Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Comparative safety of smoking cessation pharmacotherapies Carney, Gregory 2019

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2019_september_carney_greg.pdf [ 3.27MB ]
Metadata
JSON: 24-1.0380464.json
JSON-LD: 24-1.0380464-ld.json
RDF/XML (Pretty): 24-1.0380464-rdf.xml
RDF/JSON: 24-1.0380464-rdf.json
Turtle: 24-1.0380464-turtle.txt
N-Triples: 24-1.0380464-rdf-ntriples.txt
Original Record: 24-1.0380464-source.json
Full Text
24-1.0380464-fulltext.txt
Citation
24-1.0380464.ris

Full Text

Comparative Safety of Smoking Cessation Pharmacotherapies by  Gregory Carney  B.Sc., The University of Victoria, 2000  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Pharmacology and Therapeutics)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  August 2019  © Gregory Carney, 2019    ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled: Comparative Safety of Smoking Cessation Pharmacotherapies  submitted by Gregory Carney in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Pharmacology and Therapeutics  Examining Committee: Dr. Colin Dormuth Supervisor  Dr. Ken Bassett Supervisory Committee Member  Dr. Malcolm Maclure Supervisory Committee Member Dr. Michael Law University Examiner Dr. Arminee Kazanjian University Examiner  Additional Supervisory Committee Members: Dr. Suzanne Malfair Supervisory Committee Member  Supervisory Committee Member iii  Abstract Background: Smoking is the leading cause of preventable death worldwide. Although smoking cessation (SC) pharmacotherapies have been shown to have a moderate short-term benefit as a quitting aid, substantial cardiovascular and neuropsychiatric safety concerns have been identified in adverse reporting databases, leading to prescription label warnings by Health Canada and the U.S. FDA. However, recent studies indicate these warnings may be without merit.  Objectives: This thesis examined the comparative safety and effectiveness of medications commonly used to aid smoking cessation. The thesis focused on real-world safety of varenicline, bupropion, and nicotine replacement therapy (NRT). It also demonstrated a novel method of measuring comparative effectiveness using drug therapy re-initiation as a proxy for treatment failure.   Design and Setting: Retrospective cohort studies using linked de-identified claims data from the British Columbia Ministry of Health and U.S. MarketScan® databases.   Outcomes: The primary safety outcome was a composite of cardiovascular hospitalizations. Secondary safety outcomes included: all-cause mortality, a composite of neuropsychiatric hospitalizations, and individual components of the primary outcome. Effectiveness was measured using drug therapy re-initiation.   Statistical Analysis: Propensity score adjusted log-binomial and Cox proportional hazards regression models. iv  Results: 116,442 B.C. participants and 618,497 U.S. participants were analyzed. In the U.S., compared to NRT, varenicline was associated with a 20% lower 1-year CV risk [adjusted risk ratio (RR) = 0.80, 95% Confidence Interval (CI): (0.75 – 0.85)] and bupropion was associated with a 25% lower 1-year CV risk [RR=0.75, 95% CI: (0.69 – 0.81)]. Varenicline was associated with a 26% lower 1-year risk of neuropsychiatric hospitalization versus NRT [RR=0.74, 95% CI: (0.71 – 0.76)]. In B.C., compared to NRT, varenicline was associated with a 20% one-year relative risk decrease of neuropsychiatric hospitalization [RR: 0.80, CI: (0.71 – 0.89)], and a 19% one-year relative risk decrease of mortality [RR: 0.81, CI: (0.71 – 0.93)]. We found no significant difference in risk between NRT and bupropion for cardiovascular hospitalizations, neuropsychiatric hospitalizations, or mortality.  Conclusions: Compared to NRT, varenicline was associated with fewer serious adverse events, and bupropion the same number of serious adverse events. Varenicline and bupropion were associated with fewer subsequent SC drug episodes.  v  Lay Summary  This thesis examined the comparative safety and effectiveness of medications commonly used to aid smoking cessation: varenicline, bupropion, and nicotine replacement therapy (NRT). Prescribing of varenicline and bupropion have been hindered by safety concerns, including government issued product warnings. The objective of our study was to assess the difference in cardiovascular and neuropsychiatric hospitalizations, comparing NRT with the prescription smoking cessation medications using large health claims databases, and to assess the comparative effectiveness in helping people abstain from smoking. Our study found no increased safety risk associated with varenicline or bupropion compared to NRT, and some evidence supported a safety benefit with varenicline. Our study also produced evidence that varenicline and bupropion may be associated with fewer patients requiring subsequent courses of therapy, indicating greater effectiveness as a smoking cessation aid. The results of this thesis will aid physicians, patients, and policy-makers to make informed choices regarding smoking cessation pharmacotherapy.  vi  Preface  This thesis is the original unpublished work of Greg Carney.  We obtained ethics approval for this study from the Clinical Research Ethics Board, University of British Columbia (certificate number: H15-01700).  Disclaimer: All inferences, opinions, and conclusions drawn in this manuscript are those of the authors, and do not reflect the opinions or policies of the Data Stewards.      vii  Table of Contents Abstract ......................................................................................................................................... iii  Lay Summary .................................................................................................................................v  Preface ........................................................................................................................................... vi  Table of Contents ........................................................................................................................ vii  List of Tables ................................................................................................................................ xi  List of Figures ............................................................................................................................. xiii  List of Abbreviations ................................................................................................................. xiv  Introduction ....................................................................................................................................1  1.1 Pharmacology ................................................................................................................. 2 1.1.1 The Nicotine Addiction Cycle .................................................................................... 2 1.1.2 Varenicline .................................................................................................................. 2 1.1.3 Bupropion ................................................................................................................... 3  1.1.4 Nicotine Therapy ........................................................................................................ 4 1.2 Overview of Thesis Chapters .......................................................................................... 5  1.3 Data Sources ................................................................................................................... 6  1.3.1 Patient Registration and Demographics ...................................................................... 6 1.3.2 Hospital Discharge Abstracts ...................................................................................... 6 1.3.3 Physician Visits ........................................................................................................... 7  1.3.4 Dispensed Medications ............................................................................................... 7 1.3.5 Vital Statistics Deaths (British Columbia Only) ......................................................... 8 1.4 Statistical Model Selection ............................................................................................. 8 1.4.1 Log-Binomial Regression ........................................................................................... 9 1.4.2 Cox Proportional Hazards ......................................................................................... 10 1.4.3 Propensity Scores ...................................................................................................... 11 viii  Chapter 2: Cardiovascular and Neuropsychiatric Safety of Smoking Cessation Pharmacotherapies in Non-Depressed Adults: A Retrospective Cohort Study. ....................13 2.1 Introduction ................................................................................................................... 13  2.2 Methods......................................................................................................................... 16  2.2.1 Data ........................................................................................................................... 16 2.2.2 Study Design and Patients ........................................................................................ 16 2.2.3 Study outcomes ......................................................................................................... 17 2.2.4 Patient Follow-up ...................................................................................................... 18 2.2.5 Patient Characteristics and Confounders .................................................................. 18 2.2.6 Statistical Analysis .................................................................................................... 19 2.3 Results ........................................................................................................................... 19 2.3.1 Study cohort and patient characteristics ................................................................... 19 2.3.2 Relative risk of primary and secondary outcomes .................................................... 20 2.3.3 Sensitivity and subgroup analyses ............................................................................ 21 2.4 Discussion ..................................................................................................................... 22  2.4.1 Comparison with existing literature .......................................................................... 22 2.4.2 Interpretation of Results ............................................................................................ 23  2.4.3 Limitations ................................................................................................................ 24 2.5 Conclusion .................................................................................................................... 25 Chapter 3: Cardiovascular and Neuropsychiatric Safety of Smoking Cessation Pharmacotherapies: A Population-Based Retrospective Cohort Study. ................................34 3.1 Introduction ................................................................................................................... 34  3.2 Methods......................................................................................................................... 35  3.2.1 Data ........................................................................................................................... 35 3.2.2 Study population and drug exposure......................................................................... 36 3.2.3 Outcome definition ................................................................................................... 36  3.2.4 Patient Follow-up ...................................................................................................... 37 ix  3.2.5 Covariate assessment ................................................................................................ 37 3.2.6 Statistical Analysis .................................................................................................... 38 3.2.7 Sensitivity Analysis: No recent history of cardiovascular events ............................. 38 3.3 Results ........................................................................................................................... 39 3.3.1 Study cohort and patient characteristics ................................................................... 39 3.3.2 Relative risk (risk ratio) of primary and secondary outcomes .................................. 39 3.3.3 Sensitivity analyses ................................................................................................... 40  3.4 Discussion ..................................................................................................................... 41  3.4.1 Comparison with existing literature .......................................................................... 41 3.4.2 Interpretation of Results ............................................................................................ 42  3.4.3 Study strengths .......................................................................................................... 43 3.4.4 Limitations ................................................................................................................ 43 3.5 Conclusion .................................................................................................................... 44 Chapter 4: Identifying Sequential Episodes of Pharmacotherapy as a Method for Assessing Treatment Failure in Comparative Effectiveness Research ....................................................52 4.1 Introduction ................................................................................................................... 52  4.2 Methods......................................................................................................................... 53  4.2.1 Data Collection ......................................................................................................... 53 4.2.2 Population ................................................................................................................. 53  4.2.3 Study Outcomes and Cohort Follow-up ................................................................... 54 4.2.4 Statistical Analysis .................................................................................................... 54 4.2.5 High-Dimensional Propensity Scores ....................................................................... 55 4.3 Results ........................................................................................................................... 55 4.3.1 Study cohort and patient characteristics ................................................................... 55 4.3.2 Relative risk of re-starting smoking cessation therapy ............................................. 56 4.3.3 Predictors of Re-Starting Therapy ............................................................................ 57 4.3.4 Sensitivity analyses ................................................................................................... 58  4.4 Discussion ..................................................................................................................... 58  x  4.4.1 Interpretation of Results ............................................................................................ 58  4.4.2 Comparison with existing literature .......................................................................... 59 4.4.3 Limitations ................................................................................................................ 61 4.5 Conclusion .................................................................................................................... 62 Chapter 5: Concluding Chapter .................................................................................................68  5.1 Varenicline was not associated with an increased risk of hospitalization for a cardiovascular event compared to nicotine replacement therapies (NRT). .............................. 68 5.2 Varenicline was associated with a decreased risk of hospitalization for a neuropsychiatric event compared to nicotine replacement therapies (NRT). ........................... 69 5.3 There was no risk difference in cardiovascular or neuropsychiatric events between bupropion and NRT users. ........................................................................................................ 70  5.4 Bupropion users compared to varenicline users were associated with a lower risk of cardiovascular events, and a higher risk of neuropsychiatric events ........................................ 71 5.5 Drug episode sequencing can be a useful method of identifying treatment failure in comparative effectiveness research. .......................................................................................... 72  5.6 Future Research ............................................................................................................ 73  References  ................................................................................................................................... 76 Appendix A - Exclusion Criteria; Diagnosis or Treatment for Depression ...........................85 Appendix B - Case Definitions for Baseline Characteristics ....................................................86 Appendix C - Propensity Score Summary Statistics and Histograms ....................................87 Appendix D - Case Definitions for Baseline Characteristics ...................................................91 Appendix E - Case Definitions for Baseline Characteristics ....................................................92 Appendix F - History of Cardiovascular Disease ......................................................................93   xi  List of Tables  Table 2.5.1 Baseline Characteristics of Study Cohorts................................................................. 28  Table 2.5.2 Risk of Admission to Hospital for Cardiovascular (CVD) Events; Log-Binomial Regression with a One-Year Fixed Follow-up ............................................................................. 29 Table 2.5.3 Risk of Admission to Hospital for Neuropsychiatric Events; Log-Binomial Regression with a One-Year Fixed Follow-up ............................................................................. 30 Table 2.5.4 Sensitivity Analysis: Incidence Rates and Hazard Ratios of Admission to Hospital for Cardiovascular Events; Cox Proportional Hazards ................................................................. 31  Table 2.5.5 Sensitivity Analysis: Incidence Rates and Hazard Ratios of Admission to Hospital for Neuropsychiatric Events; Cox Proportional Hazards.............................................................. 32 Table 2.5.6 Sensitivity Analysis: Risk of Admission to Hospital for Cardiovascular (CVD) Events; Log-Binomial Regression, 5% Trimmed Propensity Scores ........................................... 32 Table 2.5.7 Direct Comparison of Bupropion versus Varenicline Users; Log-Binomial Regression, Risk of Cardiovascular Hospitalization, Neuropsychiatric Hospitalization. ............ 33 Table 3.5.1 Baseline Characteristics of Study Cohorts................................................................. 46  Table 3.5.2 Risk Ratios of Cardiovascular Events During 1-year of Follow-up .......................... 47 Table 3.5.3 Risk Ratios of Cardiovascular Events During 6-Months of Follow-up .................... 48 Table 3.5.4 Risk Ratios of Neuropsychiatric Events .................................................................... 49 Table 3.5.5 Risk Ratios of mortality at 6-months and 1-year ....................................................... 49 Table 3.5.6 Sensitivity Analysis: Risk ratios of cardiovascular events in patients with no cardiovascular events in 6-months prior to cohort entry .............................................................. 50  xii  Table 3.5.7 Sensitivity Analysis: Risk Ratios of Mortality in Patients with No Cardiovascular Events in 6-Months Prior to Cohort Entry .................................................................................... 50  Table 3.5.8 Direct Comparison of Bupropion versus Varenicline; Log-binomial Regression, Risk of Cardiovascular Hospitalization, Neuropsychiatric Hospitalization, Mortality ........................ 51 Table 4.5.1 Baseline Characteristics of Study Cohorts................................................................. 64  Table 4.5.2 Risk of Treatment Failure: Re-Starting Smoking Cessation Therapy ....................... 65 Table 4.5.3 Predictors of Treatment Failure in British Columbia ................................................ 66  Table 4.5.4 Sensitivity Analysis: Re-Starting Smoking Cessation Therapy (No Washout Period)....................................................................................................................................................... 67  Table 4.5.5 Medication Matrix of Patients Who Re-Start Therapy .............................................. 67 Table 5.6.1 Risk of Cardiovascular Events; Combined British Columbia and US MarketScan Inverse Variance Weighted Average, Varenicline versus NRT ................................................... 75 Table 5.6.2 Risk of Cardiovascular Events; Combined British Columbia and US MarketScan, Inverse Variance Weighted Average, Bupropion versus Varenicline .......................................... 75 Table 5.6.3 Risk of Neuropsychiatric Events; Combined British Columbia and US MarketScan, Inverse Variance Weighted Average, Bupropion versus Varenicline .......................................... 75   xiii  List of Figures  Figure 1 The Nicotine Addiction Cycle ........................................................................................ 12  Figure 2 Patient Flow Diagram ..................................................................................................... 27 Figure 3 British Columbia Patient Flow Diagram ........................................................................ 45  Figure 4 British Columbia Study Period and Follow-up .............................................................. 63 Figure 5 British Columbia Patient Flow Diagram ........................................................................ 