Factors Influencing the Use of Pharmacotherapy for Smoking Cessation During and After Pregnancy

Factors influencing the use of pharmacotherapy for smoking cessation during and after pregnancy

Mei Lin Lee BPharm (Hons), MSc (Pharm)

A thesis in fulfillment of the requirements for the degree of Doctor of Philosophy

Centre for Big Data Research in Health School of Public Health and Community Medicine Faculty of Medicine University of New South Wales March 2020

Thesis / Dissertation sheet

Surname/Family Name : Lee Given Name/s : Mei Lin Abbreviation for degree as give in the University calendar : PhD Faculty : Medicine School : Public Health and Community medicine Factors influencing the use of pharmacotherapy Thesis Title : for smoking cessation during and after pregnancy

Abstract 350 words maximum Maternal smoking cessation is key to improving maternal, neonatal and infant health. As pharmacotherapies for smoking cessation, which include nicotine replacement therapy (NRT), bupropion and varenicline, are the most effective cessation interventions in the general population, their use holds promise to promote quitting during and after pregnancy. This thesis aimed to identify factors, from the perspective of healthcare providers from healthcare providers’ perspectives, that influence the prescribing and/or use of smoking cessation pharmacotherapies among women who smoke during pregnancy and after giving birth. A study based on calls to a teratology information service found that healthcare providers have more safety concerns regarding bupropion and varenicline than other medications of same pregnancy risk category. Perceived safety concerns might limit their prescribing. A survey of Australian obstetricians and gynaecologists found that facility-level factors, including having a smoking cessation protocol and practising in remote and regional areas, were associated with familiarity with NRT patches. Among obstetricians familiar with NRT patches, those who believed that they could benefit their patients, and those who had self-efficacy to prescribe were more likely to intend to prescribe NRT patches. Although lack of knowledge of NRT patches might limit prescribing, individual beliefs might also limit the prescribing among those who were familiar with NRT. A study based on linked data comprising perinatal, hospital admission, death and dispensing records for all women who gave birth in NSW and WA, found 8.7% (95% CI 8.17-9.10) of women smoking at the time of delivery used a smoking cessation pharmacotherapy in the next 12 months. At 6.3%% (95% CI 5.89-6.69), varenicline was the most used pharmacotherapy. Maternal suitability for pharmacotherapy was not an influential factor as few maternal morbidities worsened by smoking were associated with use of these pharmacotherapies. The impact of women experiencing poor birth outcomes related to smoking was not a factor to motivate more pharmacologically assisted quit attempts. Another population-based study using linked administrative data provided evidence of varenicline’s effectiveness among postpartum maternal smokers. This evidence can be used to guide future decisions about whether to use varenicline postpartum. Robust data are required regarding safety of these pharmacotherapies during and after pregnancy to help support healthcare providers in seizing opportunities to assist pregnant and postpartum women in quitting smoking using these pharmacotherapies.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents a non-exclusive licence to archive and to make available (including to members of the public) my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known. I acknowledge that I retain all intellectual property rights which subsist in my thesis or dissertation, such as copyright and patent rights, subject to applicable law. I also retain the right to use all or part of my thesis or dissertation in future works (such as articles or books).

…………………………………………………………… 11 Mar 2020 Signature Date The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years can be made when submitting the final copies of your thesis to the UNSW Library. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.

Originality statement

I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.

Mei Lin Lee Date: 7 March 2020

Inclusions of publications statement UNSW is supportive of candidates publishing their research results during their candidature as detailed in the UNSW Thesis Examination Procedure.

