andersson C, et al. Antidepressant use and risk of out-of-hospital cardiac arrest: a nationwide case-time-control study. Clin Pharmacol Ther 2012 Jul;92(1):72-79.
(143) Fisher LD, Lin DY. Time-dependent covariates in the Cox proportional-hazards regression model. Annu Rev Public Health 1999;20(1):145-157.
64
9 TABLES AND FIGURES
9.1 TABLES
9.1.1 Table 1a: Baseline Characteristics of Older Patients on Venlafaxine Compared to Sertraline
Characteristic Sertraline Venlafaxine Standardized Difference
N=41,238 N=48,876 Crude Weighted
Demographics
Age at start of cohort drug Median (IQR) 77 (71-82) 75 (70-81) 0.17 <0.001
66 to 75 yearsa 18,090 (43.9%) 25,197 (51.6%) 0.15 <0.001
76 to 85 yearsa 17,095 (41.5%) 17,828 (36.5%) 0.1 <0.001
86 years and overa 6,053 (14.7%) 5,851 (12.0%) 0.08 <0.001
Male 14,538 (35.3%) 17,373 (35.5%) 0.01 0.007
Number of medications (past 12 months)
Median (IQR) 8 (5-13) 9 (5-13) 0.02 0.001
Hospitalization (past 12 months) 12,406 (30.1%) 12,769 (26.1%) 0.09 <0.001
Drug Dosea,b
Low (38.0%) (35.0%)
Moderate (60.0%) (62.2%)
High (2.3%) (2.9%)
Comorbidities (past 36 months)
Heart failure 6,815 (16.5%) 6,538 (13.4%) 0.09 0.001
Cardiovascular disease 22,725 (55.1%) 23,938 (49.0%) 0.12 <0.001
Stroke 7,838 (19.0%) 7,844 (16.0%) 0.08 <0.001
Valvular disease 1,367 (3.3%) 1,203 (2.5%) 0.05 <0.001
Coronary artery disease 15,174 (36.8%) 16,416 (33.6%) 0.07 <0.001
Conduction disorder 4,931 (12.0%) 4,782 (9.8%) 0.07 <0.001
Renal disease 5,228 (12.7%) 5,750 (11.8%) 0.03 <0.001
Depression 6,055 (14.7%) 8,017 (16.4%) 0.05 <0.001
Anxiety 30,214 (73.3%) 37,027 (75.8%) 0.06 <0.001
Dementia 7,115 (17.3%) 8,730 (17.9%) 0.02 <0.001
Medications to treat cardiac disease (past 12 months)
Loop diuretics 7,797 (18.9%) 8,075 (16.5%) 0.06 0.001
ACE inhibitors/ARBs 18,299 (44.4%) 22,376 (45.8%) 0.03 <0.001
Negative chronotropic drugs 16,499 (40.0%) 18,151 (37.1%) 0.06 <0.001
Statins 12,411 (30.1%) 17,422 (35.6%) 0.12 0.001
Anti-arrhythmic drugs 1,133 (2.7%) 1,109 (2.3%) 0.03 <0.001
Anti-platelet drugs 8,913 (21.6%) 8,232 (16.8%) 0.12 <0.001
Warfarin 4,489 (10.9%) 5,050 (10.3%) 0.02 <0.001
Aldosterone antagonists 1,749 (4.2%) 1,677 (3.4%) 0.04 <0.001
Medications that might trigger cardiac disease (past 12 months)
65
Thiazolidinediones 232 (0.6%) 484 (1.0%) 0.05 0.002
Systemic steroids 3,648 (8.8%) 4,317 (8.8%) <0.001 <0.001
Non-steroidal anti-inflammatory drug 12,555 (30.4%) 14,886 (30.5%) <0.001 <0.001
Medications associated with neuropathic pain (past 12 months)
Opioids 11,917 (28.9%) 13,792 (28.2%) 0.02 <0.001
Gabapentin 200 (0.5%) 388 (0.8%) 0.04 <0.001