63   xiv  List of Abbreviations  ACE: Angiotensin-converting enzyme inhibitor AMI: Acute myocardial infarction BB: Beta-adrenergic blocking agents BC: British Columbia CHF: Congestive heart failure CI: Confidence Interval CMAX: peak concentration COPD: Chronic Obstructive Pulmonary Disease CPT-4: Current Procedural Terminology, Fourth Revision CV: Cardiovascular CVD: Cardiovascular disease DAD: Discharge Abstract Database FDA: United States Food and Drug Administration HDPS: High-Dimensional Propensity Score HR: Hazard Ratio ICD-9: International Classification of Disease, Ninth Revision ICD-10: International Classification of Diseases, Tenth Revision IHD: Ischemic heart disease MSP: British Columbia Medical Services Plan N: Number of participants NRT: Nicotine replacement therapy xv  NSAID: Nonsteroidal anti-inflammatory drugs PPI: Proton-pump inhibitor PS: Propensity Score RR: Risk ratio (relative risk) SC: Smoking cessation US: United States of America                1  Introduction This thesis examines the comparative safety and effectiveness of medications used to aid smoking cessation. The thesis focuses on population-based use of varenicline, bupropion, and nicotine replacement therapy (NRT) during a government-sponsored reimbursement program in the Canadian province of British Columbia, and in a large U.S. population of government and employer-insured patients.  Smoking harms nearly every organ of the body and causes several diseases, including chronic obstructive pulmonary disease, cardiovascular disease, and some cancers.i Most of these diseases start to reverse after smoking stops.ii Smoking is the leading cause of premature death in North America, with cigarette smoking responsible for more than 480,000 deaths per year in the United States.iii According to the Centers for Disease Control and Prevention, 15.5% of all U.S. adults were current cigarette smokers in 2016, including 17.5% of adult men and 13.5% of adult women.iv  In Canada, an estimated 16.2% of residents, aged 12 and older, smoked either daily or occasionally in 2017. The proportion was higher among men (19.1%) than women (13.4%).v Nicotine replacement therapy has been available in various dosage forms as an over-the-counter aid to smoking cessation since 1996, prior to which it was restricted to prescription only in the U.S.vi Bupropion was the first non-nicotine based drug licensed for smoking cessation, receiving U.S. Food and Drug Administration (FDA) approval for smoking cessation in adults in 1997 for 100mg and 150mg sustained-release tablets.vii Varenicline was approved for smoking cessation in adults by the FDA in 2006 for 0.5mg and 1mg tablets.viii  Although smoking cessation (SC) pharmacotherapies have been shown to have a moderate short-term benefit as a quitting aid, substantial cardiovascular and neuropsychiatric safety concerns have been identified, leading to prescription label warnings by Health Canada 2  and the FDA. ix,x Recent studies indicate these warnings may be without meritxi-xiv, however, evidence from population-based use of these medications is limited.  1.1 Pharmacology 1.1.1 The Nicotine Addiction Cycle Nicotine is the principal addictive component of tobacco.xv The nicotine addiction cycle (see Figure 1.1) describes the tolerance and physical dependence created through repeated exposure to nicotine through smoking tobacco. 1) Smoking distills nicotine from tobacco and carries it into the lungs. 2) The nicotine is absorbed into the pulmonary venous circulation and then moves from the lungs to the brain by entering the arterial circulation. 3) Nicotine binds to nicotinic acetylcholine receptors, causing the release of a variety of neurotransmitters, most notably dopamine, which signals a pleasurable experience. 4) Over a short period of time, dopamine levels drop, resulting in a craving to smoke again. 5) Without nicotine, withdrawal symptoms such as anxiety and stress, will occur. 6) Over long periods of chronic exposure to nicotine, tolerance to the effects of nicotine develops, requiring a constant presence of nicotine to be maintained. xvi,xvii   1.1.2 Varenicline Varenicline is a highly selective, partial-agonist of the α4β2 nicotinic acetylcholine receptor subtype. It also prevents full-agonist nicotine from binding to the α4β2 receptors. These mechanisms of action provide relief of cravings and withdrawal symptoms when quitting, and presumably impedes the reinforcing effects of nicotine. When varenicline is taken by a smoker, 3  some dopamine is released, but less so than when smoking, which is assumed to reduce the rewarding properties of the nicotine cycle. Varenicline has a half-life of approximately 24 hours, independent of dose. When orally administered, it is almost entirely absorbed from the gastrointestinal tract, and its systemic availability is approximately 93%. Maximum serum concentration (Cmax) occurs within 4 hours, and a steady state after 4 days. There is no effect of food or time-of-day dosing on concentrations. xviii The usual oral dose in adults is 1mg twice daily for 12 weeks, often with an initial titration week. A randomized dose-response study with a 40 week follow-up found the varenicline dose with the highest efficacy was 1mg BID.xix The most common side effects reported with varenicline are nausea (30%), insomnia (18%), headache (15%), and abnormal dreams (13%). xx  1.1.3 Bupropion Bupropion is a norepinephrine and dopamine reuptake inhibitor antidepressant, with a mechanism of action for smoking cessation that may or may not be independent of its antidepressant effect.xxi Bupropion inhibits the neuronal reuptake of dopamine and norepinephrine, increasing extracellular concentrations, and the amount available to bind with post-synaptic receptor sites.xxii Bupropion also prevents nicotine activation of nicotinic receptors.xxiii  It is unclear which, if any, of these receptor effects are responsible for its influence on smoking cessation. Bupropion is also a potent inhibitor of cytochrome P450 2D6 (CYP2D6),xxiv which can raise the plasma levels of other antidepressants, antiarrhythmics, and antipsychotics.  4  Bupropion was first made available in the United States and other countries in 1989 as an antidepressant medication in a three-times-a-day immediate release formulation. A sustained release (SR) formulation was produced for smoking cessation based on observation that it helped reduce cravings for cigarettes, for some peoplexxv. The recommended dose of bupropion SR is 150mg daily for 5 days then twice daily for 7-12 weeks, starting 1-2 weeks prior to stopping smoking.xxvi,xxvii Common side effects include insomnia, dry mouth, and nausea.xxviii Bupropion has also been associated with a dose-related seizure risk. xxix   1.1.4 Nicotine Therapy Nicotine is the primary active ingredient in pharmaceutical NRT products, giving patients who are abstaining from smoking an alternative form of nicotine to mitigate symptoms of withdrawal. NRT therapy aids smoking cessation in three ways: 1. NRT partially substitutes the nicotine obtained from smoking tobacco which reduces withdrawal symptoms. 2. NRT sustains tolerance to nicotine, reducing the reinforcing effects of smoking tobacco. 3. NRT replaces some of the effects on mood, anxiety, and stress relief that were provided by smoking tobacco. xxx,xxxi  NRT is available in several forms: transdermal patch, chewing gum, lozenges, inhalers, nasal sprays, and sublingual tablets. The bioavailability of NRT through gastro-intestinal absorption is much less than transdermal patches and inhaling cigarette smoke, where it is   nearly 100% due to direct  transport to the brain through the pulmonary and arterial circulation.xxxii It is inaccurate to consider NRT a “replacement” for nicotine obtained from smoking tobacco because NRT products are not capable of producing such high plasma concentrations; NRT products generally provide a lower and slower, but more stable plasma nicotine concentration. xxxiii Efficacy is approximately the same between the dosage forms, 5  although pharmacokinetics and side-effects differ, such as skin irritation from patches, and mouth irritation from gum and tablets.xxxiv   1.2 Overview of Thesis Chapters The thesis consists of five chapters: An introduction, 3 studies in journal manuscript format, and a concluding chapter.  Chapter 2 is a retrospective new-user cohort study of the comparative safety of smoking cessation pharmacotherapies in non-depressed adults in the United States. The primary outcome was a composite of hospitalized cardiovascular events. Secondary outcomes included a composite of hospitalized neuropsychiatric events and individual components of the primary outcome. The study estimated risk ratios using propensity score adjusted log-binomial regression models. Chapter 3 is a population based cohort study assessing the relative safety between varenicline, bupropion, and NRTs, in British Columbia, during a government sponsored smoking cessation reimbursement program. The primary outcome was a composite of cardiovascular hospitalizations, secondary outcomes included all-cause mortality, a composite of neuropsychiatric hospitalizations, and individual components of the primary outcome. Risk ratios were estimated using propensity score adjusted log-binomial regression models. Chapter 4 This is a retrospective study of adult users of smoking cessation pharmacotherapy during a government sponsored smoking cessation reimbursement program. It describes and implements a novel method of analysis, namely using prescription therapy sequencing in comparative effectiveness research as a means to identify treatment failure. The method is 6  applied to B.C. Ministry of Health administrative data to identify re-initiation of smoking cessation pharmacotherapy as a proxy for treatment failure in a population-based setting. An additional multivariate analysis explores predictors of smoking cessation treatment failure. Chapter 5 is a concluding chapter that summarizes the main findings from Chapters 2 through 4, and puts study findings into context. It also recommends specific topics for future research.  1.3 Data Sources There were two data sources used in this thesis. 1) The British Columbia Ministry of Health database, provided via project-specific extracts from Population Data B.C., contained comprehensive administrative health claims data for residents of British Columbia. 2) The U.S. MarketScan® commercial claims database, provided via direct dataset access from Truven Health, encompassed U.S. insurance claims from employers, health plans, and state Medicaid agencies, covering over 100 million patients. xxxv Both data sources contained the following data dimensions unless otherwise noted: 1.3.1 Patient Registration and Demographics Health insurance registration, including starting and end dates. Also contains basic patient demographic information such as age, sex and residence region.  1.3.2 Hospital Discharge Abstracts Information on hospital discharges and deaths of in-patients from acute care hospitals. Provides patient information on admission type and date, discharge status and date, length of stay, patients diagnosis (principal and secondary diagnoses), and patient services (procedures) during the patient’s stay. In the B.C. data, the diagnosis codes are based on the International 7  Statistical Classification of Diseases and Related Health Problems, 10th Revision, Canada (ICD-10-CA) coding system and procedure codes are based on the Canadian Classification of Health Interventions (CCI) coding system. In the U.S. data, the diagnosis codes are based on the International Statistical Classification of Diseases and Related Health Problems, 9th Revision, Clinical Modification (ICD-9-CM) coding system and procedure codes are based on the Current Procedural Terminology, 4th Edition (CPT) coding system.          1.3.3 Physician Visits Information on outpatient services including physician encounters and claims for services that were delivered in a physician’s office or outpatient facility. Variables provided include: a de-identified linkable patient identifier, date of service, and diagnosis using the ICD-9 coding system.  The B.C. data also includes a ‘subsidy code’ variable to identify patients from a low-income household.    1.3.4 Dispensed Medications In British Columbia, medications dispensed from community pharmacies were recorded in the provincial PharmaNet software system. The data file provided for this thesis contains records for all dispensations regardless of who pays for the claim. Each claim contains: the date of dispensation, a unique linkable patient identifier and other patient information, drug information including drug strength and form, quantity and number of days dispensed, drug cost, and pharmacist dispensing fees. In the US MarketScan database provides an outpatient Pharmaceutical Claims Table that contains either mail-order or card program prescription drug claims. Each claim contains a de-8  identified linkable patient identifier, drug information including drug strength and form, quantity and number of days dispensed, and financial variables.  1.3.5 Vital Statistics Deaths (British Columbia Only) Date of death for all deaths registered at the Vital Statistics Agency in the province of B.C.  1.4 Statistical Model Selection Chapters 2-4 of this thesis use log-binomial regression models, regardless of outcome incidence rates, to calculate risk ratios in all fixed follow-up analyses, and Cox Proportional Hazards models to estimate hazard ratios in analyses where follow-up time varies by patient. Calculating relative risk by odds ratio was considered unsuitable in these cohort studies with a common outcome (>10%). In addition, where the outcomes were rare, and the odds ratio would have been considered a close approximation to the relative risk, given that alternative model choices exist that provide a more accurate estimate of relative risk, odds ratios were not used as an estimate of relative risk. A comparison of log-binomial and robust Poisson methods, using computer simulations and real data, found the log-binomial method yielded slightly higher power and smaller standard errors compared to the Poisson method.xxxvi For these reasons, the log-binomial model was considered the most appropriate choice for these studies. In cohort studies with a binary outcome, the odds ratio is often reported and interpreted as a risk ratio. When a high incidence of outcomes is present, the odds ratio will overstate the risk ratio, sometimes dramatically. xxxvii,xxxviii,xxxix In 1998, Zhang and Yu proposed a simple formula to convert an odds ratio to a relative risk.xl Although this method has been widely used in 9  epidemiologic studies, with 214 citations in scientific publications in the first 5 years after publishing, it has been shown to produce a biased estimate when confounding is present,  xli and can also produce biased confidence intervals that are erroneously narrow, leading one to interpret the point estimate as more precise than is true. McNutt et alxlii propose several approaches to calculating adjusted relative risk, including stratified analysis, log-binomial regression, and Poisson regression.    1.4.1 Log-Binomial Regression Log-binomial regression shares many properties to logistic regression. Both model the probability of the outcome occurring given the exposure and confounders, and both assume that the error terms have a binomial distribution. The difference between the two models is the link function between the independent variables and the probability of the outcome: Logistic regression uses the logit function and log-binomial models use the log function. xli The primary disadvantage to the log-binomial model is that it can suffer from non-convergence when adjusting for a large number of covariates. Several approaches have been described to deal with the failed convergence issues, encouraging continued use of log-binomial models.xliii In the cohort studies presented in this thesis, adjusting for high-dimensional propensity score deciles, which encompasses a large number of covariates, is an ideal strategy to overcome the log-binomial convergence issue.   10  1.4.2 Cox Proportional Hazards Time-to-event endpoints are common in pharmacoepidemiology as they add important information on timing of outcomes compared to fixed follow-up study designs where outcomes simply occur or not. Survival models, such as the Cox Proportional Hazards model (Cox model)xliv, are widely used to analyze time-to-event endpoints due to their flexibility with covariate forms, categorical or continuous, and easily interpretable regression coefficients. xlv,xlvi  The primary concept of the Cox model is the survival function, defined “as the probability of the outcome event not occurring up to a specific point in time”.xlvii Study designs with clearly defined, variable, follow-up time for all study participants are well suited for survival analysis. The observation period may end with or without the outcome event occurring, but follow-up time for all participants is included in the analysis. If participants drop out of the study (i.e. it can no longer be determined if the outcome event occured) the patient’s data are censored, i.e. follow-up time beyond that point in time is excluded. Cox models produce adjusted hazard ratios, defined as the ratio of the predicted hazard function under two different values of a predictor variable. A hazard ratio greater than one means the event is more likely to occur, and less than one means it is less likely to occur, than in the comparison group. xlviii  The Cox model assumes that the effect of the model parameters on the overall hazard is constant over time. There are several methods to assess the proportional hazards assumption, the most common methods are 1) plotting the Kaplan-Meier curves – if they cross then the assumption has been violated, and 2) plot the Schoenfeld residuals vs time – if they look like a random scatter around zero then the assumption is valid.xlix Both of these visual assessments were used to check the validity of the Cox models in Chapter 2. Failure to ensure the validity of the proportional hazards assumption may lead to erroneous or misleading results.  11  1.4.3 Propensity Scores High-dimensional exposure propensity scores (HdPS) are used in each of the studies to control for imbalances in patient characteristics between cohorts. The propensity score represents the predicted probability of receiving the treatment of interest conditional on each patients’ observed baseline characteristics.lxxxix  The high-dimensional propensity score algorithm, developed by Schneeweiss et al, empirically identifies potential covariates from the administrative data bases, ranks the covariates based on confounding adjustment potential, and integrates them into an exposure propensity score logistic regression model. Propensity score models produce a c-statistic that indicates how well the model was able to distinguish between patients who were treated versus untreated. For each pair of exposed and unexposed patients, the c-statistic gives the probability a treatment patient was assigned a higher propensity score. A model unable to discriminate between treated and untreated patients would have a c-statistic of 0.50.   There are four common methods to implementing propensity scores for confounding adjustment: 1) Stratification, 2) inverse probability of treatment weighting, 3) Matching, 4) Covariate Adjustment. The advantages and disadvantages of each of these methods have been extensively examined.l,li  Any of these methods would be appropriate to implement, however, the studies in this thesis used adjustment by propensity score decile, as it offered the best confounding adjustment without the potential loss of patients from the study cohorts.   12  Figure 1 The Nicotine Addiction Cycle   13  Chapter 2: Cardiovascular and Neuropsychiatric Safety of Smoking Cessation Pharmacotherapies in Non-Depressed Adults: A Retrospective Cohort Study.  2.1 Introduction Smoking harms nearly every organ of the body and causes several diseases, including chronic obstructive pulmonary disease, cardiovascular disease, and some cancers.lii Smoking is the leading preventable cause of death in North America, with cigarette smoking responsible for more than 480,000 deaths per year in the United States.liii According to the Centers for Disease Control and Prevention, 15.5% of all U.S. adults were current cigarette smokers in 2016, including 17.5% of adult men and 13.5% of adult women.liv Pharmacotherapies to aid smoking cessation in routine practice include varenicline, bupropion, and nicotine replacement therapy (NRT).lv Varenicline is a highly selective, partial-agonist of the α4β2 nicotinic acetylcholine receptor. It also prevents full-agonist nicotine from binding to the α4β2 receptors, lowering dopamine release, and reducing the rewarding properties of nicotine. Bupropion blocks the neuronal reuptake of dopamine and norepinephrine, increasing extracellular concentrations, and the amount available to bind with post-synaptic receptor sites. lvi Bupropion also prevents nicotine activation of nicotinic receptors.lvii It is unclear which mechanism of action is most responsible for its influence on smoking cessation. Bupropion was the first non-nicotine based drug licensed for smoking cessation, receiving U.S. Food and Drug Administration (FDA) approval for smoking cessation in adults in 1997. Varenicline was approved for smoking cessation in adults by the FDA in 2006. In conjunction with behavioural counselling, RCTs have shown varenicline, bupropion, and NRT therapy to be effective aids in smoking cessation, increasing continuous abstinence 14  rates compared with placebo, for periods up to 52 weeks.lviii-lxiv No RCTs have provided evidence on the efficacy of long-term (> 52 weeks) continuous smoking abstinence. Varenicline has the highest short-term quit rate efficacy; Varenicline doses of 1mg daily or 1mg twice daily have been shown to have superior effectiveness over bupropion sustained release titrated to 150mg twice daily or single forms of NRT.lxv-lxviii  Substantial concerns have been raised regarding the cardiovascular and neuropsychiatric safety of smoking cessation pharmacotherapies in the FDA’s adverse event reporting system database.lxix,lxx This prompted the FDA to impose Boxed Warnings, the strictest warning type issuable, on varenicline and bupropion in 2009 for psychiatric side effects including suicidal thoughts, hostility, and agitation, and included additional warnings in varenicline prescribing information concerning cardiovascular events in 2012, and serious neuropsychiatric symptoms and suicidality in 2013.lxxi The FDA also issued warnings in bupropion prescribing information concerning hypertension, seizure risk, mania, psychosis, and other neuropsychiatric reactions and suicidal thoughts and behaviours.lxxii However, these warnings are becoming increasingly inconsistent with evidence from randomized controlled trials (RCTs)lxxiii-lxxvi and observational research.lxxvii-lxxxi In accordance with the evolving safety evidence, and based on the results of a mandated RCT,lxxxii in 2016 the FDA removed the Boxed Warning on varenicline and bupropion for serious mental health side effects. The FDA required the manufacturers of varenicline (Chantix®) and bupropion (Zyban®), Pfizer Inc. and GlaxoSmithKline, respectively, to conduct a placebo- and active-controlled trial evaluating adverse events - Evaluating Adverse Events in a Global Smoking Cessation Study (EAGLES). This double-blind RCT was designed to evaluate cardiovascular safety and neuropsychiatric adverse events with smoking cessation therapies, comparing varenicline and 15  bupropion with NRT patch and placebo. The trial found no association of an increased risk of neuropsychiatric events with varenicline or bupropion compared to NRT patch, and no evidence that smoking cessation pharmacotherapies increased the risk of serious cardiovascular adverse events.lxxxii, lxxxiii  A large observational study investigating cardiovascular and neuropsychiatric safety of smoking cessation pharmacotherapies in England found no association with an increased risk of cardiovascular events, depression, or self-harm in varenicline or bupropion compared to NRT.lxxxiv  The results showed a lower risk of ischaemic heart disease, cerebral infarction, and depression in varenicline and bupropion users compared to NRT. This study has been criticized for findings that are inconsistent with RCT evidence, suggesting the study’s estimates are biased due to inadequately controlling for differences in patient characteristics, and being unable to overcome residual or unmeasured confounding.lxxxv Previous observational studies have received similar criticisms of bias by the FDA.lxxxvi Additional studies in a real-world setting are required to investigate the cardiovascular and neuropsychiatric safety profiles of smoking cessation pharmacotherapy users in a real-world setting. This study aims to provide additional evidence by investigating the comparative risk of serious cardiovascular and neuropsychiatric adverse events of smoking cessation pharmacotherapies, implementing robust confounding reduction methods using a large U.S. commercial insurance claims database.  