Publications can be used in their thesis in lieu of a Chapter if: • The candidate contributed greater than 50% of the content in the publication and is the “primary author”, ie. the candidate was responsible primarily for the planning, execution and preparation of the work for publication • The candidate has approval to include the publication in their thesis in lieu of a Chapter from their supervisor and Postgraduate Coordinator. • The publication is not subject to any obligations or contractual agreements with a third party that would constrain its inclusion in the thesis

Please indicate whether this thesis contains published material or not:

This thesis contains no publications, either published or submitted for publication ☐ (if this box is checked, you may delete all the material on page 2)

Some of the work described in this thesis has been published and it has been documented in the relevant Chapters with acknowledgement ☐ (if this box is checked, you may delete all the material on page 2)

This thesis has publications (either published or submitted for publication) ☒ incorporated into it in lieu of a chapter and the details are presented below

CANDIDATE’S DECLARATION I declare that: • I have complied with the UNSW Thesis Examination Procedure • where I have used a publication in lieu of a Chapter, the listed publication(s) below meet(s) the requirements to be included in the thesis. Candidate’s Name Signature Date (dd/mm/yy)

MEI LIN LEE 7/03/20

iii

POSTGRADUATE COORDINATOR’S DECLARATION To only be filled in where publications are used in lieu of Chapters I declare that: • the information below is accurate • where listed publication(s) have been used in lieu of Chapter(s), their use complies with the UNSW Thesis Examination Procedure • the minimum requirements for the format of the thesis have been met. PGC’s Name PGC’s Signature Date (dd/mm/yy) Assoc Prof James Wood 8/03/20

For each publication incorporated into the thesis in lieu of a Chapter, provide all of the requested details and signatures required Details of publication #1: Full title: Health-care providers’ concern regarding smoking cessation pharmacotherapies during pregnancy: Calls to a teratology information service Authors: Mei Lin Lee, Duong T. Tran, Alec Welsh, Debra Kennedy, Alys Havard Journal or book name: Drug & Alcohol Review Volume/page number: https://onlinelibrary.wiley.com/doi/full/10.1111/dar.13033 Date accepted/published: Accepted on 30 December 2019, and published in Early View section of the journal Status Published x Accepted and In In progress press (submitted) The Candidate’s Contribution to the Work MLL, AH and AW conceptualised the study. MLL cleaned and analysed the data, with the assistance of DTT. MLL interpreted the results and drafted the initial manuscript with contributions from AH and DTT. MLL, AH, DTT, AW and DK revised the manuscript. All co- authors approved the final manuscript. All co-authors were responsible for making all changes requested by the journal.

Location of the work in the thesis and/or how the work is incorporated in the thesis: Chapter 2 aims to investigate whether healthcare providers are overly concerned about the safety of pharmacotherapies for smoking cessation during pregnancycompared to medications in the same and other pregnancy-risk categories. These safety concerns may result in reluctance among healthcare providers to prescribe these pharmacotherapies to pregnant women. This research addresses the aim of the thesis regarding factors influencing the prescribing of smoking cessation pharmacotherapies to pregnant women.

PRIMARY SUPERVISOR’S DECLARATION I declare that: • the information above is accurate • this has been discussed with the PGC and it is agreed that this publication can be included in this thesis in lieu of a Chapter • All of the co-authors of the publication have reviewed the above information and have agreed to its veracity by signing a ‘Co-Author Authorisation’ form. Primary Supervisor’s name Primary Supervisor’s signature Date (dd/mm/yy)

Dr Alys Havard 12/3/20

Supervisor statement

I hereby declare that all co-authors of publication #1 referenced in the ‘Inclusion of Publications Statement’ agree to Mei Lin Lee submitting this publication as a chapter in her doctoral thesis. I also acknowledge that Mei Lin Lee made a substantial (>50%) intellectual contribution to the paper.

Dr Alys Havard Primary Supervisor Date: 9 March 2020

I hereby grant the University of New South Wales or its agents a non-exclusive licence to archive and to make available (including to members of the public) my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known. I acknowledge that I retain all intellectual property rights which subsist in my thesis or dissertation, such as copyright and patent rights, subject to applicable law. I also retain the right to use all or part of my thesis or dissertation in future works (such as articles or books).

For any substantial portions of copyright material used in this thesis, written permission for use has been obtained, or the copyright material is removed from the final public version of the thesis.

Mei Lin Lee Date: 7 March 2020

Authenticity statement

I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.