Medications associated with mortality (past 12 months)
Cholinesterase inhibitors 2,105 (5.1%) 3,233 (6.6%) 0.06 0.003
Antipsychotic drugs 3,499 (8.5%) 5,097 (10.4%) 0.07 0.002
Interquartile Range (IQR) Angiotensin Converting Enzyme (ACE), Angiotension II Receptor Blocker (ARB) Negative chronotropic drugs include beta blockers and non-dihydropyridine calcium channel blockers. a Due to rounding, proportions do not add up to 100% b Maximal dose reached during observation window
66
9.1.2 Table 1b: Baseline Characteristics of Older Patients on Venlafaxine Compared to Citalopram
Characteristic Citalopram Venlafaxine Standardized
Difference
N=3,909 N=48,876 Crude Weighted
Demographics
Age at start of cohort drug Median (IQR)
81 (74-87) 75 (70-81) 0.17 <0.001
66 to 75 yearsa 1,139 (29.1%) 25,197 (51.6%) 0.15 <0.001
76 to 85 yearsa 1,561 (39.9%) 17,828 (36.5%) 0.1 <0.001
86 years and overa 1,209 (30.9%) 5,851 (12.0%) 0.08 <0.001
Male 1,220 (31.2%) 17,373 (35.5%) 0.01 0.007
Number of medications (past 12 months)
Median (IQR)
11 (8-16) 9 (5-13) 0.02 0.001
Hospitalization (past 12 months) 1,140 (29.2%)
12,769 (26.1%) 0.09 <0.001
Drug Dosea,b
Low (24.0%) (35.0%)
Moderate (64.0%) (62.2%)
High (12.0%) (2.9%)
Comorbidities (past 36 months)
Heart failure 720 (18.4%) 6,538 (13.4%) 0.09 0.001
Cardiovascular disease 2,061 (52.7%) 23,938 (49.0%) 0.12 <0.001
Stroke 912 (23.3%) 7,844 (16.0%) 0.08 <0.001
Valvular disease 90 (2.3%) 1,203 (2.5%) 0.05 <0.001
Coronary artery disease 1,297 (33.2%) 16,416 (33.6%) 0.07 <0.001
Conduction disorder 472 (12.1%) 4,782 (9.8%) 0.07 <0.001
Renal disease 728 (18.6%) 5,750 (11.8%) 0.03 <0.001
Depression 687 (17.6%) 8,017 (16.4%) 0.05 <0.001
Anxiety 2,765 (70.7%) 37,027 (75.8%) 0.06 <0.001
Dementia 1,897 (48.5%) 8,730 (17.9%) 0.02 <0.001
Medications to treat cardiac disease (past 12 months)
Loop diuretics 1,051 (26.9%) 8,075 (16.5%) 0.06 0.001
ACE inhibitors/ARBs 2,034 (52.0%) 22,376 (45.8%) 0.03 <0.001
Negative chronotropic drugs 1,625 (41.6%) 18,151 (37.1%) 0.06 <0.001
67
Statins 1,756 (44.9%) 17,422 (35.6%) 0.12 0.001
Anti-arrhythmic drugs 79 (2.0%) 1,109 (2.3%) 0.03 <0.001
Anti-platelet drugs 640 (16.4%) 8,232 (16.8%) 0.12 <0.001
Warfarin 592 (15.1%) 5,050 (10.3%) 0.02 <0.001
Aldosterone antagonists 175 (4.5%) 1,677 (3.4%) 0.04 <0.001
Medications that might trigger cardiac disease (past 12 months)
Thiazolidinediones 79 (2.0%) 484 (1.0%) 0.05 0.002
Systemic steroids 319(8.2%) 4,317 (8.8%) <0.001 <0.001
Non-steroidal anti-inflammatory drug
722 (18.5%)
14,886 (30.5%) <0.001 <0.001
Medications associated with neuropathic pain (past 12 months)
Opioids 926 (23.7%) 13,792 (28.2%) 0.02 <0.001