16  2.2 Methods 2.2.1 Data We used the Truven Health MarketScan® Research Database (MarketScan), a commercial database encompassing U.S. insurance claims from employers, health plans, and state Medicaid agencies, covering over 100 million patients. lxxxvii The database included de-identified and linkable patient claims for outpatient prescription medications, physician outpatient visits and services, and hospital discharge records. The available data range was from January 2006 to December 2016. The MarketScan database has been widely used in observational pharmacoepidemiology research. 2.2.2 Study Design and Patients We conducted a retrospective new-user cohort study of adults aged 18 or older who initiated varenicline, bupropion 150mg sustained release (SR), or any form of nicotine replacement therapy. Cohort accrual began on January 01st, 2007, allowing for 12 months of available data to assess baseline characteristics, and ended on December 31st, 2015, allowing for 12 months of follow-up assessment. The study cohorts included all patients with a prescription dispensing for a smoking cessation pharmacotherapy between January 01st, 2007 and December 31st, 2015. The date of the first dispensing was defined as the cohort index date. Patients were excluded from the study if they: 1) did not have at least 12 months of continuous health insurance prior to the cohort index date, 2) were under 18 years old on the cohort index date, 3) used more than one smoking cessation product on the cohort index date, 4) had a prior prescription for any smoking cessation pharmacotherapy in the previous 365 days, or 5) were diagnosed or treated for depression in the 365 days prior to cohort index date. See the Patient Flow Diagram in Figure 2.  17  NRT was chosen as the comparison group because it is the oldest, least expensive treatment, often available without prescription.  2.2.3 Study outcomes The primary outcome was a composite cardiovascular endpoint, which included any hospital admission with a diagnostic code in any position of the hospital discharge abstract of: ischemic heart disease (ICD-9: 410.x, 411.x, 413.x, 414.x, ICD-10: I20.x-I25.x ), heart failure (ICD9: 428.x, ICD10: I50.x), cerebral infarction (ICD9: 430, 431, 432.x, 434.x, ICD10: I63.x (exclude I63.6), I64.x, H34.1, I60.x (exclude I60.8), I61.x, I62.x), peripheral vascular disease (ICD9: 440.x, 441.x, 443.x, 444.x, 445.x, 437.x, 557.x, ICD10: I70.x, I71.x, I73, I74.x, K55.1) unstable angina (ICD9: 413.x, ICD10: I20.x), or a hospital procedure related to coronary revascularization (CPT: 92920-92934, 92937-92944).  Secondary outcomes included individual components of the composite CVD outcome, and a composite neuropsychiatric outcome. The composite neuropsychiatric outcome included: Depression (ICD-9: 296.2, 296.3, 298.0, 300.4, 309.0, 309.28, 311.x, ICD-10: F32.x, F33.x, F34.x), anxiety (ICD-9: 300.x, ICD-10: F40.x, F41.x), schizophrenia (ICD-9: 295.x, ICD-10: F20.x), bipolar disorder (ICD-9: 296.0, 296.1, 296.4, 296.5, 296.6, 296.7, 296.8, ICD-10: F31.x), suicide attempt (ICD-9: E950.x-E959.x ICD-10: T14.91)lxxxviii, post-traumatic stress disorder (ICD9: 309.8, ICD10: F43.1), other psychosis (ICD9: 297.x, 298.x, ICD10: F39, F30, F22, F23, F24, F25, F28, F29, F34), and drug-induced mental disorders (ICD9: 292.x, ICD10: F13.1, F15.1, F19).   18  2.2.4 Patient Follow-up Patients began follow-up on the day after initiation on a study drug. Follow-up did not include the day of therapy initiation because a temporal sequence between drug use and the outcome event could not be discerned. In the primary analysis, patients were followed for a fixed 365-day period. In the sensitivity analysis, patients were followed for a period of time equal to their therapy duration plus 30 days, and were censored on the earliest of: switching to an alternative smoking cessation product, in hospital death, occurrence of outcome, loss of insurance coverage.  2.2.5 Patient Characteristics and Confounders Patient characteristics were assessed in the 12 months prior to, and not including, the cohort index date. We defined comorbidities using ICD-9, ICD-10, Current Procedural Terminology fourth edition (CPT-4) codes, and National Drug Codes (NDCs). Case definitions are described in Appendix B. To control for imbalances in patient characteristics between cohorts, we generated two sets of exposure propensity scores (NRT versus varenicline, and NRT versus bupropion) for each of the composite outcomes, using the high-dimensional propensity score algorithms provided by Schneeweiss et al.xcv The propensity score represented the predicted probability of receiving the treatment of interest conditional on each patients’ baseline characteristics.lxxxix Propensity score histograms and performance statistics for each of the models are provided in Appendix C. In addition to the 500 covariates empirically selected by the high-dimensional propensity score algorithms, we also included the following pre-specified covariates into the model, based upon data available in the 365 days preceding the cohort index date: Alcohol use disorder, 19  diabetes, chronic kidney disease, chronic obstructive pulmonary disease, hypertension, acute myocardial infarction, congestive heart failure, cerebrovascular disease, stroke, unstable angina, and year of cohort entry. Log-binomial regression models in the primary analysis and Cox proportional hazards models in the sensitivity analysis adjusted for propensity score decile, age group (18-44, 45-64, 65+), and sex.   2.2.6 Statistical Analysis Log-binomial regression models were used in the primary analysis to assess the association between the study and comparator drugs and outcomes in a one-year fixed follow-up period. Cox proportional hazards models, with a test for satisfaction of the proportional hazards assumption, were used in a sensitivity analysis. A second sensitivity analysis repeated the primary analysis using a subset of the study cohorts by trimming 5% of patients from each tail of the propensity score distribution. All analyses were performed using SAS 9.4 (SAS Institute Inc. Cary, NC.)  2.3 Results 2.3.1 Study cohort and patient characteristics A detailed comparison of patient characteristics between the study cohorts is shown in Table 2.5.1. Over 2 million patients in the database were dispensed a smoking cessation product during the available data range. After inclusion and exclusion criteria were applied, there were 454,698 varenicline users, 131,562 bupropion users, and 32,237 NRT users in the study cohorts. The average age of patients in the study was 46.0 years old. NRT users were on average 3.5 years older than bupropion patients, and 3.6 years older than varenicline patients. Women 20  accounted for 61% of the bupropion cohort compared with 44% and 46% of the varenicline and NRT cohorts, respectively. Most varenicline patients initiated therapy between 2007-2010 (65.4%), and cohort enrolment declined over the study period. NRT patients initiated therapy between 2012-2015 (59%), with increasing cohort enrolment over the study period. At baseline, NRT patients had a higher proportion of alcohol abuse, COPD, diabetes, chronic kidney disease, and history of cardiovascular disease compared to varenicline and bupropion patients.  2.3.2 Relative risk of primary and secondary outcomes Tables 2.5.2 and 2.5.3 show the risk of hospitalization for cardiovascular and neuropsychiatric events during the one-year follow-up period. In the analysis of the primary outcome - hospitalization for a cardiovascular event, varenicline was associated with a 20% one-year relative risk reduction compared to NRT (RR: 0.80, 95% CI: [0.75 – 0.85]), and bupropion was associated with a 25% one-year relative risk reduction compared to NRT (RR: 0.75, 95% CI: [0.69 – 0.81]). With respect to the individual components of the composite cardiovascular outcome, varenicline compared to NRT was associated with a one-year relative risk reduction for three outcomes: hospitalization for cerebral infarction (RR: 0.71, 95% CI: [0.60 – 0.84]), hospitalization for heart failure (RR: 0.72, 95% CI: [0.65 – 0.81]), and hospitalization for peripheral vascular disease (RR: 0.83, 95% CI: [0.75 – 0.92]). For two other outcomes, there were non-significant one-year relative risk reductions: for ischaemic heart disease (RR: 0.92, 95% CI: [0.82 – 1.03]) and angina (RR: 0.90, 95% CI: [0.72 – 1.13]). A non-significant increase in one-year relative risk was seen for coronary revascularization (RR: 1.07, 95% CI: [0.77 – 1.47]).  21  Bupropion compared to NRT was associated with lower one-year relative risks of hospitalization for ischaemic heart disease (RR: 0.79, 95% CI: [0.69 – 0.91]), cerebral infarction (RR: 0.78, 95% CI: [0.64 – 0.96]), heart failure (RR: 0.84, 95% CI: [0.74 – 0.96]), and peripheral vascular disease (RR: 0.65, 95% CI: [0.57 – 0.74]). Non-significant reductions in one-year relative risks of hospitalization were seen for coronary revascularization (RR: 0.68, 95% CI: [0.44 – 1.03]) and angina (RR: 0.80, 95% CI: [0.59 – 1.07]).  Compared to NRT, Varenicline was associated with a 35% reduction in one-year relative risk of hospitalization for neuropsychiatric events (RR: 0.65, 95% CI: [0.59 – 0.72]). In contrast, bupropion was associated with a 21% increase (RR: 1.21, 95% CI: [1.09 – 1.35]). Direct comparison of bupropion users versus varenicline users (Table 2.5.7) yielded a 17% reduction in one-year relative risk of hospitalization of cardiovascular events (RR: 0.83, 95% CI: [0.78 – 0.87]), and a 77% increase in one-year risk of hospitalization of neuropsychiatric events (RR: 1.77, 95% CI: [1.66 – 1.88]).    2.3.3 Sensitivity and subgroup analyses Results were similar in the time-to-event sensitivity analysis using Cox Proportional Hazards models, details shown in Tables 2.5.4 and 2.5.5. Varenicline was associated with a 29% decreased risk of cardiovascular hospitalization (Hazard Ratio (HR): 0.71 95% CI: [0.63-0.79]) and bupropion was associated with a 34% decreased risk [HR: 0.66, 95% CI: [0.58 – 0.75]). Varenicline was associated with a decreased risk of neuropsychiatric hospitalization (HR: 0.51, 95% CI: [0.43 – 0.61]), and bupropion was associated with a non-significant increase in neuropsychiatric hospitalization (HR: 1.11, 95% CI: [0.92 - 1.33]). In the analysis of individual components of the composite cardiovascular outcome, varenicline and bupropion were each 22  associated with a reduced risk of ischaemic heart disease, cerebral infarction, heart failure, and peripheral vascular disease, compared to NRT. The proportional hazards assumption was tested with a Kaplan-Meier plot. The curves of the plot were parallel, suggesting the proportional hazards assumption is valid.  A second sensitivity analysis using the trimmed study cohorts found similar results to our primary analysis (Table 2.5.6). Log-binomial regression models found a 15% one-year reduced risk of hospitalization for a composite of cardiovascular event (RR: 0.85, 95% CI: [0.78 – 0.93]) for varenicline compared to NRT, and a 27% reduction for bupropion compared to NRT (RR: 0.73, 95% CI: [0.67 – 0.81]).  2.4 Discussion In this large cohort study of patients without a history of depression in the year preceding enrolment, we found a decreased risk of cardiovascular hospitalization with each of varenicline and bupropion, compared to NRT therapy. In secondary outcomes analysis, we found a decreased risk of neuropsychiatric hospitalization with varenicline and an increased risk with bupropion, compared to NRT. Results were similar in the time-to-event sensitivity analysis. These results correlate well and add substantively to those of previous observational studies and RCTs, subject to important limitations.  2.4.1 Comparison with existing literature The cardiovascular findings in our real world study of patients are comparable with a recent systematic review and meta-analysis. xc  The authors included 38 RCTs in their analysis and found no evidence that varenicline increases the rate of cardiovascular serious adverse 23  events. The findings were similar in patients with or without a history of cardiovascular disease. Our results suggest varenicline does not increase, but rather reduces the risk of cardiovascular hospitalizations compared with NRT therapy. Authors of a large observational study did not find evidence of an increased risk of cardiovascular events, depression, or self-harm with varenicline or bupropion compared to NRT.lxxxiv Although a composite cardiovascular endpoint was not reported, the estimated hazard ratios for individual cardiovascular events in varenicline versus NRT were comparable to our study, particularly the hazard ratios calculated in our sensitivity analysis The incidence rates of individual cardiovascular events were lower in our study, possibly the result of differences in the sensitivity and specificity of our outcome definitions.  2.4.2 Interpretation of Results A recent Cochrane Review analyzed 39 RCTs comparing varenicline with bupropion, NRT, or placebo for long-term smoking cessation and found varenicline to be the most effective pharmacotherapy.xci Another Cochrane Review found varenicline to be equally effective to combination NRT as a quitting aid.xcii The cardio-protective findings in our study may be attributable to the effectiveness of varenicline in achieving continuous long-term smoking abstinence, and not due to the pharmacological properties of the drug itself. There are benefits to smoking cessation that occur nearly immediately, such as improvements in circulation and breathing.xciii Long-term benefits of smoking cessation are well known; a 2013 study of smoking and smoking cessation histories of over 200,000 US residents showed that smokers who quit before age 34 have nearly identical life expectancy as people who have never smoked, and quitting smoking at age 35 to 44 years results in a life expectancy increase of 9 years compared to those who continue to smoke.xciv  24  We found a decreased risk of neuropsychiatric hospitalization with varenicline compared to NRT. Our study excluded patients with a history of depression, a group that accounts for a large proportion of smokers. Figure 2 shows this exclusion factor removed nearly 38% of smoking cessation pharmacotherapy users from our study. Similar findings have been shown in several RCTs and observational studies comparing varenicline with NRT.xxi,xxiii,xxxix  These results provide further support for the FDAs decision to remove the varenicline Boxed Warning for neuropsychiatric adverse events.  We found an increased risk of neuropsychiatric hospitalization with bupropion compared to NRT. The imbalance in patient characteristics in Table 2.5.1 suggest some patients in the bupropion cohort were using the medication for treatment of depression and not for smoking cessation. Although we were careful to include only patients using sustained release 150mg tablets, it is likely that generic versions were also being prescribed for depression. The association between bupropion and increased neuropsychiatric hospitalizations should therefore be interpreted as a signal for further investigation, and not as a causal association. 2.4.3 Limitations Our study has several important limitations, mostly related to the observational nature of the study design. Selection bias can be difficult to overcome in an observational study due to its non-random patient cohort assignment and use of administrative claims data. Our study relied on the MarketScan administrative health claims database which is primarily used as a payment billing system. Important lifestyle variables are not available in the database, such as adiposity, healthy diet, and exercise. Some data may also be incomplete, inaccurate, or not captured. Specifically, the risk of mental health side effects associated with smoking cessation 25  pharmacotherapy may still be present, particularly if these side effects did not result in a hospitalization.   We chose NRT as the reference for most comparisons because it is the oldest, least expensive treatment, considered safe enough to be often available without prescription. However, the NRT users were the smallest group because many US health insurance plans do not cover it. Patients in the NRT group were older, had higher rates of excessive alcohol use, and had a higher prevalence of baseline cardiovascular conditions related to our primary outcome measure. There may also be important unmeasured patient characteristics related to the exposure or outcome. However, our study had an adequate size and implemented rigorous methods that have been shown to be effective in reducing residual confounding in administrative databases of treatment effects.xcv  Our results may not be generalizable to a population with a different health insurance status or no insurance. Commercially insured patients are more likely to have differential socioeconomic status, lifestyle, and other risk factors. However, the biological impact of smoking cessation pharmacotherapies is unlikely to differ substantially enough by health insurance status to reverse our study results.   2.5  Conclusion This study addresses the need for evidence in a real-world setting evaluating the cardiovascular and neuropsychiatric safety of varenicline and bupropion against an active comparator. The findings in our study have important clinical and policy implications. In this large cohort of patients without a history of depression, we did not find an increased risk of cardiovascular hospitalizations or neuropsychiatric hospitalizations with varenicline compared to 26  NRT. Given the comorbidities associated with smoking, varenicline should continue to be considered a treatment option for smoking cessation. These results also provide further support for the FDAs decision to remove the varenicline Boxed Warning for neuropsychiatric adverse events. We found a decreased risk of cardiovascular disease and an increased risk of neuropsychiatric hospitalizations associated with bupropion compared to NRT. These results require further investigation as the prescribing of bupropion may have been for symptoms of depression that were not captured in the database. 