Mei Lin Lee Date: 7 March 2020

vii

This thesis is dedicated to all smokers who cannot quit smoking, with a special dedication to the loving memory of my late father, Lee Hock Lye (1942-2004) who passed away after a short battle with lung cancer.

“Papa, for you, always”.

viii

Acknowledgements ‘No man ever steps in the same river twice, for it is not the same river and he is not the same man’ - Heraclitus (530-470BC).

I am extremely grateful to all who have supported me throughout my PhD program. The past four years have been a steep learning growth for me in becoming a researcher, a wiser healthcare provider, and a better person. To raise this thesis, it takes several villages and I am equally grateful to each village.

First, I would like to thank the University of New South Wales for awarding me a PhD scholarship to pursue this program and the Centre for Big Data Research in Health for providing a good research environment. Without their support, I would not have had the opportunity to cross continents to carry out and to write this research.

To my supervisors, Dr. Alys Havard, Dr. Duong Thuy Tran, and Professor Alec Welsh, I cannot express words to describe their mentorship, generosity and infinite patience. Alys, I am very grateful for your motivation, your generous advice and the opportunities that you provided in nurturing a novice who did not know what an odds ratio was into someone who can now sniff out suspiciously-tweaked analyses. I am deeply moved by your concern for my well-being, especially in tumultuous times, that spanned across continents. Danielle, your down-to-earth approach in guiding me to data cleaning (MotherSafe study), and to conduct statistical analyses has helped build my confidence that a healthcare provider can be a good epidemiologist as well, with motherhood thrown to the mix. I am very thankful that I could seek you out 24-7. Alec, thank you for the golden nuggets of wisdom that set my thesis path where it is now. Healthcare providers, not women. Effectiveness, not last-minute randomly assembled chapter. It is due to Alys, Danielle, and Alec that I have completed this thesis.

I wish to thank all the obstetricians and gynaecologists In Australia who participated in the study. I am indebted to Dr. Debra Kennedy, Majela Hill and Delwyn Culpitt at MotherSafe, for their support when I was carrying out this study. I wish to acknowledge the unnamed people who key in data into MotherSafe electronic database on daily basis, and to MotherSafe callers whose data were included in this thesis. I also extend this sentiment to participants in the Smoking MUMS study; the unnamed midwives, doctors, pharmacists, maternity support workers, and coders who enter data into NSW,

WA and PBS electronic administration systems, and to the women and babies whose data I have the privilege to analyse. Without all these people, this thesis is not possible.

I am grateful to the family at Center of Big Data Research in Health. In particular, I wish to thank Drs Bich Tran, Michael Falster, Sharon Chow, Monica Tang and Hon Hwang. Dr Marina van Leeuwen, I am so blessed with your generosity and your company during late nights. Professor Claire Vladijic, I am deeply humbled by your constant looking out for me and I am indebted to you for all the food and sustenance. Sanja Lujic, thank you for helping me out with my statistical struggles and for playing the role of the fairy godmother very well. Andrew Blance (and Dr Katie Harris), I really appreciate the advice that was doled out during times when I most needed them.

To the Ted Noffs Foundation for which I am their part-time research officer, I am very thankful that I was given the opportunity to experience the data custodian’s side of the fence. Thank you to Mark Ferry and Michael Chan for understanding that my PhD has to come first, especially when I was unwell. A great big hug to the Noffs gang; Bex, Steph, Shelley, Riri, Al and Kev, and a belly rub to Coopa the dog.

To the patients at the Prince of Wales hospital where I volunteer, thank you very much for always reminding me that there is life beyond PhD and it is a beautiful one. I have deposited your well-wishes tokens for my research into the making of this thesis.

As for friends like Jo, Eranna, Griselda, Edna and Ita, thank you for putting up with me all these years. To Emily Karanges and the late John Lewis, I appreciate your constant encouragement and support during my early PhD years. To my steadfast TaiChi buddies at the park, Theo, Angelo, and Tom, thank you for filling my spirit, and my stomach with morning exercises and BBQs.