Gabapentin 63 (1.6%) 388 (0.8%) 0.04 <0.001
Medications associated with mortality (past 12 months)
Cholinesterase inhibitors 896 (22.9%) 3,233 (6.6%) 0.06 0.003
Antipsychotic drugs 927 (23.7%) 5,097 (10.4%) 0.07 0.002
Interquartile Range (IQR) Angiotensin Converting Enzyme (ACE), Angiotension II Receptor Blocker (ARB) Negative chronotropic drugs include beta blockers and non-dihydropyridine calcium channel blockers. a Due to rounding, proportions do not add up to 100% b Maximal dose reached during observation window
68
9.1.3 Table 2a: Adverse Cardiac Events in Older Patients on Venlafaxine Compared to Sertraline (N=90,114)
Outcome
Events in Venlafaxine
Patients N (%)
Events in Sertraline Patients
N (%)
Weighted Events in
Venlafaxine N (%)
Weighted Events in Sertraline
N (%) Hazard Ratio
(95% CI)
Composite 3966 (8.1) 3707 (9.0) 4259 (8.7) 3459 (8.4) 0.97 (0.93 to
1.02) Acute Myocardial Infarction 430 (0.9) 404 (1.0) 447 (0.9) 385 (0.9)
0.91 (0.80 to 1.05)
Congestive Heart Failure 986 (2.0) 1109 (2.7) 1094 (2.2) 995 (2.4) 0.87 (0.80 to
0.95)
Death 3098 (6.3) 2770 (6.7) 3325(6.8) 2602 (6.3) 1.01 (0.96 to
1.06) Gastrointestinal Hemorrhage 333 (0.7) 298 (0.7) 353 (0.7) 282 (0.7)
0.99 (0.84 to 1.16)
Confidence Interval (CI)
9.1.4 Table 2b: Adverse Cardiac Events in Older Patients on Venlafaxine Compared to Citalopram (N=52,785)
Outcome
Events in Venlafaxine
Patients N (%)
Events in Citalopram
Patients N (%)
Weighted Events in
Venlafaxine N (%)
Weighted Events in
Citalopram N (%)
Hazard Ratio (95% CI)
Composite 3966 (8.1) 456 (11.7) 4259 (8.7) 294 (7.5) 1.39 (1.19 to
1.62) Acute Myocardial Infarction 430 (0.9) 38 (1.0) 447 (0.9) 27 (<0.1)
1.59 (1.5 to 2.41)
Congestive Heart Failure 986 (2.0) 88 (2.2) 1094 (2.2) 80 (2.0) 1.22 (0.87 to
1.72)
Death 3098 (6.3) 371 (9.5) 3325(6.8) 216 (5.5) 1.49 (1.26 to
1.76) Gastrointestinal Hemorrhage 333 (0.7) 32 (0.8) 353 (0.7) 27 (<0.1)
1.27 (0.82 to 1.96)
Confidence Interval (CI)
69
9.1.5 Table 3: Exploratory Time-Dependent Dose Analysis of Older Patients on Venlafaxine Compared to Sertraline
Variable Crude Hazard Ratio (95% Confidence Interval)
Adjusted Hazard Ratioa
(95% Confidence Interval) P value
Venlafaxineb 0.84 (0.80 to 0.87) 1.00 (0.96 to 1.05) 0.89
Moderate dosec 0.90 (0.85 to 0.94) 1.06 (1.01 to 1.12) 0.01
High dosec 0.98 (0.94 to 1.06) 1.41 (1.25 to 1.59) <0.001 aAdjusted for age, female sex, hospitalization in the past year, stroke, heart failure, cardiovascular disease, use of statins and antiplatelets, and the aggregated diagnosis groups, limited minor and stable but persistent psychosocial disease.