27   Figures Figure 2 Patient Flow Diagram      28  Tables  Table 2.5.1 Baseline Characteristics of Study Cohorts  N or mean % or SD N or mean % or SD N or mean % or SDNumber of patients 454,698               131,562               32,237                 Age (years), mean (SD) 45.8 11.9 45.9 12.8 49.4 13.2     18-55, n (%) 199,027                43.8% 58,587                  44.5% 10,382                  32.2%     56-75, n (%) 234,975                51.7% 65,938                  50.1% 18,730                  58.1%     76+, n (%) 20,696                  4.6% 7,037                    5.3% 3,125                    9.7%Female n (%) 199,376                43.8% 79,662                  60.6% 14,827                  46.0%Calendar year of cohort entry, n (%)2007 111,310                24.5% 6,817                    5.2% 2,273                    7.1%2008 76,616                  16.8% 9,105                    6.9% 2,214                    6.9%2009 67,123                  14.8% 9,376                    7.1% 3,657                    11.3%2010 42,140                  9.3% 20,869                  15.9% 2,263                    7.0%2011 36,496                  8.0% 19,928                  15.1% 2,760                    8.6%2012 38,409                  8.4% 20,654                  15.7% 4,199                    13.0%2013 29,501                  6.5% 15,408                  11.7% 4,042                    12.5%2014 29,665                  6.5% 17,091                  13.0% 4,777                    14.8%2015 23,438                  5.2% 12,314                  9.4% 6,052                    18.8%Length of Continuous Therapy in Days, mean (sd) 65.7                      (68.8) 266                        (355.9) 76                          (180.6)One-Year Medical History: Excessive Alcohol Use, n (%) 2,259                    0.5% 727                        0.6% 669                        2.1%COPD, n (%) 28,856                  6.3% 5,182                    3.9% 4,140                    12.8%Diabetes, n (%) 34,187                  7.5% 11,820                  9.0% 3,805                    11.8%Chronic Kidney Disease, n (%) 2,568                    0.6% 1,259                    1.0% 601                        1.9%Hypertension, n (%) 86,753                  19.1% 27,252                  20.7% 8,490                    26.3%Cerebral Infarction, n (%) 25,710                  5.7% 5,985                    4.5% 3,053                    9.5%Angina, n (%) 4,229                    0.9% 1,043                    0.8% 430                        1.3%Peripheral Vascular Disease, n (%) 13,601                  3.0% 3,108                    2.4% 1,840                    5.7%Acute Myocardial Infarction, n (%) 4,337                    1.0% 899                        0.7% 982                        3.0%Congestive Heart Failure, n (%) 3,783                    0.8% 1,252                    1.0% 1,009                    3.1%Cerebrovascular Disease, n (%) 8,680                    1.9% 2,199                    1.7% 1,470                    4.6%Coronary Revascularization, n (%) 179                        0.0% 72                          0.1% 149                        0.5%Depression, n (%) -                        0.0% -                        0.0% -                        0.0%One-Year Medication History: Statins, n (%) 84,114                  18.5% 24,361                  18.5% 7,621                    23.6%ACE inhibitors, n (%) 58,488                  12.9% 18,061                  13.7% 6,039                    18.7%Beta Blockers, n (%) 47,988                  10.6% 15,851                  12.0% 5,301                    16.4%Calcium Channel Blockers, n (%) 32,607                  7.2% 10,320                  7.8% 3,597                    11.2%Loop or thiazide diuretics, n (%) 55,410                  12.2% 20,327                  15.5% 5,753                    17.8%Fibrates, n (%) 11,205                  2.5% 3,029                    2.3% 815                        2.5%SSRIs / SNRIs / TCAs, n (%) -                        0.0% -                        0.0% -                        0.0%Number of hospitalization episodes of care:      0, n (%) 423,075                93.0% 121,407                92.3% 24,983                  77.5%      1, n (%) 26,934                  5.9% 8,465                    6.4% 5,805                    18.0%      2, n (%) 3,614                    0.8% 1,227                    0.9% 1,096                    3.4%      3, n (%) 744                        0.2% 292                        0.2% 226                        0.7%      4+, n (%) 331                        0.1% 171                        0.1% 127                        0.4%CharacteristicsVarenicline Bupropion NRT29   Table 2.5.2 Risk of Admission to Hospital for Cardiovascular (CVD) Events; Log-Binomial Regression with a One-Year Fixed Follow-up   30   Table 2.5.3 Risk of Admission to Hospital for Neuropsychiatric Events; Log-Binomial Regression with a One-Year Fixed Follow-up                    RR 95% CI  RR 95% CI1-Year Neuropsychiatric RiskNicotine 544 (1.7%) 1.00            1.00            Varenicline 3,099 (0.7%) 0.40            (0.37 - 0.44) 0.65            (0.59 - 0.72)Bupropion 1,963 (1.5%) 0.88            (0.80 - 0.97) 1.21            (1.09 - 1.35)Number of cases within one-year of follow-upCrude  Adjusted Risk Ratio (RR) Crude and Adjusted Log-Binomial Regression Models31  Table 2.5.4 Sensitivity Analysis: Incidence Rates and Hazard Ratios of Admission to Hospital for Cardiovascular Events; Cox Proportional Hazards  32  Table 2.5.5 Sensitivity Analysis: Incidence Rates and Hazard Ratios of Admission to Hospital for Neuropsychiatric Events; Cox Proportional Hazards   Table 2.5.6 Sensitivity Analysis: Risk of Admission to Hospital for Cardiovascular (CVD) Events; Log-Binomial Regression, 5% Trimmed Propensity Scores          Number of eventsPatient YearsIncidence Rate per 1000 PYs  HR 95% CI  HR 95% CIComposite Neuropsychiatric RiskNicotine 172               5,862.87      29.3                     1.00              1.00              Varenicline 799               95,832.80    8.3                       0.29              (0.25 - 0.35) 0.51              (0.43 - 0.61)Bupropion 1,061           52,285.52    20.3                     0.76              (0.65 - 0.90) 1.11              (0.92 - 1.33)Crude and Adjusted Cox P-H Regression ModelsIncidence of Events Crude Hazard Ratio (HR)  Adjusted Hazard Ratio n=  RR 95% CI  RR 95% CI 1-Year Composite CVD Risk (NRT vs Varenicline)Nicotine 22,796     576 (2.5%) 1.00              Reference 1.00              ReferenceVarenicline 415,444   7,003 (1.7%) 0.67              (0.61 - 0.73) 0.85              (0.78 - 0.93) 1-Year Composite CVD Risk (NRT vs Bupropion)Nicotine 26,004     715 (2.7%) 1.00              Reference 1.00              ReferenceBupropion 121,416   1,534 (1.3%) 0.46              (0.42 - 0.50) 0.73              (0.67 - 0.81)Crude and Adjusted Log-Binomial Regression ModelsNumber of cases within one-year of follow-upCrude Risk Ratio (RR)  Adjusted Risk Ratio (RR) 33  Table 2.5.7 Direct Comparison of Bupropion versus Varenicline Users; Log-Binomial Regression, Risk of Cardiovascular Hospitalization, Neuropsychiatric Hospitalization.  34  Chapter 3: Cardiovascular and Neuropsychiatric Safety of Smoking Cessation Pharmacotherapies: A Population-Based Retrospective Cohort Study.  3.1 Introduction On September 30th, 2011, the provincial government of British Columbia launch the Smoking Cessation Program which provided reimbursement for one of four smoking cessation aids: varenicline (Champix®), bupropion (Zyban®), nicotine chewing gum (Thrive®), or nicotine patch (Habitrol®). Reimbursement eligibility was for one continuous course of treatment for up to 12 weeks in duration, once per calendar year. Smoking cessation pharmacotherapies are moderately effective aids for short-term smoking abstinence.xcvi A recent Cochrane review of 39 studies comparing varenicline to placebo, bupropion, or nicotine replacement therapy (NRT) concluded varenicline delivers one extra successful quitter for every 11 users, compared with smokers attempting to quit without varenicline.xcvii No active-controlled RCTs have studied long-term cessation (>52 weeks without smoking), although one observational study from 654 general practices in England found varenicline users were more likely to be smoke free at 2-years compared with NRT users (28.8% versus 24.3%, respectively), and the association persisted for up to 4 years.xcviii Shortly after varenicline was approved for use in Canada in 2007, several reports of neuropsychiatric adverse events, including agitation and suicide, were registered in Health Canada’s Adverse Reaction Database, and in similar adverse reporting databases in other jurisdictions.xcix,c These case reports resulted in neuropsychiatric safety warnings for varenicline and bupropion by Health Canada and the U.S. Food and Drug Administration (FDA).ci Several placebo-controlled RCTs have studied the neuropsychiatric safety of smoking cessation 35  therapies, but have not shown varenicline or bupropion to be associated with an increased neuropsychiatric risk.cii,ciii,civ Evidence of the neuropsychiatric safety between smoking cessation therapies from active-controlled RCTs is limited.  In 2011, Health Canada issued a safety alert informing Canadians that varenicline was undergoing a safety review for a possible increased risk of heart-related side effects in patients with a history of cardiovascular disease.cv This warning was based on an RCT which showed risk of cardiovascular disease in varenicline users was double compared to those not taking the drug.cvi  RCT’s have produced conflicting evidence on the cardiovascular risks associated with smoking cessation pharmacotherapies. A 2011 systematic review and meta-analysis of 14 RCTs found varenicline was associated with a 72% increased risk of cardiovascular events compared to placebo.cvii However, a 2016 systematic review and meta-analysis found no evidence that varenicline increases the risk of cardiovascular events compared to placebo.cviii There is limited population-based evidence about the comparative safety of smoking cessation pharmacotherapies from observational studies in a community-based setting. This study will investigate the comparative safety between these products during a government sponsored reimbursement program in the Canadian province of British Columbia. This study is, to our knowledge, the largest population-based observational study on this topic.   3.2 Methods 3.2.1 Data Access to the B.C. Ministry of Health administrative health claims database was provided by Population Data BC through a secure access environment. The database contained de-identified and linkable patient records for prescriptions dispensed at community pharmacies, 36  outpatient physician visits, hospital separations, and vital statistics mortality data.cix-cxiv We assume the completeness and accuracy of the administrative data is comparable to other administrative databases.cxv   3.2.2 Study population and drug exposure Our study period extended from September 30, 2011 to March 31, 2013. We identified patients aged 18 or older with a medication dispensing for varenicline, bupropion (Zyban® only), or nicotine replacement therapy (Habitrol® or Thrive® only). Patients entered the study cohort on the date of first use of any of these medications during the study period, provided they had a minimum of six-months continuous enrolment in the provincial medical services plan prior to study entry. The baseline characteristics of individuals in each study cohort are summarized in Table 3.5.1. NRT was chosen as the comparison group because it is the oldest, least expensive treatment, often available without prescription.  3.2.3 Outcome definition The primary outcome was a composite outcome of cardiovascular events which included a hospital admission with a diagnosis, in any diagnostic position, for ischemic heart disease (ICD-10: I20.x-I25.x),cxvi heart failure (ICD10: I50.x),cxvii cerebral infarction (ICD10: I63.x (exclude I63.6), I64.x, H34.1, I60.x (exclude I60.8), I61.x, I62.x),cxviii,cxix peripheral vascular disease (ICD10: I70.x, I71.x, I73, I74.x, K55.1),cxx unstable angina (ICD10: I20.x), or a hospital procedure related to coronary revascularization (CCI: 1IJ50.x, 1IJ57GQ.x, 1IJ76.x , 1IJ80.x, 1IJ57LA.x).cxxi 37  Secondary outcomes included all-cause mortality, the individual components of the primary outcome, and a composite outcome of neuropsychiatric events. The composite neuropsychiatric outcome included a hospital admission with a diagnosis, in any diagnostic position, for depression (ICD-10: F32.x, F33.x, F34.x), anxiety (ICD-10: F40.x, F41.x), schizophrenia (ICD-10: F20.x), bipolar disorder (ICD-10: F31.x), suicide attempt (ICD-10: T14.91)cxxii, post-traumatic stress disorder (ICD10: F43.1), other psychosis (ICD10: F39, F30, F22, F23, F24, F25, F28, F29, F34), and drug-induced mental disorders (ICD10: F13.1, F15.1, F19).   3.2.4 Patient Follow-up Patients began follow-up on the day after initiation on a study drug. Follow-up did not include the day of therapy initiation because a temporal sequence between drug use and the outcome event could not be discerned. In the primary analysis, patients were followed for a fixed 365-day period. In a secondary analysis, patients were followed for a 180-day period.  3.2.5 Covariate assessment Patient characteristics were assessed in the twelve months prior to, and not including, the cohort index date. We defined comorbidities using ICD-9, ICD-10, Canadian Classification of Health Intervention (CCI) codes, and Canadian Drug Information Numbers (DINs). Case definitions are described in Appendix E. To control for imbalances in patient characteristics between cohorts, we generated two sets of exposure propensity scores (NRT versus varenicline, NRT versus bupropion), for each of the composite outcomes (cardiovascular, neuropsychiatric) as the predicted probability of 38  receiving the treatment of interest conditional on each patients’ baseline characteristics.cxxiii In addition to the 500 covariates empirically selected by the high-dimensional propensity score algorithms, we also included the following pre-specified covariates based upon data available in the 180-days preceding the cohort index date: Alcohol use disorder, diabetes, chronic kidney disease, chronic obstructive pulmonary disease, hypertension, acute myocardial infarction, congestive heart failure, cerebrovascular disease, stroke, unstable angina, and year of cohort entry.   3.2.6 Statistical Analysis The primary analysis compared the composite CVD endpoint in varenicline compared to nicotine replacement therapy, and in bupropion compared to nicotine replacement therapy, during a one-year fixed follow-up period. A second analysis examined the same comparisons during a 6-month fixed follow-up period. Risk ratios were calculated using log-binomial regression models adjusted for propensity score decile, age group (18-44, 45-64, 65+), patient sex, and low annual family income (<= $30,000).cxxiv,cxxv All analyses were performed using SAS 9.4 (SAS Institute Inc. Cary, NC.)  3.2.7 Sensitivity Analysis: No recent history of cardiovascular events A sensitivity analysis repeated the primary analysis for the mortality and composite cardiovascular outcomes using the subset of patients with no history of cardiovascular disease in the one year prior to cohort entry. The case definition for history of cardiovascular disease is described in Appendix F.  39  3.3 Results 3.3.1 Study cohort and patient characteristics We identified 130,638 patients who used a smoking cessation pharmacotherapy during the study period. After applying exclusion criteria, our study included a total of 116,442 patients using either varenicline (n=39,094), bupropion (n=5,838), or NRT (n=71,510). A detailed comparison of patient characteristics between the study cohorts is shown in Table 3.5.1. The mean age of all study participants was 46.6, and 48% were women. NRT patients were more likely to be from a low-income household (40.9%) than varenicline patients (34.5%) or bupropion patients (38.8%). NRT patients were less likely to have received a prescription for a smoking cessation pharmacotherapy in the year prior to study cohort entry. Use of medications was similar between the study cohorts, although NRT patients had a higher use of medications used for depression.  A higher proportion of NRT patients were also diagnosed with depression (20.5%) compared to varenicline (15.4%) and bupropion (17.8%) patients. NRT patients had a higher proportion of cardiovascular diagnoses such as acute myocardial infarction, heart failure, cerebrovascular disease, angina, and coronary revascularization, although the numbers for cardiovascular diagnoses were small.   3.3.2 Relative risk (risk ratio) of primary and secondary outcomes During the 12-month follow-up period 2,039 of 116,442 patients (1.75%) were hospitalized for a cardiovascular event. Compared to NRT users, the adjusted risk ratio (RR) for the composite cardiovascular events associated with varenicline was 0.90 (95% CI 0.82 - 1.00), and the RR associated with bupropion compared to NRT was 0.95 (95% CI 0.77 – 1.17). The RR for the composite neuropsychiatric events associated with varenicline compared to NRT was 40  0.80 (95% CI 0.71 – 0.89), and for bupropion compared to NRT was 0.95 (95% CI 0.76 – 1.17). The RR for mortality associated with varenicline was 0.81 (95% CI 0.71 – 0.93), and bupropion was 1.04 (0.71 – 0.93).  No significant difference was found for individual components of the composite cardiovascular outcome for varenicline compared to NRT, except for ischaemic heart disease, RR = 0.77 (95% CI 0.63 – 0.93).  There were 1,192 of 116,442 patients (1.0%) hospitalized for a cardiovascular event during the first 6-months of follow-up. Compared to NRT, the RR for the composite cardiovascular events associated with varenicline was 0.88 (95% CI 0.77 - 1.01), and the RR associated with bupropion compared to NRT was 1.02 (95% CI 0.78 – 1.34). The RR for the composite neuropsychiatric events associated with varenicline compared to NRT was 0.78 (95% CI 0.67 – 0.90), and for bupropion compared to NRT was 0.81 (95% CI 0.60 – 1.11). The RR for mortality associated with varenicline compared to NRT was 0.70 (95% CI 0.57 – 0.86), and for bupropion compared to NRT was 0.92 (0.62 – 1.36). Detailed results are in Tables 3.5.2 – 3.5.5.  3.3.3 Sensitivity analyses The sensitivity analysis removed 5,521 patients (4.7%) with a cardiovascular diagnosis or procedure in the 365 days prior to study cohort entry. Results for the sensitivity analysis are shown in Tables 3.5.6 and 3.5.7. During the 12-month follow-up period, the RR for the composite cardiovascular outcome between varenicline and NRT was 0.91 (95% CI 0.80 – 1.04) and between bupropion and NRT was 0.94 (95% CI 0.72 – 1.22). Results were similar in the 6-month follow-up period. Neither varenicline nor bupropion compared to NRT were significantly associated with a difference in mortality, RR = 0.91 (95% CI 0.78 – 1.05) and RR = 1.00 (95% CI 0.74 – 1.35), respectively.  41  Post hoc direct comparison of bupropion users versus varenicline users (Table 3.5.8) found no significant difference in one-year relative risk of hospitalization for cardiovascular events (RR: 0.99, 95% CI: [0.79 – 1.23]), neuropsychiatric events (RR: 1.15, 95% CI: [0.91 – 1.46]), or mortality (RR: 1.18, 95% CI: [0.90 – 1.56]).    3.4 Discussion 3.4.1 Comparison with existing literature A recent RCT compared varenicline and bupropion to NRT for major adverse cardiovascular events (MACE) and found the time to cardiovascular events did not differ significantly by treatment.cxxvi The same trial also assessed psychiatric disorders using a composite of moderate and severe neuropsychiatric adverse events. The risk difference for varenicline and bupropion compared to nicotine patch in a non-psychiatric cohort were -1.07 (-2.21 to 0.08) and 0.13 (-1.19 to 1.45), respectively, over 24 weeks of follow-up.cii  Using outcome event results from the trial, we calculated the neuropsychiatric relative risks for varenicline and bupropion, compared to NRT, as 0.53 (0.27 – 1.03) and 0.90 (0.51 – 1.58), respectively. These associations are about double the magnitude of our findings of a 20% 1-year neuropsychiatric risk reduction for varenicline versus NRT and a 5% non-significant decreased risk for bupropion versus NRT. Our findings were  consistent with a recent observational study of 753 general practices in England that found neither bupropion nor varenicline were associated with an increased risk of any cardiovascular or neuropsychiatric event compared with NRT.cxxvii  42  3.4.2 Interpretation of Results The study cohorts are well balanced on demographic and other factors, but a higher proportion of NRT users were from low-income households. Financial stress has been associated with smoking intensity and may also be linked to poorer diet and delayed treatment for health issues.cxxviii Smoking by itself is a major risk factor for cardiovascular disease, and a greater smoking intensity among NRT users could have resulted in the observed increase in cardiovascular events. Although the 1-year composite cardiovascular outcome risk was not statistically significant, the point estimates did show a 10% and 5% relative risk reduction for varenicline and bupropion, respectively, compared to NRT.  The observed risk reduction could be the results of physicians channelling patients at higher cardiovascular risk towards NRT therapy, and not due to a relative pharmacological cardio-protective benefit of varenicline or bupropion. Evidence of this is seen in the baseline characteristics table where NRT patients are more likely to have a diagnosis of cardiovascular related disease in the 1-year prior to study entry. Less channeling bias would be expected in the direct comparison of varenicline users with bupropion users, with respect to cardiovascular outcomes. Compared to NRT, varenicline was associated with a 1-year relative risk reduction in mortality, but this result did not remain statistically significant in the sensitivity analyses where users with a recent history of cardiovascular events were excluded. NRT users were more likely to have had a prior cardiovascular event compared to varenicline or bupropion users. Since patients with a history of cardiovascular disease are at higher risk of a future cardiovascular event, it is probable that a higher proportion of NRT patients were at higher risk during the study follow-up period. Because of this, the risk ratio attenuating towards the null was an expected 43  result of the sensitivity analysis, and the imbalance in cardiovascular risk at baseline was primarily responsible for the mortality risk benefit seen in the primary analysis.  3.4.3 Study strengths The main strength of this study was that all patients entered the study with a medication adjudicated at the pharmacy through the provincial Smoking Cessation Program. This minimizes the possibility of the medications being used for purposes other than smoking cessation; a particular concern with bupropion and treatment of depression. Another strength of this study is its large size and generalizability due to British Columbia’s population-based health claims database. Unlike patients enrolled in RCTs, our study participants represent real-world use of these medications. Detailed longitudinal data on patients’ pharmacy claims linked to comprehensive medical records is another advantage of the study data source. This study used rigorous pharmaco-epidemiologic methods such as the implementation of high dimensional propensity score adjustment to control for a large number of confounders, specific outcome definitions, and use of an active control.  