Last but not least, to my real village in Malaysia, that of my dearest mother, daughter, and sisters. Mamee, I humbly bow to you in immense gratitude for looking after Elvy in my absence. Poops, although I deeply regret missing out on these four years of your life, I am so, so glad that Grandma is now your bestie. For my wonderful sisters, Cia, Pepi, and Mitch, you are simply magnificent in keeping me sane all this time. To the tune of our family song, Ed Sheeran’s Photograph, ‘wait for me to come home’, I know that all of you will be just as excited for this thesis to be submitted as I am.

Table of contents Thesis / Dissertation sheet ...... i Originality statement ...... ii Inclusions of publications statement...... iii Supervisor statement ...... v Copyright statement ...... vi Authenticity statement ...... vii Acknowledgements ...... ix Table of contents ...... xi Thesis abstract ...... xvi Publications ...... xvii Presentations ...... xvii List of Chapter tables ...... xviii List of Chapter figures ...... xix List of Abbreviations ...... xx Chapter 1 Introduction and review of literature ...... 1 Smoking and smoking cessation during pregnancy ...... 2 1.1.1 Health impact of smoking during pregnancy ...... 2 1.1.2 Prevalence of smoking during pregnancy ...... 3 1.1.3 Smoking cessation during pregnancy ...... 5 1.1.4 Summary of evidence on smoking and smoking cessation during pregnancy ...... 23 Smoking and smoking cessation after pregnancy ...... 25 1.2.1 Health risks of smoking after pregnancy ...... 25 1.2.2 Prevalence of smoking after pregnancy ...... 26 1.2.3 Smoking cessation after pregnancy ...... 27 1.2.4 Summary of evidence on smoking and smoking cessation after pregnancy 34 Overview of smoking cessation pharmacotherapy ...... 36 1.3.1 Clinical pharmacology ...... 36 1.3.2 Efficacy in the general population ...... 40 1.3.3 Safety and tolerability ...... 40 1.3.4 Cost...... 42 Summary of existing literature and evidence gaps ...... 43 Thesis aims and overview ...... 46

Chapter 2 Healthcare providers’ concern regarding smoking cessation pharmacotherapies during pregnancy: Calls to a teratology information service ...... 48 Preface to the chapter ...... 48 Published manuscript in Microsoft Word format ...... 49 Chapter 3 Factors associated with obstetricians’ and gynaecologists’ prescribing of smoking cessation pharmacotherapies to pregnant women ...... 63 Introduction ...... 63 Aims ...... 66 Theoretical framework ...... 66 Methods ...... 69 3.4.1 Study population ...... 69 3.4.2 Questionnaire sections and items ...... 70 3.4.3 Questionnaire pilot and administration ...... 71 3.4.4 Study variables ...... 72 3.4.5 Consent and ethics ...... 80 3.4.6 Data storage, handling, and confidentiality ...... 80 3.4.7 Statistical analyses ...... 81 Results ...... 83 3.5.1 Characteristics of respondents ...... 83 3.5.2 Facility-level factors associated with familiarity with NRT patches ...... 88 3.5.3 Factors associated with intention to prescribe NRT patches among obstetricians who were familiar with NRT patches ...... 90 Discussion ...... 93 3.6.1 Principal findings and interpretation ...... 93 3.6.2 Incidental findings and interpretation ...... 96 3.6.3 Limitations ...... 98 3.6.4 Conclusion ...... 100 Chapter 4 Use of smoking cessation pharmacotherapies in the 12 months after giving birth…...... 102 Introduction ...... 102 Aim ...... 104 Methods ...... 104 4.3.1 Data sources and linkages ...... 104 4.3.2 Study population ...... 107