bSertraline is reference cLow dose is reference
70
9.2 FIGURES
9.2.1 Figure 1: Boxplot Distribution of Inverse Probability of Treatment Weights for Venlafaxine to Sertraline and Venlafaxine to Citalopram Comparisons
0
50
100
150
200
250
300
350
Inv
ers
e P
rob
ab
ilit
y o
f T
rea
tme
nt
We
igh
ts
Venlafaxine Drug Comparisons
Boxplot Distribution of Inverse Probability of Treatment Weights
Sertraline
Citalopram
71
9.2.2 Figure 2a: Crude and Weighted Survival Curves for Adverse Cardiac Events among Individuals on Venlafaxine Compared to Sertraline
0
0.05
0.1
0.15
0.2
1 91 181 271 361
Pro
po
rtio
n w
ith
Co
mp
osi
te O
utc
om
e
Time from Start of Therapy (Days)
Crude Survival Curve for Primary Outcome from Start of Therapy with Venlafaxine or Sertraline
Sertraline
Venlafaxine
0
0.05
0.1
0.15
0.2
1 91 181 271 361
Pro
po
rtio
n w
ith
Co
mp
osi
te o
utc
om
e
Time from start of therapy (days)
Weighted Survival Curve of Primary Outcome from Start of Therapy with Sertraline or Venlafaxine
Sertraline
Venlafaxine
72
9.2.3 Figure 2b: Crude and Weighted Survival Curves for Adverse Cardiac Events among Individuals on Venlafaxine Compared to Citalopram
0
0.05
0.1
0.15
0.2
1 91 181 271 361
Pro
bab
ility
of
Co
mp
osi
te O
utc
om
e
Time from Start of Therapy (Days)
Crude Survival Curve for Primary Outcome from Start of Therapy with Venlafaxine or Citalopram
Venlafaxine
Citalopram
0
0.05
0.1
0.15
0.2
1 91 181 271 361
Pro
bab
ility
of
Co
mp
osi
te O
utc
om
e
Time from Start of Therapy (Days)
Weighted Survival Curve for Primary Outcome from Start of Therapy with Venlafaxine or Citalopram
Citalopram
Venlafaxine
73
9.2.4 Figure 3a: Adverse Cardiac Events in Older Patients without Cardiovascular Disease on Venlafaxine Compared to Sertraline (N=43,451)
9.2.5 Figure 3b: Adverse Cardiac Events in Older Patients with Cardiovascular Disease on Venlafaxine Compared to Sertraline (N=46,663)
0.5 1 2Hazard Ratio Higher Risk with Lower Risk with
Weighted Hazard Ratio
Composite 0.94 (0.86 to 1.03)
Acute Myocardial 1.01 (0.76 to 1.34) Infarction
Congestive Heart 0.88 (0.71 to 1.10)
Failure
Death 0.93 (0.85 to 1.02)
Gastrointestinal 0.80 (0.60 to 1.07)
Hemorrhage
0.5 1 2Hazard Ratio Higher Risk with VenlafaxineLower Risk with Venlafaxine
Composite 0.99 (0.94 to 1.04)
Acute Myocardial 0.89 (0.76 to 1.05) Infarction
Congestive Heart 0.87 (0.80 to 0.96)
Death 1.05 (0.98 to 1.11)
Gastrointestinal 1.10 (0.90 to 1.33)
Weighted Hazard Ratio (95% CI)
74