The use of an active comparator design increased the overlap of measured patient characteristics, which reduces the potential for unmeasured confounding.cxxix   3.4.4 Limitations Our study has important limitations, mostly related to the observational design and the nature of claims data. First, evidence for channelling of lower-income patients and patients with prior cardiovascular events towards NRT therapy is indicated in the baseline characteristics table, and although the application of high dimensional propensity score adjustment is a well-44  established method of controlling for confounding, residual confounding cannot be ruled out. Second, important lifestyle factors which cannot be directly measured in the B.C. claims data may also have led to bias, such as diet, body mass index, and smoking intensity. Third, NRT products are available over-the-counter and were not comprehensively recorded in the B.C database; some patients may have chosen not to participate or were not eligible for the Smoking Cessation Program, and paid directly for their NRT medication. There is also no record of NRT use in our database prior to the Smoking Cessation Program. We included prevalent users in our study design to compensate for this, which is an approach with advantages and disadvantages. The advantage of a prevalent user design is generalizability by including all eligible users of the study medications. The primary disadvantage is the possible depletion of patients susceptible to the outcome. If one of the medications causes the outcome, that patient may have discontinued the product prior to entering into our study, and both the exposure and outcome are not accounted for. However, it is unlikely that depletion of susceptible patients would be a strong enough bias to reverse the results of our findings.     3.5  Conclusion The study showed no evidence of an increased cardiovascular or neuropsychiatric risk among varenicline or bupropion compared to NRT users. The results were consistent between the 180-day and 365-day follow-up periods. The decreased 1-year mortality risk in varenicline users did not hold in a sensitivity analysis where patients with pre-existing cardiovascular events were excluded.    45  Figures Figure 3 British Columbia Patient Flow Diagram            46  Tables Table 3.5.1 Baseline Characteristics of Study Cohorts       N or mean % or SD N or mean % or SD N or mean % or SD N or mean % or SD N or mean % or SDNumber of patients 39,094       5,838         71,510       16,078            55,432            Age (years), mean (SD) 46.5 (13.5) 47.7 (13.7) 46.5 (14.6) 44.9 (14.7) 47 (14.5)     18-55, n (%) 28,597       73.1% 4,087          70.0% 51,307       71.7% 12,005            74.7% 39,302            70.9%     56-75, n (%) 10,082       25.8% 1,647          28.2% 18,709       26.2% 3,820              23.8% 14,889            26.9%     76+, n (%) 415             1.1% 104             1.8% 1,494          2.1% 253                  1.6% 1,241              2.2%Female (n,%) 18,394       47.1% 2,931          50.2% 34,695       48.5% 6,910              43.0% 27,785            50.1%Family Income < $30,000 13,501       34.5% 2,263          38.8% 29,221       40.9% 6,685              41.6% 22,536            40.7%Calendar year of cohort entry, n (%)      2011 15,546       39.8% 2,084          35.7% 23,399       32.7% 4,764              29.6% 18,635            33.6%      2012 19,939       51.0% 3,244          55.6% 40,895       57.2% 9,591              59.7% 31,304            56.5%      2013 3,609          9.2% 510             8.7% 7,216          10.1% 1,723              10.7% 5,493              9.9%Smoking cessation products in prior year:      Varenicline, n (%) 5,740          14.7% 166             2.8% 1,313          1.8% 212                  1.3% 1,101              2.0%      Buproprion, n (%) 1,207          3.1% 715             12.2% 2,939          4.1% 701                  4.4% 2,238              4.0%Personal Smoking Health Risk Assessment:     In prior year 7,737          19.8% 1,151          19.7% 5,536          7.7% 997                  6.2% 4,539              8.2%     In 3 days prior and incl. index date 3,799          9.7% 582             10.0% 916             1.3% 140                  0.9% 776                  1.4%Prescriptions in prior year:Statins, n (%) 3,384          8.7% 580             9.9% 5,860          8.2% 1,137              7.1% 4,723              8.5%ACE inhibitors, n (%) 3,318          8.5% 538             9.2% 5,664          7.9% 1,122              7.0% 4,542              8.2%Beta Blockers, n (%) 2,367          6.1% 380             6.5% 4,388          6.1% 843                  5.2% 3,545              6.4%Calcium Channel Blockers, n (%) 1,825          4.7% 290             5.0% 3,104          4.3% 591                  3.7% 2,513              4.5%Loop or thiazide diuretics, n (%) 1,409          3.6% 210             3.6% 2,444          3.4% 483                  3.0% 1,961              3.5%Fibrates, n (%) 147             0.4% 22                0.4% 260             0.4% 54                    0.3% 206                  0.4%SSRIs, n (%) 3,598          9.2% 576             9.9% 7,866          11.0% 1,730              10.8% 6,136              11.1%SNRIs, n (%) 1,718          4.4% 239             4.1% 3,589          5.0% 810                  5.0% 2,779              5.0%TCAs, n (%) 457             1.2% 67                1.1% 926             1.3% 201                  1.3% 725                  1.3%Medical diagnosis in prior year:Excessive alcohol use, n (%) 1,042          2.7% 142             2.4% 2,856          4.0% 652                  4.1% 2,204              4.0%Depression/Anxiety, n (%) 6,025          15.4% 1,037          17.8% 14,686       20.5% 3,429              21.3% 11,257            20.3%COPD, n (%) 4,060          10.4% 620             10.6% 6,897          9.6% 1,066              6.6% 5,831              10.5%Acute Myocardial Infarction, n (%) 251             0.6% 47                0.8% 1,011          1.4% 93                    0.6% 918                  1.7%Congestive Heart Failure, n (%) 424             1.1% 71                1.2% 1,254          1.8% 156                  1.0% 1,098              2.0%Cerebrovascular Disease, n (%) 520             1.3% 96                1.6% 1,223          1.7% 193                  1.2% 1,030              1.9%Peripheral Vascular Disease, n (%) 742             1.9% 126             2.2% 1,543          2.2% 250                  1.6% 1,293              2.3%Coronary Revascularization, n (%) 157             0.4% 36                0.6% 746             1.0% 63                    0.4% 683                  1.2%Angina, n (%) 703             1.8% 117             2.0% 1,604          2.2% 265                  1.6% 1,339              2.4%Diabetes, n (%) 3,769          9.6% 583             10.0% 6,467          9.0% 1,304              8.1% 5,163              9.3%Chronic Kidney Disease, n (%) 419             1.1% 79                1.4% 928             1.3% 161                  1.0% 767                  1.4%Habitrol®CharacteristicsVarenicline Bupropion NRT Thrive®47  Table 3.5.2 Risk Ratios of Cardiovascular Events During 1-year of Follow-up   RR 95% CI  RR 95% CI1-Year Composite CVD RiskNicotine 1,348 (1.9%) 1.00              Reference 1.00              ReferenceVarenicline 595 (1.5%) 0.81              (0.73 - 0.89) 0.90              (0.82 - 1.00)Bupropion 96 (1.6%) 0.87              (0.71 - 1.07) 0.95              (0.77 - 1.17)1-Year Ischaemic Heart Disease RiskNicotine 398 (0.6%) 1.00              Reference 1.00              ReferenceVarenicline 159 (0.4%) 0.73              (0.61 - 0.88) 0.77              (0.63 - 0.93)Bupropion 30 (0.5%) 0.92              (0.64 - 1.34) 1.02              (0.70 - 1.50)1-Year Cerebral Infarction RiskNicotine 186 (0.3%) 1.00              Reference 1.00              ReferenceVarenicline 83 (0.2%) 0.82              (0.63 - 1.06) 0.95              (0.72 - 1.26)Bupropion 9 (0.2%) 0.59              (0.30 - 1.16) 0.73              (0.37 - 1.45)1-Year Heart Failure RiskNicotine 433 (0.6%) 1.00              Reference 1.00              ReferenceVarenicline 137 (0.4%) 0.58              (0.48 - 0.70) 0.82              (0.67 - 1.01)Bupropion 30 (0.5%) 0.85              (0.59 - 1.23) 1.13              (0.77 - 1.65)1-Year Peripheral Vascular Disease RiskNicotine 302 (0.4%) 1.00              Reference 1.00              ReferenceVarenicline 179 (0.5%) 1.08              (0.90 - 1.30) 1.14              (0.94 - 1.40)Bupropion 28 (0.5%) 1.14              (0.77 - 1.67) 1.04              (0.70 - 1.54)1-Year Coronary RevascularizationNicotine 336 (0.5%) 1.00              Reference 1.00              ReferenceVarenicline 160 (0.4%) 0.87              (0.72 - 1.05) 0.88              (0.72 - 1.08)Bupropion 27 (0.5%) 0.98              (0.67 - 1.46) 0.99              (0.66 - 1.49)1-Year AnginaNicotine 146 (0.2%) 1.00              Reference 1.00              ReferenceVarenicline 77 (0.2%) 0.96              (0.73 - 1.27) 0.96              (0.71 - 1.29)Bupropion 11 (0.2%) 0.92              (0.50 - 1.70) 0.96              (0.51 - 1.81)Crude and Adjusted Log-Binomial Regression ModelsNumber of cases within one-year of follow-upCrude Risk Ratio (RR)  Adjusted Risk Ratio (RR) 48  Table 3.5.3 Risk Ratios of Cardiovascular Events During 6-Months of Follow-up  49     Table 3.5.4 Risk Ratios of Neuropsychiatric Events    Table 3.5.5 Risk Ratios of mortality at 6-months and 1-year       RR 95% CI  RR 95% CI1-Year Composite Neuropsychiatric RiskNicotine 1,743 (2.4%) 1.00              Reference 1.00              ReferenceVarenicline 465 (1.2%) 0.49              (0.44 - 0.54) 0.80              (0.71 - 0.89)Bupropion 91 (1.6%) 0.64              (0.52 - 0.79) 0.95              (0.76 - 1.17)6-Month Composite Neuropsychiatric RiskNicotine 1,012 (1.4%) 1.00              Reference 1.00              ReferenceVarenicline 260 (0.7%) 0.47              (0.41 - 0.54) 0.78              (0.67 - 0.90)Bupropion 44 (0.8%) 0.53              (0.39 - 0.72) 0.81              (0.60 - 1.11)Crude and Adjusted Log-Binomial Regression ModelsNumber of cases within one-year of follow-upCrude Risk Ratio (RR)  Adjusted Risk Ratio (RR)  RR 95% CI  RR 95% CI1-Year Mortality RiskNicotine 1,032 (1.4%) 1.00              Reference 1.00              ReferenceVarenicline 312 (0.8%) 0.55              (0.49 - 0.63) 0.81              (0.71 - 0.93)Bupropion 64 (1.1%) 0.76              (0.59 - 0.98) 1.04              (0.80 - 1.36)6-Month Mortality RiskNicotine 555 (0.8%) 1.00              Reference 1.00              ReferenceVarenicline 133 (0.3%) 0.44              (0.36 - 0.53) 0.70              (0.57 - 0.86)Bupropion 27 (0.5%) 0.60              (0.41 - 0.88) 0.92              (0.62 - 1.36)Crude and Adjusted Log-Binomial Regression ModelsNumber of cases within one-year of follow-upCrude Risk Ratio (RR)  Adjusted Risk Ratio (RR) 50  Table 3.5.6 Sensitivity Analysis: Risk ratios of cardiovascular events in patients with no cardiovascular events in 6-months prior to cohort entry   Table 3.5.7 Sensitivity Analysis: Risk Ratios of Mortality in Patients with No Cardiovascular Events in 6-Months Prior to Cohort Entry       RR 95% CI  RR 95% CI1-Year Composite CVD RiskNicotine 773/67758 (1.1%) 1.00              Reference 1.00              ReferenceVarenicline 404/37585 (1.1%) 0.94              (0.84 - 1.06) 0.91              (0.80 - 1.04)Bupropion 63/5578 (1.1%) 0.99              (0.77 - 1.28) 0.94              (0.72 - 1.22)6- Month Composite CVD RiskNicotine 439/67758 (0.6%) 1.00              Reference 1.00              ReferenceVarenicline 279/37585 (0.7%) 0.87              (0.73 - 1.03) 0.85              (0.71 - 1.02)Bupropion 33/5578 (0.6%) 0.97              (0.68 - 1.38) 0.93              (0.65 - 1.34)Crude and Adjusted Log-Binomial Regression ModelsNumber of cases within one-year of follow-upCrude Risk Ratio (RR)  Adjusted Risk Ratio (RR)  RR 95% CI  RR 95% CI1-Year Mortality RiskNicotine 770/67,758 (1.1%) 1.00              Reference 1.00              ReferenceVarenicline 280/37,585 (0.7%) 0.66              (0.57 - 0.75) 0.91              (0.78 - 1.05)Bupropion 49/5,578 (0.9%) 0.77              (0.58 - 1.03) 1.00              (0.74 - 1.35)6- Month Mortality RiskNicotine 409/67,758 (0.6%) 1.00              Reference 1.00              ReferenceVarenicline 118/37,585 (0.3%) 0.52              (0.42 - 0.64) 0.78              (0.63 - 0.97)Bupropion 22/5,578 (0.4%) 0.65              (0.43 - 1.00) 0.95              (0.61 - 1.47)Crude and Adjusted Log-Binomial Regression ModelsNumber of cases within one-year of follow-upCrude Risk Ratio (RR)  Adjusted Risk Ratio (RR) 51  Table 3.5.8 Direct Comparison of Bupropion versus Varenicline; Log-binomial Regression, Risk of Cardiovascular Hospitalization, Neuropsychiatric Hospitalization, Mortality  52  Chapter 4: Identifying Sequential Episodes of Pharmacotherapy as a Method for Assessing Treatment Failure in Comparative Effectiveness Research  4.1 Introduction Tobacco dependence remains one of the biggest single preventable causes of death in the world.cxxx Smoking cessation pharmacotherapies are moderately effective aids for short-term smoking abstinencecxxxi,cxxxii but the majority of people quit smoking unassisted.cxxxiii  Recent studies have shown that public funding of smoking cessation pharmacotherapies increases the likelihood of use.cxxxiv Several randomized controlled trials (RCTs) have assessed the comparative efficacy of smoking cessation pharmacotherapies, finding varenicline to be the most effective monotherapy, compared with nicotine replacement therapy (NRT) and bupropion.cxxxv-cxliii However, these RCTs suffer from short durations and incomplete reporting of patients during the study period.cxliv Evidence of comparative effectiveness in a real-world setting is limited.   The Canadian province of British Columbia (B.C.) provided a large-scale natural experiment of a smoking cessation by introducing a new drug benefit policy. On September 30th 2011, The B.C. Ministry of Health launched the Smoking Cessation Program (the Program), which provided financial coverage for varenicline, bupropion, nicotine gum, and nicotine patches. Each calendar year, B.C. residents were eligible to receive full coverage for a single continuous course of treatment for up to 84 days with nicotine gum or nicotine patches, or coverage of one of the two prescription drugs, bupropion (Zyban® only) and varenicline (Champix® only), in accordance with their income-based benefit plan.  53  In this analysis, we evaluate the effectiveness of smoking cessation pharmacotherapies using a novel method of identifying sequential drug therapy episodes as a proxy for initial treatment failure. We hypothesise that patients who are unable to achieve smoking abstinence during their first course of treatment during the Program will attempt a subsequent course of therapy, allowing us to identify the subsequent therapy as a proxy for failure to achieve smoking cessation during their initial treatment episode.   4.2 Methods 4.2.1 Data Collection The data used for this study were obtained from the population-based administrative health claims data from the Canadian province of British Columbia, covering over 4.4 million residents. cxlv-cl The comprehensiveness of the British Columbia administrative database makes it possible to study real-world comparative effectiveness that is representative of routine clinical care. The database included de-identified and linkable patient claims for outpatient prescription medications, physician outpatient visits and services, hospital discharge records, and patient demographics and vital statistics.    4.2.2 Population The source population included all B.C. residents aged >= 18 years. We extracted health claims records for those who used a smoking cessation product between September 30th, 2011 (the date of policy initiation) and March 31st, 2013. Data for individuals were available for the period covering September 30th, 2010 until March 31st, 2015. Patients entered the study on their 54  first use of a smoking cessation product during the Smoking Cessation Program. The date of first use was defined as the Cohort Index Date. Patients were excluded from the study if they had less than 6 months of continuous insurance coverage prior to Cohort Index Date, were < 18 years old on Cohort Index Date, or were dispensed multiple smoking cessation pharmacotherapies on Cohort Index Date. Three study cohorts were created based on first use of varenicline, bupropion, or nicotine replacement therapy during the Program. See Study Flow Diagram in Figure 5.   4.2.3 Study Outcomes and Cohort Follow-up Patients were assumed to have taken a 12-week course of therapy, regardless of the number of prescriptions, days supply, or quantity dispensed. The 12-week course of therapy is consistent with the period of reimbursement covered by the Program and is consistent with the duration of treatment studied in nearly all smoking cessation efficacy RCTs. A 12-week washout period was applied after the 12-week course of therapy to ensure that an extended period of therapy was not misidentified as a subsequent new episode of therapy. See Figure 4. The primary outcome was a prescription dispensing for any smoking cessation product covered by the Program, occurring after the initial course of treatment and washout period. Predictors of treatment failure were also assessed in the smoking cessation study cohorts.  4.2.4 Statistical Analysis This was an observational, retrospective cohort study. Log-binomial regression models were used to assess the relative risk between the study and comparator drugs and outcomes in a one-year and two-year fixed follow-up period. Log-binomial regression was also used to assess 55  selected patient characteristics as predictors of treatment failure. NRT, the most commonly used smoking cessation pharmacotherapy, was selected as the reference medication. To test the robustness of our results, a sensitivity analysis replicated the primary analysis by eliminating the 12-week washout period such that the follow-up period started the day after the initial 12-week course of therapy.  4.2.5 High-Dimensional Propensity Scores High-dimensional propensity scores (HdPS) were calculated for both comparisons (NRT versus varenicline, and NRT versus bupropion) to control for imbalances between the study cohorts. Methods for calculated HdPS have been previously described.cli In addition to the 500 covariates empirically selected by the HdPS algorithms, we also included the following pre-specified covariates based on data available in the 180-days preceding the Cohort Index Date: Alcohol use disorder, diabetes, chronic kidney disease, chronic obstructive pulmonary disease, hypertension, acute myocardial infarction, congestive heart failure, cerebrovascular disease, stroke, unstable angina, and year of cohort entry.  4.3 Results 4.3.1 Study cohort and patient characteristics A detailed comparison of patient characteristics between the study cohorts is shown in Table 4.5.1. The study cohorts were well balanced on age and sex, although the proportion of patients aged 76 and older are higher in NRT (2.1%) compared to varenicline (1.1%) and bupropion (1.8%). A higher proportion of NRT users were from a low-income household, and were diagnosed with excessive alcohol use. NRT users also had a higher rate of diagnosis for 56  depression and anxiety in the year prior to cohort entry (20.5% compared to 15.4% and 17.8% for varenicline and bupropion, respectively), and higher use of medications used to treat depression and anxiety. Medications used to treat hypertension and high cholesterol were well balanced, as was diagnosis for COPD, diabetes, chronic kidney disease, and certain cardiovascular conditions.  Varenicline and bupropion users had a similar proportion of personalized smoking health risk assessments with their general practitioner within three days of Cohort Index, at 9.7% and 10.0%, respectively. A personalized health risk assessment consists of a face-to-face meeting with their general practitioner, where a plan is developed that recommends age and sex specific targeted clinical preventive actions, consistent with primary prevention guidelines.clii   Varenicline users were more likely to have a dispensing for varenicline prior to the Program, and bupropion users were more likely to have a dispensing for bupropion prior to the Program. Prior use of NRT is unknown since it is an over-the-counter product and not recorded in the B.C. Ministry of Health databases prior to the Program.    4.3.2 Relative risk of re-starting smoking cessation therapy Table 4.5.2 shows the risk of re-starting smoking cessation therapy during follow-up. During the 365-day follow-up period, 10.3% of NRT users re-started smoking cessation therapy, compared to 8.7% of varenicline users and 8.4% of bupropion users. The propensity score adjusted risk ratio was 0.87 (95% CI: 0.84 – 0.91) for varenicline users and 0.82 (95% CI: 0.75 – 0.89) for bupropion users. During the 2-year follow-up period, 22.9% of NRT users re-started smoking cessation therapy, compared to 17.0% of varenicline users and 16.6% of bupropion 57  users. The propensity score adjusted risk ratio was 0.77 (95% CI: 0.75 – 0.79) for varenicline users and 0.74 (95% CI: 0.70 – 0.78) for bupropion users.  Table 4.5.5 summarizes the smoking cessation medications used as subsequent therapy, stratified by the initial medication used. NRT users were also separated into Thrive© and Habitrol© (grey shading). The table only includes patients with a subsequent episode of therapy in the one-year or two-year follow-up period. Patients who started on varenicline were equally likely to have a subsequent episode of therapy of varenicline (48%) or NRT (49%). Bupropion starters were most likely to move to NRT (53%) compared to re-using bupropion (33%) or moving to varenicline (13%). Almost all NRT ‘starters’ (including continuers) who reappeared as SC therapy users one year later, after the ineligibility window, were using NRT (89.1%) rather than varenicline (9.2%) or bupropion (1.7%).  4.3.3 Predictors of Re-Starting Therapy The multivariate analysis of resuming SC therapy in relation to demographic and other factors is shown in Table 4.5.3. Age and sex were not significant predictors of resuming coverage of SC therapy. Patients diagnosed with COPD were 15% more likely to resume covered therapy (95% CI: 1.08 – 1.23) and patients diagnosed with depression were 11% more likely to resume (95% CI: 1.05 – 1.17). Patient’s receiving individual counselling from their family physician within 12 months of initial therapy were 6% less likely to resume (95% CI 0.89 – 0.997). Prior use of a prescription smoking cessation medication, and some drugs used to treat hypertension and depression, were all associated with an increased risk of resuming covered therapy.    58  4.3.4 Sensitivity analyses Table 4.5.4 shows the risk of re-starting smoking cessation therapy in the sensitivity analysis with no washout period. During the 365-day follow-up period, 10.8% of NRT users re-started smoking cessation therapy, compared to 9.8% of varenicline users and 10.3% of bupropion users. The propensity score adjusted risk ratio was 0.94 (95% CI: 0.91 – 0.98) for NRT versus varenicline, and 0.97 (95% CI: 0.89 – 1.05) for bupropion versus varenicline.  