xii

4.3.3 Ascertainment of smoking cessation pharmacotherapy use and timing of use ...... 110 4.3.4 Factors associated with pharmacotherapy use: Maternal morbidities and poor birth outcomes ...... 110 4.3.5 Potential confounders ...... 117 4.3.6 Statistical analyses ...... 118 4.3.7 Ethics and data access ...... 120 Results ...... 121 4.4.1 Prevalence and timing of smoking cessation pharmacotherapy use ..... 124 4.4.2 Relationship between maternal morbidities, and poor birth outcomes and smoking cessation pharmacotherapy use in the 12 months postpartum 128 Discussion ...... 133 4.5.1 Summary of findings ...... 133 4.5.2 Comparison with other studies and interpretation ...... 133 4.5.3 Limitations ...... 136 4.5.4 Conclusions ...... 139 Chapter 5 The effectiveness of smoking cessation pharmacotherapy during the inter-pregnancy interval ...... 140 Introduction ...... 140 Aim ...... 142 Methods ...... 142 5.3.1 Data sources and linkages ...... 142 5.3.2 Study population ...... 142 5.3.3 Ascertainment of smoking cessation pharmacotherapy exposures ...... 144 5.3.4 Smoking cessation outcomes ...... 145 5.3.5 Potential confounders ...... 148 5.3.6 Statistical analyses ...... 149 5.3.7 Ethics and data access ...... 150 Results ...... 151 5.4.1 Comparison between women who were included and excluded based on missing information on their number of cigarettes smoked during the second half of their first pregnancy ...... 151 5.4.2 Characteristics of the study population ...... 153 5.4.3 Association between varenicline exposure during the inter-pregnancy interval and smoking cessation prior to the second pregnancy ...... 156

xiii

Discussion ...... 160 5.5.1 Summary of findings ...... 160 5.5.2 Comparison with other studies and interpretation ...... 160 5.5.3 Limitations ...... 161 5.5.4 Implications for practice ...... 162 5.5.5 Conclusions ...... 163 Chapter 6 Discussion ...... 165 Review of the thesis aims ...... 165 Summary and interpretation of the main findings of the thesis ...... 167 6.2.1 Factors influencing the use of pharmacotherapy for smoking cessation during pregnancy...... 167 6.2.2 Factors influencing the use of pharmacotherapy for smoking cessation after pregnancy ...... 170 Strengths and limitations of the thesis ...... 173 Implications for clinical practice and for future research ...... 177 Conclusion ...... 182 References ...... 184 Appendix ...... 206 Chapter Two Appendix ...... 206 Appendix 2 ...... 206 Chapter Three Appendices ...... 214 Appendix 3A Email invitation to participate in the survey ...... 214 Appendix 3B WHA website recruitment of survey participants ...... 221 Appendix 3C Full questionnaire ...... 222 Appendix 3D Ethics approval ...... 231 Chapter Four Appendices ...... 232 Appendix 4A Ascertainment of pre-existing maternal morbidities from hospital and PBS dispensing data ...... 232 Appendix 4B Components and ICD10 codes for Maternal Morbidity Outcome Indicator (MMOI), adapted from Roberts et al. (2008) ...... 235 Appendix 4C Components and ICD10 codes for Neonatal Adverse Outcome Indicator (NAOI), adapted from Lain et al. (2012)[353] ...... 236 Appendix 4D Correlation matrix for each potential confounder...... 240

xiv

Appendix 4F Association between neonatal adverse events in relation to varenicline use in the 12 months postpartum, 2011-2012, adjusted for maternal morbidities and potential confounders ...... 243 Appendix 4G Association between maternal adverse events in relation to NRT patches use in the 12 months postpartum, 2011-2012, adjusted for maternal morbidities and potential confounders ...... 245 Appendix 4H Association between neonatal adverse events in relation to NRT patches use in the 12 months postpartum, 2011-2012, adjusted for maternal morbidities and potential confounders ...... 247 Chapter Five Appendix ...... 249 Appendix 5 Comparison between women who were excluded and included in the study cohort based on missing information on the quantity of cigarettes smoked during the second half of their first pregnancy ...... 249