9.2.6 Figure 4a: Adverse Cardiac Events in Older Patients without Cardiovascular Disease on Venlafaxine Compared to Citalopram (N=26,786)
9.2.7 Figure 4b: Adverse Cardiac Events of Older Patients with Cardiovascular Disease on Venlafaxine Compared to Citalopram (N=25,999)
Composite
Acute Myocardial Infarction
Congestive Heart Failure
Death
Gastrointestinal Hemorrhage
0.5 1 2 4 8
2.75 (1.06 to 7.12)
1.39 (1.09 to 1.77)
1.54 (0.67 to 3.53)
1.33 (1.03 to 1.72)
0.96 (0.49 to 1.86)
Lower RIsk with Higher Risk with VenlafaxineHazard
Composite
Acute Myocardial Infarction
Congestive Heart Failure
Death
Gastrointestinal Hemorrhage
0.5 1 2 4
Weighted Hazard Ratio (95% CI)
1.34 (1.08 to 1.65)
1.24 (0.77 to 2.02)
1.07 (0.71 to 1.61)
1.56 (1.24 to 1.96)
1.53 (0.84 to 2.78)
Lower RIsk with Higher RIsk with VenlafaxineHazard
Weighted Hazard Ratio (95% CI)
75
10 APPENDICES
10.1.1 APPENDIX 1a: Aggregated Diagnosis Groups of Older Individuals on Venlafaxine Compared to Sertraline
76
Characteristic Sertraline Venlafaxine Standardized Difference
N=41,238 N=48,876 Crude Weighted
Aggregated Diagnosis Groups
Time Limited
Minor 18,315 (44.4%) 22,435 (45.9%) 0.03 0.001
Minor-Primary Infections 28,361 (68.8%) 33,464 (68.5%) 0.01 <0.001
Major 10,191 (24.7%) 11,094 (22.7%) 0.05 <0.001
Major-Primary Infections 7,710 (18.7%) 8,677 (17.8%) 0.02 0.001
Allergies 4,017 (9.7%) 4,713 (9.6%) 0 0.001
Asthma 4,826 (11.7%) 5,393 (11.0%) 0.02 <0.001
Likely to Recur
Discrete 25,095 (60.9%) 29,848 (61.1%) 0 <0.001
Discrete-Infections 15,601 (37.8%) 18,381 (37.6%) 0 <0.001
Progressive 9,931 (24.1%) 10,003 (20.5%) 0.09 <0.001
Chronic Medical
Stable 36,730 (89.1%) 43,573 (89.2%) 0 0.001
Unstable 28,371 (68.8%) 32,026 (65.5%) 0.07 <0.001
Chronic Specialty
Stable-Orthopedic 2,053 (5.0%) 2,272 (4.6%) 0.02 <0.001
Stable-Ear,Nose,Throat 4,052 (9.8%) 4,558 (9.3%) 0.02 0.001
Stable-Eye 15,148 (36.7%) 16,721 (34.2%) 0.05 <0.001
Unstable-Orthopedic 2,921 (7.1%) 3,644 (7.5%) 0.01 <0.001
Unstable-Ear,Nose,Throat 1,112 (2.7%) 1,165 (2.4%) 0.02 <0.001
Unstable-Eye 8,005 (19.4%) 9,171 (18.8%) 0.02 <0.001
Dermatologic 9,659 (23.4%) 11,902 (24.4%) 0.02 <0.001
Injuries or Adverse Effects
Minor 14,103 (34.2%) 16,694 (34.2%) 0 <0.001
Major 14,939 (36.2%) 17,518 (35.8%) 0.01 <0.001
Psychosocial
Time Limited, Minor 4,354 (10.6%) 4,890 (10.0%) 0.02 <0.001
Stable, Recurrent or Persistent 25,155 (61.0%) 32,589 (66.7%) 0.12 0.001
Unstable, Recurrent or Persistent 9,327 (22.6%) 11,486 (23.5%) 0.02 0.001
Signs or Symptoms
Minor 27,612 (67.0%) 32,471 (66.4%) 0.01 <0.001