During the 2-year follow-up period, 23.0% of NRT users re-started smoking cessation therapy, compared to 18.0% of varenicline users and 18.6% of bupropion users. The propensity score adjusted risk ratio was 0.82 (95% CI: 0.80 – 0.84) for NRT versus varenicline, and 0.82 (95% CI: 0.78 – 0.87) for bupropion versus varenicline.    4.4 Discussion 4.4.1 Interpretation of Results Both our primary analysis and sensitivity analysis found NRT users were more likely to have a subsequent episode of therapy after their initial treatment episode during the study period. In our primary analysis, varenicline users were 13% less likely, and bupropion users were 18% less likely, to re-start a subsequent episode of therapy compared to NRT, although the bupropion result was non-significant in our sensitivity analysis which removed the washout period. Without the washout period, patients with a length of therapy extending beyond 12 weeks are considered to have received a subsequent episode of therapy.  We expected the proportion of patients who attempt a second episode of therapy to be the same regardless of initial choice of smoking cessation pharmacotherapies. As a result, NRT can be seen to be associated with greater treatment failures than varenicline or bupropion. 59  The well-balanced study cohorts were likely a result of the Program providing financial coverage for all of the smoking cessation products, which minimized socioeconomic selection bias. This was further evident by the small difference observed between the crude and fully adjusted risk ratios in the results tables.  Personalized smoking health risk assessments were lower in the NRT group within 3 days of Cohort Index (1.3%), which is expected considering NRT does not require a prescription. A recent systematic review found individual counselling may aid in smoking cessation.cliii We use log-binomial regression to estimate relative risk. The odds ratio is often reported and interpreted as a relative risk in cohort studies with a binary outcome, but given the high incidence of outcomes in our study, the odds ratio could have over-stated the relative risk.cliv,clv  We used log-binomial regression to estimate relative risk using a log link function instead of a logit link function (logistic regression). This method has previously been shown to produce an unbiased estimate of the relative risk.clvi   The relative risk of treatment failure in our study cannot be used to estimate abstinence rates, because the number of patients who fail to quit smoking after treatment, and who choose not to try another course of treatment, is unknown. Although treatment failure rates cannot be estimated in our study design, our results suggest NRT may have a higher relative risk of treatment failure compared to varenicline and bupropion.  4.4.2 Comparison with existing literature Several RCTs have found varenicline to have a superior short-term efficacy for increasing the quitting rate compared to NRT and bupropion monotherapy, and some evidence exists for the use of combination therapy. A 2013 Cochrane network meta-analysis found equal 60  efficacy (smoking abstinence of 6 months or longer) between bupropion and NRT, and found varenicline was superior to single forms of NRT (OR 1.59, 95% CI: 1.29-1.96) and to bupropion (OR 1.51, 95% CI: 1.22-1.87), but was not more effective than combination NRT (OR: 1.06, 95% CI: 0.75-1.48).cxxxi Koegelenberg (2014) found combination varenicline plus NRT was associated with a higher continuous abstinence rate at 12 weeks (OR: 1.85, 95% CI: 1.19-2.89) and 24 weeks (OR: 1.98, 95% CI: 1.25-3.14) compared to varenicline monotherapy. Cinciripini (2018) found the addition of bupropion to varenicline therapy is not more effective than varenicline monotherapy.clvii We cannot rule out that combination therapy was a factor in our results, perhaps partially explaining the strong performance of bupropion in our study relative RCT findings. The inability to identify over-the-counter NRT products is a limitation of our data source.  There is limited evidence about the effectiveness of smoking cessation pharmacotherapies as prescribed in routine practice. A prospective cohort study of electronic medical records in the UK found varenicline users were more likely to quit smoking than those prescribed NRT at 2-years, odds ratio 1.26 (95% CI: 1.23 to 1.29).clviii There are no other known observational studies that have assessed the comparative effectiveness of smoking cessation pharmacotherapies.  A recent study by West (2018) examined factors associated with the efficacy of smoking cessation treatments. clix The study found demographic characteristics, such as age, sex, and ethnicity, were not associated with a difference in treatment effectiveness. This was consistent with the findings in our study showing no difference in risk of using a subsequent episode of therapy based on age group or sex. However, a recent meta-analysis found varenicline was more effective for females than for males.clx   61  We found patients diagnosed with or treated for depression were more likely to re-start therapy, whereas the study by West found depression was associated with lower cessation rates. Our study found prior use of bupropion was not associated with a statistically significant difference in risk of subsequent therapy use, and prior use of varenicline was associated with a small increase in risk, RR=1.13 (1.03 – 1.25). These results are consistent with a recent placebo-controlled RCT that found similar effectiveness in smokers that had used varenicline previously compared to drug-naïve smokers.clxi   4.4.3 Limitations The validity of our findings required significant assumptions. The first assumption was that patients who fail on each of the smoking cessation products have an equal probability of resuming therapy within the follow-up period.  Second, we assumed non-differential loss to follow-up. Although low follow-up rates can limit validity and introduce bias if the reasons for loss are related to both the exposure and the outcome, given the similarities in age characteristics between the study cohorts and the stable source population, we do not anticipate a high loss to follow-up. Third, it was not possible to identify the use of any NRT products prior to the Program, or use of non-benefit over-the-counter products during the Program. This has important implications, mainly that we were not able to distinguish between new users and prevalent users, and also that we could not be certain that any course of therapy is not being used in combination with an NRT product that is not captured in our data.  62  4.5  Conclusion Therapy sequencing was used to identify treatment failures to smoking cessation therapy. Based on treatment failures during a drug benefit policy of smoking cessation medications, varenicline and bupropion were more effective aids to smoking cessation than NRT. The therapy sequencing method identified patient characteristics associated with treatment effectiveness.    63  Figures Figure 4 British Columbia Study Period and Follow-up  Figure 5 British Columbia Patient Flow Diagram  64  Tables  Table 4.5.1 Baseline Characteristics of Study Cohorts       N or mean % or SD N or mean % or SD N or mean % or SD N or mean % or SD N or mean % or SDNumber of patients 39,094       5,838         71,510       16,078            55,432            Age (years), mean (SD) 46.5 (13.5) 47.7 (13.7) 46.5 (14.6) 44.9 (14.7) 47 (14.5)     18-55, n (%) 28,597       73.1% 4,087          70.0% 51,307       71.7% 12,005            74.7% 39,302            70.9%     56-75, n (%) 10,082       25.8% 1,647          28.2% 18,709       26.2% 3,820              23.8% 14,889            26.9%     76+, n (%) 415             1.1% 104             1.8% 1,494          2.1% 253                  1.6% 1,241              2.2%Female (n,%) 18,394       47.1% 2,931          50.2% 34,695       48.5% 6,910              43.0% 27,785            50.1%Family Income < $30,000 13,501       34.5% 2,263          38.8% 29,221       40.9% 6,685              41.6% 22,536            40.7%Calendar year of cohort entry, n (%)      2011 15,546       39.8% 2,084          35.7% 23,399       32.7% 4,764              29.6% 18,635            33.6%      2012 19,939       51.0% 3,244          55.6% 40,895       57.2% 9,591              59.7% 31,304            56.5%      2013 3,609          9.2% 510             8.7% 7,216          10.1% 1,723              10.7% 5,493              9.9%Smoking cessation products in prior year:      Varenicline, n (%) 5,740          14.7% 166             2.8% 1,313          1.8% 212                  1.3% 1,101              2.0%      Buproprion, n (%) 1,207          3.1% 715             12.2% 2,939          4.1% 701                  4.4% 2,238              4.0%Personal Smoking Health Risk Assessment:     In prior year 7,737          19.8% 1,151          19.7% 5,536          7.7% 997                  6.2% 4,539              8.2%     In 3 days prior and incl. index date 3,799          9.7% 582             10.0% 916             1.3% 140                  0.9% 776                  1.4%Prescriptions in prior year:Statins, n (%) 3,384          8.7% 580             9.9% 5,860          8.2% 1,137              7.1% 4,723              8.5%ACE inhibitors, n (%) 3,318          8.5% 538             9.2% 5,664          7.9% 1,122              7.0% 4,542              8.2%Beta Blockers, n (%) 2,367          6.1% 380             6.5% 4,388          6.1% 843                  5.2% 3,545              6.4%Calcium Channel Blockers, n (%) 1,825          4.7% 290             5.0% 3,104          4.3% 591                  3.7% 2,513              4.5%Loop or thiazide diuretics, n (%) 1,409          3.6% 210             3.6% 2,444          3.4% 483                  3.0% 1,961              3.5%Fibrates, n (%) 147             0.4% 22                0.4% 260             0.4% 54                    0.3% 206                  0.4%SSRIs, n (%) 3,598          9.2% 576             9.9% 7,866          11.0% 1,730              10.8% 6,136              11.1%SNRIs, n (%) 1,718          4.4% 239             4.1% 3,589          5.0% 810                  5.0% 2,779              5.0%TCAs, n (%) 457             1.2% 67                1.1% 926             1.3% 201                  1.3% 725                  1.3%Medical diagnosis in prior year:Excessive alcohol use, n (%) 1,042          2.7% 142             2.4% 2,856          4.0% 652                  4.1% 2,204              4.0%Depression/Anxiety, n (%) 6,025          15.4% 1,037          17.8% 14,686       20.5% 3,429              21.3% 11,257            20.3%COPD, n (%) 4,060          10.4% 620             10.6% 6,897          9.6% 1,066              6.6% 5,831              10.5%Acute Myocardial Infarction, n (%) 251             0.6% 47                0.8% 1,011          1.4% 93                    0.6% 918                  1.7%Congestive Heart Failure, n (%) 424             1.1% 71                1.2% 1,254          1.8% 156                  1.0% 1,098              2.0%Cerebrovascular Disease, n (%) 520             1.3% 96                1.6% 1,223          1.7% 193                  1.2% 1,030              1.9%Peripheral Vascular Disease, n (%) 742             1.9% 126             2.2% 1,543          2.2% 250                  1.6% 1,293              2.3%Coronary Revascularization, n (%) 157             0.4% 36                0.6% 746             1.0% 63                    0.4% 683                  1.2%Angina, n (%) 703             1.8% 117             2.0% 1,604          2.2% 265                  1.6% 1,339              2.4%Diabetes, n (%) 3,769          9.6% 583             10.0% 6,467          9.0% 1,304              8.1% 5,163              9.3%Chronic Kidney Disease, n (%) 419             1.1% 79                1.4% 928             1.3% 161                  1.0% 767                  1.4%Habitrol®CharacteristicsVarenicline Bupropion NRT Thrive®65   Table 4.5.2 Risk of Treatment Failure: Re-Starting Smoking Cessation Therapy     RR 95% CI  RR 95% CI1-Year Treatment FailureNRT 7,372          10.3% 1.00              Reference 1.00              ReferenceVarenicline 3,413          8.7% 0.85              (0.81 - 0.88) 0.87              (0.84 - 0.91)Bupropion 492             8.4% 0.82              (0.75 - 0.89) 0.82              (0.75 - 0.89)2-year Treatment FailureNRT 16,361       22.9% 1.00              Reference 1.00              ReferenceVarenicline 6,641          17.0% 0.74              (0.72 - 0.76) 0.77              (0.75 - 0.79)Bupropion 971             16.6% 0.73              (0.69 - 0.77) 0.74              (0.70 - 0.78)Crude and Adjusted Log-Binomial Regression ModelsNumber of users re-starting therapyCrude Risk Ratio (RR)  Adjusted Risk Ratio (RR) 66  Table 4.5.3 Predictors of Treatment Failure in British Columbia      Risk RatioAgeAge 56-75 (reference) 1.00               -Age 18-55 1.03               0.995     - 1.07  Age 76+ 0.92               0.87       - 0.97  Patient is female 1.01               0.98       - 1.04  Patient is from a low income household 1.04               1.01       - 1.08  DiagnosisCOPD 1.12               1.06       - 1.18  Diabetes 1.01               0.95       - 1.08  Depression 1.03               0.98       - 1.08  Myocardial infarction 0.87               0.69       - 1.11  Unstable angina 1.03               0.90       - 1.18  Cerebrovascular disease 0.84               0.72       - 0.99  CHF 1.01               0.87       - 1.17  Chronic kidney disease 0.89               0.75       - 1.06  Coronary revascularization 0.95               0.72       - 1.27  Prior Drug UseBupropion 1.07               0.99       - 1.16  Varenicline 1.30               1.21       - 1.40  Ace Inhibitor 1.00               0.95       - 1.05  Beta Blocker 1.03               0.98       - 1.08  Calcium Channel Blocker 1.08               1.02       - 1.15  Diuretic 1.00               0.94       - 1.06  Statin 1.04               0.99       - 1.09  TCA 1.09               1.02       - 1.15  SNRI 0.98               0.93       - 1.03  SSRI 1.05               1.02       - 1.10  MAOI 1.18               0.73       - 1.91  Predictors of Re-Starting TherapyCharacteristic 95% CI67  Table 4.5.4 Sensitivity Analysis: Re-Starting Smoking Cessation Therapy (No Washout Period)    Table 4.5.5 Medication Matrix of Patients Who Re-Start Therapy       RR 95% CI  RR 95% CI1-Year Treatment FailureNRT 7,697          10.8% 1.00              Reference 1.00              ReferenceVarenicline 3,816          9.8% 0.91              (0.87 - 0.94) 0.94              (0.91 - 0.98)Bupropion 604             10.3% 0.96              (0.89 - 1.04) 0.97              (0.89 - 1.05)2-year Treatment FailureNRT 16,445       23.0% 1.00              Reference 1.00              ReferenceVarenicline 7,035          18.0% 0.78              (0.76 - 0.80) 0.82              (0.80 - 0.84)Bupropion 1,086          18.6% 0.81              (0.77 - 0.86) 0.82              (0.78 - 0.87)Number of users re-starting therapyCrude Risk Ratio (RR)  Adjusted Risk Ratio (RR) Crude and Adjusted Log-Binomial Regression ModelsTherapy Re-Starting, 365 day follow-upn % n % n % n % n %Initial TherapyVarenicline 1,653         48% 86              3% 1,674         49% 349            10% 1,325         39%Bupropion 66              13% 163            33% 263            53% 63              13% 200            41%NRT 678            9% 126            2% 6,568         89% 1,896         26% 4,672         63%Thrive 106            6% 21              1% 1,602         93% 1,167         67% 435            25%Habitrol 572            10% 105            2% 4,966         88% 729            13% 4,237         75% Therapy Re-Starting, 730 day follow-upn % n % n % n % n %Initial TherapyVarenicline 3,042         46% 182            3% 3,417         51% 735            11% 2,682         40%Bupropion 148            15% 264            27% 559            58% 128            13% 431            44%NRT 1,401         9% 258            2% 14,702       90% 4,417         27% 10,285       63%Thrive 223            6% 41              1% 3,603         93% 2,644         68% 959            25%Habitrol 1,178         9% 217            2% 11,099       89% 1,773         14% 9,326         75%HabitrolThriveNRTBupropionVareniclineTherapy Re-startTherapy Re-startVarenicline Bupropion NRT Thrive Habitrol68  Chapter 5: Concluding Chapter This thesis examined the comparative safety and effectiveness of medications commonly used to aid smoking cessation. The thesis focused on population-based data on safety of varenicline, bupropion, and nicotine replacement therapy (NRT) during a government-sponsored reimbursement program in the Canadian province of British Columbia, and in a large U.S. population of government and employer-insured patients. It also described and implemented a novel method of measuring comparative effectiveness using drug therapy re-initiation as a proxy for treatment failure.  The British Columbia Smoking Cessation Program provided a natural experiment in a jurisdiction with a population-based linkable database capturing inpatient, outpatient, and pharmaceutical dispensing records for all residents of the province. The U.S. MarketScan® database contains health records for over 100 million people, making our study the largest, to our knowledge, conducted on this topic. The results of this thesis will aid physicians, patients, and policy-makers to make informed choices regarding smoking cessation pharmacotherapy.  5.1 Varenicline was not associated with an increased risk of hospitalization for a cardiovascular event compared to nicotine replacement therapies (NRT).  Despite the short-term efficacy evidence supporting use of varenicline in clinical trials, prescribing of varenicline has been hindered by cardiovascular safety concerns, particularly the 2011 FDA product warnings.clxii Our analysis of U.S. administrative data, in adults with no history of depression, found varenicline was associated with a 20% (0.75 – 0.85) lower 1-year risk of hospital admission for a cardiovascular event. Results were less conclusive in our population-based analysis during the British Columbia Smoking Cessation Program. We found 69  varenicline was associated with a non-significant 10% (0.82 – 1.00) decreased risk of hospital admission for a cardiovascular event, which was not inconsistent with the US MarketScan result of 20% (0.75 – 0.85). An inverse variance weighted average of the two results (Table 5.6.1) shows varenicline was associated with an 18% (0.77 – 0.87) decreased one-year risk of hospitalization for a cardiovascular event. Therefore, there was no increased cardiovascular risk associated with varenicline compared to NRT, and possibly some cardiovascular safety benefit. It is not known if this is a drug effect or the result of less exposure to tobacco among successful quitters. Also, smoking cessation might include nicotine dose reduction as cigarette smokers manage to switch to non-tobacco forms of smoking (e.g. vaping).  Clinical and policy implications Varenicline and NRT are the most commonly used first-line medications to aid smoking cessation. Our analyses found no evidence that the use of varenicline increased serious cardiovascular events during, or shortly after treatment, in the general population of smokers seeking pharmacotherapy to aid smoking cessation. Based on the results of this thesis, there is no reason for clinicians or policy makers to favour NRT over varenicline in terms of cardiovascular safety.   5.2 Varenicline was associated with a decreased risk of hospitalization for a neuropsychiatric event compared to nicotine replacement therapies (NRT).  We found a statistically significant decreased risk of hospitalization for a neuropsychiatric event in both the B.C. and U.S. analyses, for varenicline compared to NRT. Study cohorts in B.C. were well balanced on age and sex, but NRT users were more likely to have a prior depression diagnosis. The US study excluded users with a history of depression, but 70  NRT users were on average older, with more than twice as many users in the oldest age category, and more likely to have a history of alcohol abuse. Although these potential risk factors for neuropsychiatric events were adjusted for in our statistical models, residual confounding can’t be ruled out. The results, however, are consistent with findings from two other observational studies in the UK general practice that found a significantly reduced risk of depression and self harm in varenicline users comparison with NRT.clxiii,clxiv   Clinical and policy implications Neuropsychiatric safety concerns have been raised related to the use of varenicline for smoking cessation, although limited evidence exists directly comparing varenicline with NRT. This thesis addressed the need for a large real-world safety study with rigorous confounding adjustment methods and an active comparator. The results of this study found varenicline to be associated with a lower risk of neuropsychiatric events in the general population of smokers in BC, and in a non-depressed adult population in the US. The study did not address the neuropsychiatric safety of varenicline compared to NRT in adults with psychiatric disorders, where evidence suggests there may be an association.clxv Further research into this subgroup of smokers is needed.  5.3 There was no risk difference in cardiovascular or neuropsychiatric events between bupropion and NRT users.   In our population-based study of adult users of smoking cessation pharmacotherapy in B.C., we found no significant difference in risk between bupropion and NRT for cardiovascular hospitalizations, neuropsychiatric hospitalizations, or mortality. Bupropion accounts for approximately 5% of first-line prescribing for the smoking cessation market in B.C. The cohort 71  is well balanced on age, sex, and comorbidities, compared to NRT.  The results were consistent in 6-month and 1-year follow-up periods, and also in sensitivity analysis of patients with and without a recent history of cardiovascular events.  The safety results in the U.S. analysis were likely flawed due to exposure misclassification. It was difficult to distinguish between generic bupropion used for depression versus smoking cessation, even with the exclusion of patients with a history of depression and limiting bupropion cohort inclusion to the 150mg sustained release formulation. The high number of female bupropion patients (61%) versus NRT (47%) supports the assumption that exposure misclassification was an issue because we would not expect a disproportionate sex ratio based on bupropion prescribing for smoking cessation.  Clinical and policy implications Based on the results of this thesis, there is no reason for clinicians or policy makers to favour bupropion or NRT, in terms of cardiovascular safety, neuropsychiatric safety, or mortality.  