Thesis abstract Maternal smoking cessation is key to improving maternal, neonatal and infant health. As pharmacotherapies for smoking cessation, which include nicotine replacement therapy (NRT), bupropion and varenicline, are the most effective cessation interventions in the general population, their use holds promise to promote quitting during and after pregnancy. This thesis aimed to identify factors that influence the prescribing and/or use of smoking cessation pharmacotherapies among women who smoke during pregnancy and after giving birth. A study based on calls to a teratology information service found that healthcare providers have more safety concerns regarding bupropion and varenicline than other medications of same pregnancy risk category. Safety concerns might limit their prescribing. A survey of Australian obstetricians and gynaecologists found that facility-level factors, including having a smoking cessation protocol and practising in remote and regional areas, were associated with familiarity with NRT patches. Among obstetricians familiar with NRT patches, those who believed that they could benefit their patients, and those who had self-efficacy to prescribe were more likely to intend to prescribe NRT patches. Although lack of knowledge of NRT patches might limit prescribing, individual beliefs might also limit the prescribing among those who were familiar with NRT. A study based on linked data comprising perinatal, hospital admission, death and dispensing records for all women who gave birth in NSW and WA, found 8.7% (95% CI 8.2-9.1) of women smoking at the time of delivery used a smoking cessation pharmacotherapy in the next 12 months. At 6.3% (95% CI 5.9-6.7), varenicline was the most used pharmacotherapy. Maternal suitability for pharmacotherapy was not an influential factor as few maternal morbidities worsened by smoking were associated with use of these pharmacotherapies. Poor birth outcomes were not capitalised on as a factor to motivate more pharmacologically assisted quit attempts. Another population-based study using linked administrative data provided evidence of varenicline’s effectiveness among postpartum maternal smokers. This evidence can be used to guide future decisions about whether to use varenicline postpartum. Robust data arerequired regarding safety of these pharmacotherapies during and after pregnancy to help support healthcare providers in seizing opportunities to assist pregnant and postpartum women in quitting smoking using these pharmacotherapies.

xvi

Publications Lee ML, Tran DT, Welsh A, Kennedy D, Havard A. Health-care providers' concern regarding smoking cessation pharmacotherapies during pregnancy: Calls to a teratology information service. Drug and Alcohol Review 2020. DOI: https://onlinelibrary.wiley.com/doi/full/10.1111/dar.13033

Presentations Mei Lin Lee*. Smoking cessation pharmacotherapies during pregnancy. CMC Midwifery Practice Development, Royal Hospital for Women, Sydney, 10th May 2017.

Mei Lin Lee*, Alys Havard,Duong T. Tran, Debra Kennedy, Alec Welsh. Healthcare providers' concerns regarding smoking cessation pharmacotherapies in pregnancy: Calls to a teratology information service. Full Oral presentation. 17th World Conference on Tobacco or Health, WCTOH 2018, Cape Town, South Africa, 07-09 March, 2018.

Mei Lin Lee*, Duong T. Tran, Alec Welsh, Alys Havard. Maternal morbidities associated with post-delivery use of smoking cessation pharmacotherapies. One minute rapid fire oral and poster presentation. Australian Epidemiology Association ASM 2018, Perth, 22nd - 24th October 2018

Mei Lin Lee*, Duong T. Tran, Alec Welsh, Alys Havard. Maternal morbidities associated with post-delivery use of smoking cessation pharmacotherapies. Three Minute oral presentation. 2018 UNSW Postgraduate Research Symposium: Exploring Opportunities, Sydney, 16th November 2018.