Uncertain 34,365 (83.3%) 40,700 (83.3%) 0 <0.001
Major 24,808 (60.2%) 28,952 (59.2%) 0.02 0.001
Discretionary 14,760 (35.8%) 17,376 (35.6%) 0.01 0.001
See and Reassure 1,300 (3.2%) 1,343 (2.7%) 0.02 <0.001
Prevention or Administrative 20,534 (49.8%) 24,515 (50.2%) 0.01 0.001
77
Malignancy 10,128 (24.6%) 12,527 (25.6%) 0.02 0.001
Pregnancy 213 (0.5%) 210 (0.4%) 0.01 0.001
Dental 891 (2.2%) 1,049 (2.1%) 0 <0.001
78
10.1.2 APPENDIX 1b: Aggregated Diagnosis Groups of Older Individuals on Venlafaxine Compared to Citalopram
79
Characteristic Citalopram Venlafaxine Standardized Difference
N=3,909 N=48,876 Crude Weighted
Aggregated Diagnosis Groups
Time Limited
Minor 1,749 (44.7%) 22,435 (45.9%) 0.02 0.001
Minor-Primary Infections 2,499 (63.9%) 33,464 (68.5%) 0.10 <0.001
Major 956 (24.5%) 11,094 (22.7%) 0.04 <0.001
Major-Primary Infections 829 (21.2%) 8,677 (17.8%) 0.09 0.001
Allergies 231 (5.9%) 4,713 (9.6%) 0.13 0.001
Asthma 292 (7.5%) 5,393 (11.0%) 0.12 <0.001
Likely to Recur
Discrete 2,224 (56.9%) 29,848 (61.1%) 0.09 <0.001
Discrete-Infections 1,574 (40.3%) 18,381 (37.6%) 0.05 <0.001
Progressive 1,139 (29.1%) 10,003 (20.5%) 0.21 <0.001
Chronic Medical
Stable 3,352 (85.8%) 43,573 (89.2%) 0.11 0.001
Unstable 2,754 (70.5%) 32,026 (65.5%) 0.10 <0.001
Chronic Specialty
Stable-Orthopedic 175 (4.5%) 2,272 (4.6%) 0.01 <0.001
Stable-Ear,Nose,Throat 294 (7.5%) 4,558 (9.3%) 0.06 0.001
Stable-Eye 1,081 (27.7%) 16,721 (34.2%) 0.14 <0.001
Unstable-Orthopedic 198 (5.1%) 3,644 (7.5%) 0.09 <0.001
Unstable-Ear,Nose,Throat 68 (1.7%) 1,165 (2.4%) 0.04 <0.001
Unstable-Eye 722 (18.5%) 9,171 (18.8%) 0.01 <0.001
Dermatologic 764 (19.5%) 11,902 (24.4%) 0.11 <0.001
Injuries or Adverse Effects
Minor 1,452 (37.1%) 16,694 (34.2%) 0.06 <0.001
Major 1,656 (42.4%) 17,518 (35.8%) 0.14 <0.001
Psychosocial
Time Limited, Minor 392 (10.0%) 4,890 (10.0%) <0.001 <0.001
Stable, Recurrent or Persistent 2,389 (61.1%) 32,589 (66.7%) 0.12 0.001
Unstable, Recurrent or Persistent 2,001 (51.2%) 11,486 (23.5%) 0.64 0.001
Signs or Symptoms
Minor 2,511 (64.2%) 32,471 (66.4%) 0.05 <0.001
Uncertain 3,258 (83.3%) 40,700 (83.3%) <0.001 <0.001
Major 2,356 (60.3%) 28,952 (59.2%) 0.02 0.001
Discretionary 1,117 (28.6%) 17,376 (35.6%) 0.15 0.001
See and Reassure 92 (2.4%) 1,343 (2.7%) 0.02 <0.001
Prevention or Administrative 2,108 (53.9%) 24,515 (50.2%) 0.08 0.001
Malignancy 913 (23.4%) 12,527 (25.6%) 0.05 0.001
Pregnancy 21 (0.5%) 210 (0.4%) 0.02 0.001
80
Dental 99 (2.5%) 1,049 (2.1%) 0.03 <0.001