5.4 Bupropion users compared to varenicline users were associated with a lower risk of cardiovascular events, and a higher risk of neuropsychiatric events  In a direct comparison of bupropion with varenicline in our B.C. study, we found no difference in the 1-year risk of hospitalization for cardiovascular events (RR: 0.99, 95% CI: [0.79 – 1.23]) and a non-significant 15% increased risk for neuropsychiatric events (RR: 1.15, 95% CI: [0.91 – 1.46]). In our U.S. analysis, we found bupropion was associated with a 17% (RR: 0.83, 95% CI: [0.78 – 0.87]) lower 1-year risk of hospital admission for cardiovascular 72  events and a 77% increased risk of neuropsychiatric events (RR: 1.77, 95% CI: [1.66 – 1.88]) compared to varenicline.  Combined results from the B.C. and U.S. analysis using inverse variance weighting (Tables 5.6.2 and 5.6.3) found bupropion was associated with a 67% increased 1-year risk of neuropsychiatric events (RR: 1.67, 95% CI: [1.46 – 1.91]), and a 16% decreased 1-year risk of cardiovascular events (RR: 0.84, 95% CI: [0.79 – 0.89]).  Clinical and policy implications Based on the results of this thesis, neither bupropion nor varenicline have a superior safety profile. Varenicline is associated with a lower risk of neuropsychiatric events and bupropion is associated with a lower risk of cardiovascular events.  5.5 Drug episode sequencing can be a useful method of identifying treatment failure in comparative effectiveness research. We described a novel method of using successive episodes of drug therapy as a proxy for treatment failure. We implemented the method in the context of comparative effectiveness of smoking cessation use during a government sponsored reimbursement program, which provided a unique opportunity to identify non-prescription NRT use in a population-based administrative claims database. The comparative effectiveness results were comparable to findings in other studies, although evidence of comparative effectiveness of smoking cessation pharmacotherapies in a real-world setting using an active comparator is limited.  We identified several important assumptions that need to hold true for validity of the results. Further investigation, linking additional data such as survey data on smoking status, would be beneficial to explore the reliability of this method, and its potential use in other comparative effectiveness research settings.    73   Research implications The use of drug sequencing algorithms to identify treatment failure has particular benefit when “hard data” on treatment failure is unavailable to researchers. The method, as applied in this thesis, produced reasonable results in terms of comparability with existing literature, and should be considered for future use in comparative effectiveness drug research.    5.6 Future Research The following topics present gaps in knowledge identified in this thesis that would be beneficial from the perspective of a researcher, clinician, or formulary decision-maker.  1. Efficacy of combination NRT therapy. Some evidence exists showing that combination NRT therapy may be more efficacious than monotherapy varenicline, bupropion, or NRT. Differences in pharmacokinetics between NRT forms may play a role in improved efficacy. Evidence in a real-world setting is needed.  2. The B.C. analysis of neuropsychiatric events did not stratify patients with and without a history of psychiatric disorders, and the US analysis excluded patients diagnosed or treated for depression. There is some evidence that patients with psychiatric disorders may be at higher risk of a neuropsychiatric event with varenicline or bupropion. This was not answered in our analyses and should be evaluated further.  3. The episode of therapy sequencing method could be further validated if patient reported smoking data could be linked to the prescription claims data. This would allow researchers to examine important assumptions, such as the proportion of patients who 74  attempt a second episode of therapy to be non-differential based on their initial choice of smoking cessation pharmacotherapies. 4. There is no evidence pertaining to the long-term comparative effectiveness of smoking cessation therapies (5+ years). This is critically important information for patients, clinicians, and policy-makers, considering tobacco addiction is the leading cause of preventable deaths worldwide. With improvements in administrative health databases and health survey data capturing lifestyle behaviours, this research is essential.    75  Tables  Table 5.6.1 Risk of Cardiovascular Events; Combined British Columbia and US MarketScan Inverse Variance Weighted Average, Varenicline versus NRT   Table 5.6.2 Risk of Cardiovascular Events; Combined British Columbia and US MarketScan, Inverse Variance Weighted Average, Bupropion versus Varenicline   Table 5.6.3 Risk of Neuropsychiatric Events; Combined British Columbia and US MarketScan, Inverse Variance Weighted Average, Bupropion versus Varenicline    76  References i U.S. Department of Health and Human Services. How Tobacco Smoke Causes Disease: What It Means to You. Atlanta: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, 2010 ii Health Canada. https://www.canada.ca/en/health-canada/services/smoking-tobacco/effects-smoking/smoking-your-body/risks-smoking.html. Accessed March 14, 2019. iii U.S. Department of Health and Human Services. The Health Consequences of Smoking – 50 years of Progress: A Report of the Surgeon General. Atlanta: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, national Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, 2014 iv Centers for Disease Control and Prevention. Current Cigarette Smoking Among Adults—United States, 2005–2016. Morbidity and Mortality Weekly Report 2018;67(2):53-9 [accessed 2018 Oct 15]. v Statistics Canada. Health Fact Sheet – Smoking, 2017. https://www150.statcan.gc.ca/n1/en/pub/82-625-x/2018001/article/54974-eng.pdf?st=nknale2P. Accessed March 29, 2019. Catalogue no. 82-625-X. ISSN 1920-9118. vi Shiffman S, Rolf C, Hellebusch S, et al. Real-world efficacy of prescription and over-the-counter nicotine replacement therapy. Addiction, May 2002, Volume 97, Issue 5. vii US Food and Drug Administration. Center for Drug Evaluation and Research. https://www.accessdata.fda.gov/drugsatfda_docs/nda/97/20711_ZYBAN,%20100MG%20AND%20150MG_APPROV.PDF. Accessed March 29, 2019.  Department of Health & Human Services. Approval Letter. 1997. viii US Food and Drug Administration. Center for Drug Evaluation and Research. https://www.accessdata.fda.gov/drugsatfda_docs/nda/97/20711_ZYBAN,%20100MG%20AND%20150MG_APPROV.PDF. Accessed March 29, 2019.  Department of Health & Human Services. Approval Letter. 2006. ix Motooka Y, Matsui T, Slaton RM, et al. Adverse events of smoking cessation treatments (nicotine replacement therapy and non-nicotine prescription medication) and electronic cigarettes in the Food and Drug Administration Adverse Event Reporting System, 2004−2016. SAGE Open Med. 2018; 6: 2050312118777953. Published online 2018 May 21. doi:  10.1177/2050312118777953 x Moore T, Furberg CD, Glenmullen J, et al. Suicidal Behavior and Depression in Smoking Cessation Treatments. PLoS One. 2011; 6(11): e27016. Published online 2011 Nov 2. xi Foulds J, Russ C, Yu CR, et al. Effect of varenicline on individual nicotine withdrawal symptoms: a combined analysis of eight randomized, placebo-controlled trials. Nicotine Tob Res. 2013 Nov;15(11):1849-57. doi: 10.1093/ntr/ntt066. Epub 2013 May 21. xii Ciciripini PM, Robinson JD, Karam-Hage M, et al. Effects of varenicline and bupropion sustained-release use plus intensive smoking cessation counseling on prolonged abstinence from smoking and on depression, negative affect, and other symptoms of nicotine withdrawal. JAMA Psychiatry. 2013 May;70(5):522-33. doi: 10.1001/jamapsychiatry.2013.678. xiii Anthenelli RM, Morris C, Ramey TS, et al. Effects of varenicline on smoking cessation in adults with stably treated current or past major depression: a randomized trial. Ann Intern Med. 2013 Sep 17;159(6):390-400. doi: 10.7326/0003-4819-159-6-201309170-00005. xiv Garza D, Murphy M, Tseng LJ, et al. A Double-Blind Randomized Placebo-Controlled Pilot Study of Neuropsychiatric Adverse Events in Abstinent Smokers Treated with Varenicline or Placebo. Biological Psychiatry. Volume 69, Issue 11, June 2011, pg 1075-1082. xv Benowitz NL. Pharmacology of nicotine: addiction, smoking-induced disease, and therapeutics. Annu Rev Pharmacol Toxicol. 2009;49:57-71. doi: 10.1146/annurev.pharmtox.48.113006.094742.   77   xvi Benowitz N. Nicotine Addiction. N Engl J Med. 2010 Jun 17;362(24):2295-2303. doi: 10.1056/NEJMra0809890 xvii De Biasi M, Dani J. Reward, Addiction, Withdrawal to Nicotine. Annu Rev Neurosci. 2011; 34: 105–130. doi: 10.1146/annurev-neuro-061010-113734 xviii Faessel HM, Obach RS, Rollema H, et al. A review of the clinical pharmacokinetics and pharmacodynamics of varenicline for smoking cessation. Clin Pharmacokinet. 2010 Dec;49(12):799-816. doi: 10.2165/11537850-000000000-00000. xix Nakamura M, Oshima A, Fujimoto Y, et al. Efficacy and tolerability of varenicline, an alpha4beta2 nicotinic acetylcholine receptor partial agonist, in a 12-week, randomized, placebo-controlled, dose-response study with 40-week follow-up for smoking cessation in Japanese smokers. Clin Ther. 2007 Jun;29(6):1040-56.  xx Lam S, Patel PN. Varenicline: a selective alpha4beta2 nicotinic acetylcholine receptor partial agonist approved for smoking cessation.  xxi Carrozzi L, Pistelli F, Viegi G. Pharmacotherapy for smoking cessation. Ther Adv Respir Dis. 2008 Oct;2(5):301-17. doi: 10.1177/1753465808096136. xxii Stahl SM, Pradko J, Haight BR, et al. A review of the Neuropharmacology of Bupropion, a Dual Norepinephrine and Dopamine Reuptake Inhibitor. Prim Care Companion J Clin Psychiatry. 2004; 6(4): 159-166. xxiii Slemmer JE, Martin BR, Damaj MI. Bupropion is a nicotinic antagonist. J Pharmacol Exp Ther. 2000 Oct;295(1):321-7. xxiv Kotlyar M, Brauer LH, Tracy TS, et al. Inhibition of CYP2D6 activity by bupropion. J Clin Psychopharmacol. 2005 Jun;25(3):226-9. xxv Ferry LH, Non-nicotine pharmacotherapy for smoking cessation. Prim Care. 1999 Sep;26(3):653-69. xxvi Fiore, M, Jae´n C, Baker T, et al. (2008) Treating Tobacco Use and Dependence: 2008 Update. Clinical Practice Guideline. Rockville, MDUS Department of Health and Human Services, Public Health Service. May 2008. xxvii Tonnesen P, Carrozzi L, Fagerstrom K, et al. Smoking cessation in patients with respiratory diseases: a high priority, integral component of therapy, Eur Respir. Eur Respir J 2007. 29: 390–417. xxviii Hughes JR, Stead LF, Hartmann-Boyce J, et al. Antidepressants for smoking cessation. Cochrane Database Syst Rev. 2014 Jan 8;(1):CD000031. doi: 10.1002/14651858.CD000031.pub4. Review. xxix Rissmiller D, Campo T. Extended-Release Bupropion Induced Grand Mal Seizures. The Journal of the American Osteopathic Association, October 2007, Vol. 107, 441-442. xxx Silagy C, Lancaster T, Stead L, et al. Nicotine replacement therapy for smoking cessation. Cochrane Database Syst Rev. 2004;(3):CD000146. xxxi Henningfield J. Nicotine Medications for Smoking Cessation. N Engl J Med 1995; 333:1196-1203 DOI  10.1056/NEJM199511023331807 xxxii Aubin HJ, Luqiens A, Berlin I. Pharmacotherapy for smoking cessation: pharmacological principles and clinical practice. Br J Clin Pharmacol. 2014 Feb; 77(2): 324–336. Published online 2014 Jan 22. doi: 10.1111/bcp.12116. xxxiii Henningfield JE, Keenan RM. Nicotine delivery kinetics and abuse liability. J Consult Clin Psychol 1993;61:743-750. xxxiv Hartmann-Boyce J, Chepkin S, Ye W, et al. Nicotine replacement therapy versus control for smoking cessation. Cochrane Database Syst Rev. 31 May 2018. DOI: 10.1002/14651858.CD000146.pub5  xxxv Hansen, Leigh. The Truven Health MarketScan Databases for life sciences researchers. March 2017. https://truvenhealth.com/Portals/0/Assets/2017-MarketScan-Databases-Life-Sciences-Researchers-WP.pdf. Accessed Feb 25, 2019. xxxvi Petersen M, Deddens J. A comparison of two methods for estimating prevalence ratios. BMC Med Res Methodol. 2008: 8: 9. Published online 2008 Feb 28. doi: 10.1186/1471-2288-8-9  78   xxxvii Ranganathan, P, Aggarwal R, Pramesh C. Common pitfalls in statistical analysis: Odds versus risk. Perspect Clin Res. 2015 Oct-Dec; 6(4): 222-224. xxxviii Knol M, Le Cessie S, Algra A, et al. Overestimation of risk ratios by odds ratios in trials and cohort studies: alternatives to logistic regression. CMAJ. 2012 May 15: 184(8): 895-899. xxxix Blizzard L, Hosmer DW. Parameter estimation and goodness-of-fit in log binomial regression. Biom J. 2006 Feb;48(1):5-22. xl Zhang J, Yu KF. What’s relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA 1998;280:1690-1. xli McNutt L-A, Wu C, Xue X, et al. Estimating the Relative Risk in Cohort Studies and Clinical Trials of Common Outcomes. American Journal of Epidemiology, Volume 157, Issue 10, 15 May 2003, Pages 940–943, https://doi.org/10.1093/aje/kwg074 xlii McNutt LA, Hafner JP, Xue X. Correcting the odds ratio in cohort studies of common outcomes. (Letter). JAMA 1999;282:529. xliii Williamson T, Eliasziw M, Fick G. Log-binomial models: exploring failed convergence. Emerg Themes Epidemiol. 2013; 10: 14. Published online 2013 Dec 13. doi: 10.1186/1742-7622-10-14 xliv Cox D.R. Regression Models and Life-Tables. Journal of the Royal Statistical Society. Series B (Methodological) Vol. 34, No. 2 (1972). xlv Lirette ST, Aban I. Quantifying predictive accuracy in survival models. J Nucl Cardiol. 2017 Dec;24(6):1998-2003. doi: 10.1007/s12350-015-0296-z. Epub 2015 Oct 19. xlvi Mittal S, Madigan D. high-dimensional, massive sample-size Cox propotional hazards regression for survival analysis. Biostatistics. 2014 Apr; 15(2): 207–221. Published online 2013 Oct 4. Doi: 10.1093/biostatistics/kxt043 xlvii In J, Lee D. Survival analysis: Part I – analysis of time-to-event. Korean J Anesthesiol. 2018 Jun; 71(3): 182–191. Published online 2018 May 17. doi: 10.4097/kja.d.18.00067 xlviii George B, Seals S, Aban I. Survival analysis and regression models. J Nucl Cardiol. 2014 Aug; 21(4): 686–694. Published online 2014 May 9. doi: 10.1007/s12350-014-9908-2 xlix Klein JP, Moeschberger ML. Survival analysis: Techniques for censored and truncated data. New York: Springer; 2003. l Austin, Peter. An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivariate Behav Res. 2011 May; 46(3): 399-424. doi: 10.1080/00273171.2011.568786 li Ali MS, Rolf G, Klungel O. Best (but oft-forgotten) practices: propensity score methods in clinical nutrition research. The American Journal of Clinical Nutrition, Volume 104, Issue 2, August 2016, Pages 247-258.  lii U.S. Department of Health and Human Services. How Tobacco Smoke Causes Disease: What It Means to You. Atlanta: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, 2010 liii U.S. Department of Health and Human Services. The Health Consequences of Smoking – 50 years of Progress: A Report of the Surgeon General. Atlanta: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, national Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, 2014 liv Centers for Disease Control and Prevention. Current Cigarette Smoking Among Adults—United States, 2005–2016. Morbidity and Mortality Weekly Report 2018;67(2):53-9 [accessed 2018 Oct 15]. lv Stead L. Combined pharmacotherapy and behavioural interventions for smoking cessation. Cochrane Database of Systematic Reviews. March 2016. 10.1002/14651858.CD008286.pub3 lvi Stahl SM, Pardko JF, Haight BR, et al. A review of the Neuropharmacology of Bupropion, a Dual Norepinephrine and Dopamine Reuptake Inhibitor. Prim Care Companion J Clin Psychiatry. 2004; 6(4): 159-166. lvii Slemmer JE, Martin BR, Damaj MI. Bupropion is a nicotinic antagonist. J Pharmacol Exp Ther. 2000 Oct;295(1):321-7.  79   lviii Littlewood RA, Claus ED, Wilcox CE, et al. Moderators of smoking cessation outcomes in a randomized-controlled trial of varenicline versus placebo. Psychopharmacology (Berl). 2017 Dec;234(23-24):3417-3429. doi: 10.1007/s00213-017-4721-7. Epub 2017 Sep 9. lix Gray KM, McClure EA, Baker NL, et al. An exploratory short-term double-blind randomized trial of varenicline versus nicotine patch for smoking cessation in women. Addiction. 2015 Jun;110(6):1027-34. doi: 10.1111/add.12895. Epub 2015 Mar 29 lx Ebbert JO, Hughes JR, West RJ, et al. Effect of varenicline on smoking cessation through smoking reduction: a randomized clinical trial. JAMA. 2015 Feb 17;313(7):687-94. doi: 10.1001/jama.2015.280. lxi Tashkin DP, Rennard S, Hays JT, et al. Effects of varenicline on smoking cessation in patients with mild to moderate COPD: a randomized controlled trial. Chest. 2011 Mar;139(3):591-599. doi: 10.1378/chest.10-0865. Epub 2010 Sep 23. lxii Fagerström K, Gilljam H, Metcalfe M, et al. Stopping smokeless tobacco with varenicline: randomised double blind placebo controlled trial. BMJ 2010;341:c6549 lxiii Nakamura M, Oshima A, Fujimoto Y, et al. Efficacy and tolerability of varenicline, an alpha4beta2 nicotinic acetylcholine receptor partial agonist, in a 12-week, randomized, placebo-controlled, dose-response study with 40-week follow-up for smoking cessation in Japanese smokers. Clin Ther 2007;29:1040–56. lxiv Oncken C, Gonzales D, Nides M, et al. Efficacy and safety of the novel selective nicotinic acetylcholine receptor partial agonist, varenicline, for smoking cessation. Arch Intern Med 2006; 166:1571–7 lxv Aubin HJ, Bobak A, Britton JR, et al. Varenicline versus transdermal nicotine patch for smoking cessation: results from a randomised, open-label trial. Thorax 2008;63:717–24. lxvi Jorenby DE, Hays JT, Rigotti NA, et al.; Varenicline Phase 3 Study Group. Efficacy of varenicline, an alpha4beta2 nicotinic acetylcholine receptor partial agonist, vs placebo or sustained-release bupropion for smoking cessation: a randomized controlled trial. JAMA 2006;296:56–63 lxvii Nides M, Oncken C, Gonzales D, et al. Smoking cessation with varenicline, a selective alpha4beta2 nicotinic receptor partial agonist: results from a 7-week, randomized, placebo- and bupropion-controlled trial with 1-year follow-up. Arch Intern Med 2006;166:1561–8. lxviii Gonzales D, Rennard SI, Nides M, et al. Varenicline, an alpha4beta2 nicotinic acetylcholine receptor partial agonist, vs sustained-release bupropion and placebo for smoking cessation: a randomized controlled trial. JAMA. 2006 Jul 5;296(1):47-55. lxix Motooka Y, Matsui T, Slaton RM, et al. Adverse events of smoking cessation treatments (nicotine replacement therapy and non-nicotine prescription medication) and electronic cigarettes in the Food and Drug Administration Adverse Event Reporting System, 2004−2016. SAGE Open Med. 2018; 6: 2050312118777953. Published online 2018 May 21. doi:  10.1177/2050312118777953 lxx Moore T, Furberg CD, Glenmullen J, et al. Suicidal Behavior and Depression in Smoking Cessation Treatments. PLoS One. 2011; 6(11): e27016. Published online 2011 Nov 2. lxxi U.S. Food and Drug Administration. Highlights of Prescribing Information. Chantix (varenicline). https://www.accessdata.fda.gov/drugsatfda_docs/label/2013/021928s030lbl.pdf. Accessed Dec 27, 2017. lxxii U.S. Food and Drug Administration. Highlights of Prescribing Information. Zyban (bupropion hydrochloride). https://www.accessdata.fda.gov/drugsatfda_docs/label/2014/020711s042lbl.pdf. Accessed Dec 27, 2017. lxxiii Foulds J, Russ C, Yu CR, et al. Effect of varenicline on individual nicotine withdrawal symptoms: a combined analysis of eight randomized, placebo-controlled trials. Nicotine Tob Res. 2013 Nov;15(11):1849-57. doi: 10.1093/ntr/ntt066. Epub 2013 May 21. lxxiv Ciciripini PM, Robinson JD, Karam-Hage M, et al. Effects of varenicline and bupropion sustained-release use plus intensive smoking cessation counseling on prolonged abstinence from smoking and on depression, negative affect, and other symptoms of nicotine withdrawal. JAMA Psychiatry. 2013 May;70(5):522-33. doi: 10.1001/jamapsychiatry.2013.678.  80   lxxv Anthenelli RM, Morris C, Ramsey TS, et al. Effects of varenicline on smoking cessation in adults with stably treated current or past major depression: a randomized trial. Ann Intern Med. 2013 Sep 17;159(6):390-400. Doi: 10.7326/0003-4819-159-6-201309170-00005. lxxvi Garza D, Murphy M, Tseng LJ, et al. A Double-Blind Randomized Placebo-Controlled Pilot Study of Neuropsychiatric Adverse Events in Abstinent Smokers Treated with Varenicline or Placebo. Biological Psychiatry. Volume 69, Issue 11, June 2011, pg 1075-1082. lxxvii Svanstrom H, Pasternak B, Hviid A, et al. Use of varenicline for smoking cessation and risk of serious cardiovascular events: nationwide cohort study. BMJ. 2012 Nov 8;345:e7176. doi: 10.1136/bmj.e7176. lxxviii Davies NM, Taylor GMJ, Taylor AE, et al. The effects of prescribing varenicline on two‐year health outcomes: an observational cohort study using electronic medical records. Addiction. 2018 Jun;113(6):1105-1116. doi: 10.1111/add.14146. Epub 2018 Feb 20. lxxix Kotz D, Viechtbauer W, Simpson CR, et al. Cardiovascular and neuropsychiatric risks of varenicline and bupropion in smokers with chronic obstructive pulmonary disease. Thorax. 2017 Oct;72(10):905-911. doi: 10.1136/thoraxjnl-2017-210067. lxxx Cunningham FE, Hur K, Dong D, et al. A comparison of neuropsychiatric adverse events during early treatment with varenicline or a nicotine patch. Addiction. 2016 Jul;111(7):1283-92. doi: 10.1111/add.13329. lxxxi Pasternak B, Svanstrom H, Hviid A. Use of varenicline versus bupropion and risk of psychiatric adverse events. Addiction. 2013 Jul;108(7):1336-43. doi: 10.1111/add.12165. Epub 2013 Apr 4. lxxxii Anthenelli RM, Benowitz NL, West R, et al. Neuropsychiatric safety and efficacy of varenicline, bupropion, and nicotine patch in smokers with and without psychiatric disorders (EAGLES): a double-blind, randomised, placebo-controlled clinical trial. Lancet. 2016 Apr 19:e1-e14.[Epub ahead of print]. lxxxiii Benowitz NL, Pipe A, West R, et al. Cardiovascular Safety of Varenicline, Bupropion, and Nicotine Patch in Smokers: A Randomized Clinical Trial. JAMA Intern Med. 2018 May 1;178(5):622-631. doi: 10.1001/jamainternmed.2018.0397. lxxxiv Kotz D, Viechtbauer W, Simpson C, et al. Cardiovascular and neuropsychiatric risks of varenicline: a retrospective cohort study. Lancet Respir Med. 2015 Oct;3(10):761-8. doi: 10.1016/S2213-2600(15)00320-3. Epub 2015 Sep 6. lxxxv Davies N, Taylor G, Taylor AE, et al. Cardiovascular and neuropsychiatric risks of varenicline: too good to be true? Lancet Respir Med. 2015 Dec;3(12):e39-40. doi: 10.1016/S2213-2600(15)00468-3 lxxxvihttps://www.fda.gov/downloads/advisorycommittees/committeesmeetingmaterials/drugs/psychopharmacologicdrugsadvisorycommittee/ucm520103.pdf. Accessed Oct 16, 2018. lxxxvii Hansen, Leigh. The Truven Health MarketScan Databases for life sciences researchers. March 2017. https://truvenhealth.com/Portals/0/Assets/2017-MarketScan-Databases-Life-Sciences-Researchers-WP.pdf. Accessed Feb 25, 2019. lxxxviii Walkup JT, Townsend L, Crystal S, et al. A systematic review of validated methods for identifying suicide or suicidal ideation using administrative or claims data. Pharmacoepidemiol Drug Saf. 2012 Jan; 21 Suppl 1:174-82. doi: 10.1002/pds.2335. lxxxix Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997 Oct 15; 127(8 Pt 2):757-63. xc Sterling LH, Windle SB, Filion KB, et al. Varenicline and Adverse Cardiovascular Events: A Systematic Review and Meta‐Analysis of Randomized Controlled Trials. J Am Heart Assoc. 2016 Feb; 5(2): e002849. Published online 2016 Feb 22. doi:  [10.1161/JAHA.115.002849] xci Cahill K, Lindson-Hawley N, Thomas KH, et al. Nicotine receptor partial agonists for smoking cessation. Cochrane Database Syst Rev. 2016 May 9;(5):CD006103. doi: 10.1002/14651858.CD006103.pub7. xcii Cahill K, Stead LF, Lancaster T. Pharmacological interventions for smoking cessation: an overview and network meta-analysis. Cochrane Database Syst Rev. 2013 May 31;(5):CD009329. doi: 10.1002/14651858.CD009329.pub2.  81   xciii National Cancer Institute. Harms of Cigarette Smoking and Health Benefits of Quitting. https://www.cancer.gov/about-cancer/causes-prevention/risk/tobacco/cessation-fact-sheet. Published December 3, 2014. Accessed October 27, 2018. xciv Jha P, Ramasundarahettig C, Landsman V, et al. 21st-century hazards of smoking and benefits of cessation in the United States. N Engl J Med. 2013 Jan 24;368(4):341-50. doi: 10.1056/NEJMsa1211128. xcv Schneeweiss S, Rassen JA, Glynn RJ, et al. High-dimensional Propensity Score Adjustment in Studies of Treatment Effects Using Health Care Claims Data. Epidemiology. 2009 Jul;20(4):512-22. doi: 10.1097/EDE.0b013e3181a663cc. xcvi Cahill K, Stead LF, Lancaster T. Pharmacological interventions for smoking cessation: an overview and network meta-analysis. Cochrane Database Syst Rev. 2013 May 31;(5):CD009329. doi: 10.1002/14651858.CD009329.pub2. xcvii Cahill K, Lindson-Hawley N, Thomas KH, et al. Nicotine receptor partial agonists for smoking cessation. Cochrane Database of Systematic Reviews 2016, Issue 5. Art. No.: CD006103. DOI: 10.1002/14651858.CD006103.pub7 xcviii Taylor GMJ, Taylor AE, Thomas KH, et al. The effectiveness of varenicline versus nicotine replacement therapy on long-term smoking cessation in primary care: a prospective cohort study of electronic medical records. Int J Epidemiol. 2017 Dec; 46(6): 1948–1957. Published online 2017 Jun 26. doi:  [10.1093/ije/dyx109] xcix Health Canada. New safety information regarding Champix (varenicline tartrate) – For Health Professionals. June 20, 2008. http://www.healthycanadians.gc.ca/recall-alert-rappel-avis/hc-sc/2008/14499a-eng.php. Accessed November 30, 2018. c Motooka Y, Mastsui T, Slaton RM, et al. Adverse events of smoking cessation treatments (nicotine replacement therapy and non-nicotine prescription medication) and electronic cigarettes in the Food and Drug Administration Adverse Event Reporting System, 2004−2016. SAGE Open Med. 2018; 6: 2050312118777953. Published online 2018 May 21. doi:  10.1177/2050312118777953 ci U.S. Food and Drug Administration. Highlights of Prescribing Information. Chantix (varenicline). https://www.accessdata.fda.gov/drugsatfda_docs/label/2013/021928s030lbl.pdf. Accessed Nov 30, 2018. cii Anthenelli RM, Benowitz NL, West R, et al. Neuropsychiatric safety and efficacy of varenicline, bupropion, and nicotine patch in smokers with and without psychiatric disorders (EAGLES): a double-blind, randomised, placebo-controlled clinical trial. Lancet. 2016 Apr 19:e1-e14.[Epub ahead of print]. ciii Cinciripini PM, Robinson JD, Karam-Hage M, et al. Effects of varenicline and bupropion sustained-release use plus intensive smoking cessation counseling on prolonged abstinence from smoking and on depression, negative affect, and other symptoms of nicotine withdrawal. JAMA Psychiatry. 2013 May;70(5):522-33. doi: 10.1001/jamapsychiatry.2013.678. civ Gibbons RD, Mann JJ. Varenicline, smoking cessation, and neuropsychiatric adverse events. Am J Psychiatry. 2013 Dec;170(12):1460-7. doi: 10.1176/appi.ajp.2013.12121599. cv Health Canada Safety Alert. June 27, 2011.  http://healthycanadians.gc.ca/recall-alert-rappel-avis/hc-sc/2011/13619a-eng.php. Accessed November 30, 2018 cvi Pfizer. Efficacy and Safety of Varenicline in Smokers With Cardiovascular Disease Who Wish to Quit Smoking. April 29, 2009. https://clinicaltrials.gov/ct2/show/NCT00282984. Accessed November 30, 2018. cvii Singh S, Loke YK, Spangler JG, Furberg CD. Risk of serious adverse cardiovascular events associated with varenicline: a systematic review and meta‐analysis. CMAJ. 2011 Jul 4. cviii Sterling LH, Windle SB, Filion KB, et al. Varenicline and Adverse Cardiovascular Events: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. J Am Heart Assoc. 2016 Feb 22;5(2). pii: e002849. doi: 10.1161/JAHA.115.002849. cix BC Ministry of Health (2015). PharmaNet. BC Ministry of Health. Data Stewardship Committee (2015). Data Extract. http://www.popdata.bc.ca/data.  cx BC Ministry of Health (2015). PharmaCare. Population Data BC. Data Extract. MOH (2015). http://www.popdata.bc.ca/data.  82   cxi British Columbia Ministry of Health (2015): Medical Service Plan (MSP) Payment Information File. Population Data BC. Data Extract. MOH (2015). http://www.popdata.bc.ca/data. cxii Canadian Institute for Health Information (2015): Discharge Abstract Database (Hospital Separations). Population Data BC. Data Extract. MOH (2015). http://www.popdata.bc.ca/data. cxiii British Columbia Ministry of Heatlh (2015): Consolidation File (MSP Registration & Premium Billing). Population Data BC. Data Extract. MOH (2015). http://www.popdata.bc.ca/data. cxiv BC Vital Statistics Agency (2015): Vital Statistics Deaths. Population Data BC. Data Extract BC Vital Statistics Agency (2015). http://www.popdata.bc.ca/data. cxv Williams J. Inventory of studies on the accuracy of Canadian health administrative databases. Techincal report. Toronto: Institute for Clinical Evaluative Sciences (ICES), 1996. cxvi Rawson NS, Malcolm E. Validity of the recording of ischaemic heart disease and chronic obstructive pulmonary disease in the Saskatchewan health care datafiles. Stat Med. 1995 Dec 30;14(24):2627-43. cxvii McCormick N, Lacaille, Bhole V, Avina-Zubieta JA. Validity of Heart Failure Diagnoses in Administrative Datbases: A Systematic Review and Meta-Analysis. PLoS One. 2014; 9(8): e104519. Published online 2014 Aug 15. doi: 10.1371/journal.pone.0104519 cxviii Tirschwell DL, Longstreth WT Jr. Validating administrative data in stroke research. Stroke. 2002 Oct;33(10):2465-70. cxix McCormick N, Bhole V, Lacaille D, Avina-Zubieta JA. Validity of Diagnostic Codes for Acute Stroke in Administrative Databases: A Systematic Review. PLoS One. 2015 Aug 20;10(8):e0135834. doi: 10.1371/journal.pone.0135834. eCollection 2015. cxx Jun M, Lix LM, Durand M, et al. Comparative safety of direct oral anticoagulants and warfarin in venous thromboembolism: multicentre, population based, observational study. BMJ 2017 Oct 17; 359:j4323. Doi: 10.1136/bmj.j4323.  cxxi Gurevich Y, McFarlane A, Morris K, et al. Estimating the number of coronary artery bypass graft and percutaneous coronary intervention procedures in Canada: A comparison of cardiac registry and Canadian Institute for Health Information data sources. Can J Cardiol. 2010 Aug-Sep; 26(7): 249–253. cxxii Walkup JT, Townsend L, Crystal S, et al. A systematic review of validated methods for identifying suicide or suicidal ideation using administrative or claims data. Pharmacoepidemiol Drug Saf. 2012 Jan; 21 Suppl 1:174-82. doi: 10.1002/pds.2335. cxxiii Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997 Oct 15; 127(8 Pt 2):757-63. cxxiv Knowl MJ. [Down with odds ratios: risk ratios in cohort studies and randomised clinical trials]. [Article in Dutch]. Ned Tijdschr Geneeskd. 2012;156(28):A4775. cxxv McNutt L-A, Wu C, Xue X, et al. Estimating the Relative Risk in Cohort Studies and Clinical Trials of Common Outcomes. American Journal of Epidemiology, Volume 157, Issue 10, 15 May 2003, Pages 940–943, https://doi.org/10.1093/aje/kwg074 cxxvi Benowitz NL, Pipe A, West R, et al. Cardiovascular Safety of Varenicline, Bupropion, and Nicotine Patch in Smokers: A Randomized Clinical Trial. JAMA Intern Med. 2018 May 1;178(5):622-631. doi: 10.1001/jamainternmed.2018.0397. cxxvii Kotz D, Viechtbauer W, Simpson C, et al. Cardiovascular and neuropsychiatric risks of varenicline: a retrospective cohort study. Lancet Respir Med. 2015 Oct;3(10):761-8. doi: 10.1016/S2213-2600(15)00320-3. Epub 2015 Sep 6. cxxviii Widome R, Joseph AM, Hammett P, et al. Associations between smoking behaviors and financial stress among low-income smokers. Prev Med Rep. 2015; 2: 911–915. Published online 2015 Oct 29. doi:  [10.1016/j.pmedr.2015.10.011] cxxix Yoshida K, Solomon DH, Kim SC. Active-comparator design and new-user design in observational studies. Nat Rev Rheumatol. 2015 Jul; 11(7): 437–441. Published online 2015 Mar 24. doi:  [10.1038/nrrheum.2015.30]  83   cxxx World Health Organization. WHO Report on the Global Tobacco Epidemic: Monitoring Tobacco Use and Prevention Policies. 2017. WHO, 2017. cxxxi Cahill K, et al. Pharmacological interventions for smoking cessation: an overview and network meta-analysis. Cochrane Database Syst Rev. 2013 May 31;(5):CD009329. doi: 10.1002/14651858.CD009329.pub2. cxxxii Cahill K, Lindson-Hawley N, Thomas KH, Fanshawe TR, Lancaster T. Nicotine receptor partial agonists for smoking cessation. Cochrane Database Syst Rev. 2016 May 9;(5):CD006103. doi: 10.1002/14651858.CD006103.pub7. cxxxiii Edwards SA, Bondy SJ, Callaghan RC, Mann RE. Prevalence of unassisted quit attempts in population-based studies: a systematic review of the literature. Addict Behav. 2014;39(3):512–519. doi: 10.1016/j.addbeh.2013.10.036. cxxxiv Altman D, Clement F, Barnieh L, et al. Cost-Effectiveness of Universally Funding Smoking Cessation Pharmacotherapy. CHEST Journal. October 2016Volume 150, Issue 4, Supplement, Page 1303A. cxxxv Gonzales D1, Rennard SI, Nides M, Oncken C, Azoulay S, Billing CB, Watsky EJ, Gong J, Williams KE, Reeves KR; Varenicline Phase 3 Study Group. Varenicline, an alpha4beta2 nicotinic acetylcholine receptor partial agonist, vs sustained-release bupropion and placebo for smoking cessation: a randomized controlled trial. JAMA. 2006 Jul 5;296(1):47-55. cxxxvi Baker TB, Piper ME, Stein JH, et al. Effects of Nicotine Patch vs Varenicline vs Combination Nicotine Replacement Therapy on Smoking Cessation at 26 Weeks: A Randomized Clinical Trial. JAMA. 2016 Jan 26;315(4):371-9. doi: 10.1001/jama.2015.19284. cxxxvii Gray KM, McClure EA, Baker NL, et al. An exploratory short-term double-blind randomized trial of varenicline versus nicotine patch for smoking cessation in women. Addiction. 2015 Jun;110(6):1027-34. doi: 10.1111/add.12895. Epub 2015 Mar 29 cxxxviii Koegelenberg CF, Noor F, bateman ED, et al. Efficacy of varenicline combined with nicotine replacement therapy vs varenicline alone for smoking cessation: a randomized clinical trial. JAMA. 2014 Jul;312(2):155-61. doi: 10.1001/jama.2014.7195. cxxxix Ebbert JO, Hatsukami DK, Croghan IT, et al. Combination varenicline and bupropion SR for tobacco-dependence treatment in cigarette smokers: a randomized trial. JAMA. 2014 Jan 8;311(2):155-63. doi: 10.1001/jama.2013.283185. cxl Aubin HJ, Bobak A, Britton JR, et al. Varenicline versus transdermal nicotine patch for smoking cessation: results from a randomised, open-label trial. Thorax 2008;63:717–24. cxli Jorenby DE, Hays JT, Rigotti NA, et al.; Varenicline Phase 3 Study Group. Efficacy of varenicline, an alpha4beta2 nicotinic acetylcholine receptor partial agonist, vs placebo or sustained-release bupropion for smoking cessation: a randomized controlled trial. JAMA 2006;296:56–63 cxlii Gonzales D, Rennard SI, Nides M, et al. Varenicline, an alpha4beta2 nicotinic acetylcholine receptor partial agonist, vs sustained-release bupropion and placebo for smoking cessation: a randomized controlled trial. JAMA. 2006 Jul 5;296(1):47-55. cxliii Rigotti NA, Pipe AL, Benowitz NL, Arteaga C, Garza D, Tonstad S. Efficacy and safety of varenicline for smoking cessation in patients with cardiovascular disease: a randomized trial. Circulation. 2010 Jan 19;121(2):221‐9. cxliv Twardella D, Brenner H. Implications of Nonresponse Patterns in the Analysis of Smoking Cessation Trials. Nicotine Tob Res2008;10:891-6.  cxlv BC Ministry of Health (2015). PharmaNet. BC Ministry of Health. Data Stewardship Committee (2015). Data Extract. http://www.popdata.bc.ca/data.  cxlvi BC Ministry of Health (2015). PharmaCare. Population Data BC. Data Extract. MOH (2015) http://www.popdata.bc.ca/data. cxlvii British Columbia Ministry of Health (2015): Medical Service Plan (MSP) Payment Information File. Population Data BC. Data Extract. MOH (2015). http://www.popdata.bc.ca/data.  84   cxlviii Canadian Institute for Health Information (2015): Discharge Abstract Database (Hospital Separations). Population Data BC. Data Extract. MOH (2015). http://www.popdata.bc.ca/data. cxlix British Columbia Ministry of Heatlh (2015): Consolidation File (MSP Registration & Premium Billing). Population Data BC. Data Extract. MOH (2015). http://www.popdata.bc.ca/data.  cli Schneeweiss S, Rassen J, Glynn R, Avorn J, Mogun H, Brookhart A. High-dimensional Propensity Score Adjustment in Studies of Treatment Effects Using Health Care Claims Data. Epidemiology. 2009 Jul;20(4):512-22. doi: 10.1097/EDE.0b013e3181a663cc. clii GP Services Committee. Prevention Incentives. January 2018. http://www.gpscbc.ca/sites/default/files/uploads/GPSC%20Billing%20Guide%20-%20Prevention%20201801.pdf. Accessed February 25th, 2018.  cliii Lancaster T, Stead LF. Individual behavioural counselling for smoking cessation. Cochrane Database Syst Rev. 2017 Mar 31; 3:CD001292. doi: 10.1002/14651858.CD001292.pub3. cliv Ranganathan, P, Aggarwal R, Pramesh C. Common pitfalls in statistical analysis: Odds versus risk. Perspect Clin Res. 2015 Oct-Dec; 6(4): 222-224. clv Knol M, Le Cessie S, Algra A, et al. Overestimation of risk ratios by odds ratios in trials and cohort studies: alternatives to logistic regression. CMAJ. 2012 May 15: 184(8): 895-899. clvi McNutt L-A, Wu C, Xue X, et al. Estimating the Relative Risk in Cohort Studies and Clinical Trials of Common Outcomes.  clvii Cinciripini PM, Minnix JA, Green CE, et al. An RCT with the combination of varenicline and bupropion for smoking cessation: clinical implications for front line use. Addiction. 2018 Apr 21. doi: 10.1111/add.14250. [Epub ahead of print] clviii Taylor GMJ, Taylor AE, Thomas KH, et al. The effectiveness of varenicline versus nicotine replacement therapy on long-term smoking cessation in primary care: a prospective cohort study of electronic medical records. Int J Epidemiol. 2017 Dec; 46(6): 1948–1957. Published online 2017 Jun 26. doi:  [10.1093/ije/dyx109] clix West R, Evins AE, Benowitz NL, et al. Factors associated with the efficacy of smoking cessation treatments and predictors of smoking abstinence in EAGLES. Addiction. 2018 Aug;113(8):1507-1516. doi: 10.1111/add.14208. Epub 2018 Mar 30. clx Mckee S. A, Smith P. H, Kaufman M, et al.. Sex differences in varenicline efficacy for smoking cessation: a meta‐analysis. Nicotine Tob Res 2016; 18: 1002–1011. clxi Gonzales D, Hajek P, Pliamm L, et al. Retreatment with varenicline for smoking cessation in smokers who have previously taken varenicline: a randomized, placebo‐controlled trial. Clin Pharmacol Ther 2014; 96: 390–396. clxii Benowitz N, Pipe A, West R. Cardiovascular Safety of Varenicline, Bupropion, and Nicotine Patch in Smokers. A Randomized Clinical Trial. JAMA Intern Med. 2018;178(5):622-631. doi:10.1001/jamainternmed.2018.0397 clxiii Kotz D, Viechtbauer W, Simpson C, et al. Cardiovascular and neuropsychiatric risks of varenicline: a retrospective cohort study. Lancet Respir Med. 2015 Oct;3(10):761-8. doi: 10.1016/S2213-2600(15)00320-3. Epub 2015 Sep 6. clxiv Thomas KH, Martin RM, Davies NM, et al. Smoking cessation treatment and risk of depression, suicide, and self harm in the Clinical Practice Research Datalink: prospective cohort study. BMJ 2013 Oct 11;347:f5704. doi: 10.1136/bmj.f5704 clxv Anthenelli R, Benowitz N, West R, et al. Neuropsychiatric safety and efficacy of varenicline, bupropion, and nicotine patch in smokers with and without psychiatric disorders (EAGLES): a double-blind, randomised, placebo-controlled clinical trial. Lancet. 2016 Apr 19:e1-e14.  85  Appendix A - Exclusion Criteria; Diagnosis or Treatment for Depression Patients were excluded from the study cohorts if they had either a physician or hospital diagnosis for depression, or a prescription for a medication used to treat depression, defined as follows:  Diagnosis: Condition ICD-9-CM Codes Location Depression 296.2, 296.3, 298.0, 300.4, 309.0 309.1 Inpatient or outpatient diagnosis in any position  Medications: Drug Group Drug Name Tricyclic Antidepressants (TCAs) Amitriptyline Butriptyline Clompipramine Dosulepin Doxepin Imipramine Iprindole Lofepramine Nortriptyline Protriptyline Trimipramine Serotonin and Norepinephrine Reuptake Inhibitors (SNRIs) Venlafaxine Desvenlafaxine Duloxetine Milnacipran Levomilnacipran Sibutramine Selective Serotonin Reuptake Inhibitors (SSRIs) Citalopram Fluvoxamine Escitalopram Paroxetine Sertraline Fluoxetine    86  Appendix B - Case Definitions for Baseline Characteristics Outpatient or inpatient diagnosis in any position, unless otherwise specified Condition ICD-9-CM Codes CPT-4 Location Excessive Alcohol Abuse 291.x, 303.x, 571.0x-571.3x, 535.3x   COPD 491.x, 492.x, 496.x   Acute Myocardial Infarction 410.x   Congestive Heart Failure 428.x   Cerebrovascular Diseases 430.x - 438.x   Diabetes 250.x   Cerebral Infarction 430.x-434.x   Chronic Kidney Disease 580.x – 583.x, 585.x – 587.x, 589.x   Angina 413.x   Hypertension 401.x   Peripheral Vascular Disease 440.x, 441.x, 443.x 444.x, 445.x, 437.x, 557.x   Coronary Revascularization  92920 – 92934, 92937 - 92944 Inpatient only        87  Appendix C - Propensity Score Summary Statistics and Histograms C.1 NRT versus Bupropion (outcome=composite cardiovascular hospitalization) Min score NRT  0.001 Max score NRT  0.994     Min score Bupropion 0.004 Max score Bupropion 0.999     Model C statistic  0.766  Histogram of propensity scores:  C.2 NRT versus Varenicline (outcome=composite cardiovascular hospitalization) Min score NRT  0.002 Max score NRT  0.995     Min score Varenicline 0.01 Max score Varenicline 0.998     Model C statistic  0.791  Histogram of propensity scores: 88   C.3 NRT versus Bupropion (outcome=composite neuropsychiatric hospitalization) Min score NRT  0.002 Max score NRT  0.999     Min score Bupropion 0.001 Max score Bupropion 0.997     Model C statistic  0.764  Histogram of propensity scores:  89  C.4 NRT versus Varenicline (outcome=composite neuropsychiatric hospitalization) Min score NRT  0.01 Max score NRT  0.997     Min score Varenicline 0.004 Max score Varenicline 0.996     Model C statistic  0.789  Histogram of propensity scores:  C.6 NRT versus Bupropion; 5% Trimmed Propensity Score (outcome=composite neuropsychiatric hospitalization) Min score NRT  0.453 Max score NRT  0.453     Min score Bupropion 0.967 Max score Bupropion 0.967     Model C statistic  0.762  Histogram of propensity scores: 90   NRT versus Varenicline; 5% Trimmed Propensity Score (outcome=composite neuropsychiatric hospitalization) Min score NRT  0.8 Max score NRT  0.989     Min score Varenicline 0.8 Max score Varenicline 0.989     Model C statistic  0.788  Histogram of propensity scores:  91  Appendix D - Case Definitions for Baseline Characteristics Outpatient or inpatient diagnosis in any position, unless otherwise specified Condition ICD-9-CM Codes CPT-4 Location Excessive Alcohol Abuse 291.x, 303.x, 571.0x-571.3x, 535.3x   COPD 491.x, 492.x, 496.x   Acute Myocardial Infarction 410.x   Congestive Heart Failure 428.x   Cerebrovascular Diseases 430.x - 438.x   Diabetes 250.x   Cerebral Infarction 430.x-434.x   Chronic Kidney Disease 580.x – 583.x, 585.x – 587.x, 589.x   Angina 413.x   Hypertension 401.x   Peripheral Vascular Disease 440.x, 441.x, 443.x 444.x, 445.x, 437.x, 557.x   Coronary Revascularization  92920 – 92934, 92937 - 92944 Inpatient only     92  Appendix E - Case Definitions for Baseline Characteristics Outpatient or inpatient diagnosis in any position, unless otherwise specified Condition ICD-9 Codes ICD-10 Codes CCI Codes Location Excessive Alcohol Abuse 291.x, 303.x, 571.0x-571.3x, 535.3x F10.x  Hospital admission or physician billings Depression /Anxiety 296.x, 300.x, 311.x, 50b [50b is unique to BC physician billing codes] F32, F33, F40, F41  Hospital admission or physician billings COPD 491.x, 492.x, 496.x J44.x, J41.x, J43.x  Hospital admission or physician billings Acute Myocardial Infarction 410.x I21.x  Hospital admission or physician billings Congestive Heart Failure 428.x I50.x  Hospital admission or physician billings Cerebrovascular Diseases 430.x - 438.x I60.x – I69.x  Hospital admission or physician billings Diabetes 250.x E10.x-E14.x  Hospital admission or physician billings Chronic Kidney Disease 580.x – 583.x, 585.x – 587.x, 589.x N00.x - N07.x  Hospital admission or physician billings Hypertension 401.x I10.x - I13.x  Hospital admission or physician billings Peripheral Vascular Disease 440.x, 441.x, 443.x 444.x, 445.x, 437.x, 557.x I70, I71, I74, I75   Hospital admission or physician billings Angina 413.x I20.x  Hospital admission or physician billings Coronary Revascularization   1IJ50.x, 1IJ57GQ.x, 1IJ76.x , 1IJ80.x, 1IJ57LA.x Inpatient only       93  Appendix F - History of Cardiovascular Disease Condition ICD-9 Codes ICD-10 Codes CCI Codes Location Acute Myocardial Infarction 410.x I21.x  Hospital admission or physician billings Congestive Heart Failure 428.x I50.x  Hospital admission or physician billings Cerebrovascular Diseases 430.x - 438.x I60.x – I69.x  Hospital admission or physician billings Peripheral Vascular Disease 440.x, 441.x, 443.x 444.x, 445.x, 437.x, 557.x I70, I71, I74, I75   Hospital admission or physician billings Angina 413.x I20.x  Hospital admission or physician billings Coronary Revascularization   1IJ50.x, 1IJ57GQ.x, 1IJ76.x , 1IJ80.x, 1IJ57LA.x Inpatient only        

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0380464/manifest

Comment

Related Items