*presenter

xvii

List of Chapter tables Table 1 Characteristics and efficacy of psychosocial interventions for smoking cessation that have been evaluated in pregnancy* ...... 10 Table 2 Health conditions worsened by smoking* ...... 26 Table 3 Pharmacotherapeutic properties of smoking cessation pharmacotherapies ... 38 Table 4: Characteristics of calls to MotherSafe for 2001-2016 and 2008-2016 ...... 57 Table 5 Organisation of survey items and response options ...... 75 Table 6 Characteristics of survey respondents and RANZCOG-affiliated obstetricians84 Table 7 Characteristics of survey respondents ...... 86 Table 8 Smoking cessation practices and interest of survey respondents ...... 87 Table 9 Association between facility-level factors and familiarity with NRT patches among obstetricians ...... 89 Table 10 Association between facility-level factors and intention to prescribe NRT patches among obstetricians who were familiar with NRT patches ...... 91 Table 11 Relationship between individual-level factors and intention to prescribe NRT patches among obstetricians who were familiar with NRT patches ...... 92 Table 12 Maternal morbidities based on reasons for selection ...... 112 Table 13 Characteristics of women who smoked at delivery, 2011-2012 ...... 122 Table 14 Distribution of NRT patches and varenicline use in the 12 months postpartum among women who smoked at the time of delivery, NSW and WA, 2011-2012 ...... 125 Table 15 Association between maternal morbidities and poor birth outcomes in relation to varenicline use in the 12 months postpartum, 2011-2012 ...... 129 Table 16 Association between maternal morbidities and poor birth outcomes in relation to NRT patches use in the 12 months postpartum, 2011-2012...... 131 Table 17 Description of the smoking information available and methods used to define smoking cessation outcome ...... 146 Table 18 Characteristics of women who were current smokers at the time of delivery for the first pregnancy in the pair according to varenicline exposure status in the inter- pregnancy interval, NSW, 2008-2012 ...... 154 Table 19 Association between varenicline exposure during the inter-pregnancy interval and smoking cessation prior to the second pregnancy ...... 157 Table 20 Sensitivity analysis in examining association between varenicline dispensing during inter-pregnancy interval and smoking cessation prior to the second pregnancy, concessional beneficiaries ...... 159

xviii

List of Chapter figures

Figure 1 The 5A’s process for smoking cessation interventions and the reported compliance with each step by obstetric care providers ...... 8 Figure 2 Selection of eligible calls made to MotherSafe between 1 January 2001 and 31 December 2016 ...... 56 Figure 3 Likelihood for call regarding each smoking cessation pharmacotherapy to be from providers compared to other categories of medications, 2001-2016 (and 2008- 2016 for varenicline) ...... 59 Figure 4 The Theory of Planned Behaviour* ...... 66 Figure 5 Selection of women who smoked at delivery, NSW and WA ...... 109 Figure 6 Graphical representation of maternal morbidities and poor birth outcomes that could potentially motivate a woman who smoked at delivery to using smoking cessation pharmacotherapy in the 12 months postpartum ...... 116 Figure 7 Cumulative prevalence of varenicline and NRT patches use in the 12 months postpartum among women who smoked at delivery, NSW and WA, 2011-2012 ...... 127 Figure 8 Selection of a cohort of women smoking at delivery and had a subsequent pregnancy in NSW, 2008-2012 ...... 152

xix

List of Abbreviations

ABS Australian Bureau of Statistics ACOG American College of Obstetricians and Gynecologists ASGS Australian Statistical Geography Standard aOR Adjusted odds ratio APDC Admitted Patient Data Collection ATC Anatomical Therapeutic Chemical CHeReL Centre for Health Record Linkage CI Confidence interval CPDRC Continuing Professional Development and Revalidation Committee FDA Food and Drug Administration GBP Great Britain Pounds GORD Gastro-intestinal oesophageal reflux disease GP General Practitioner HR Hazard Ratio ICD-10-AM International Classification of Diseases, version 10, Australian Modification IQR Inter-quartile range IRSD Index of Relative Socio-economic Disadvantage MMOI Maternal morbidity outcomes indicator MNS Midwives Notification System MUMS Maternal Use of Medications and Safety NAOI Neonatal adverse outcomes indicator NSAIDs Non-steroidal anti-inflammatory drugs NRT Nicotine replacement therapy NSC Neonatal special care NSW New South Wales OR Odds ratio OTC Over-the-counter POA Postal Area PBS Pharmaceutical Benefits Scheme PDC Perinatal Data Collection PPROM Preterm premature rupture of membrane RANZCOG Royal Australian and New Zealand College of Obstetricians and Gynaecologists RCT Randomised controlled trial RR Relative risk SAS Statistical Analysis System SEIFA Socio-economic Indexes for areas SES Socio-economic status SGA Small for gestational age SIDS Sudden infant death syndrome SLA Statistical Local Area SPSS Statistical Package for the Social Sciences xx

SURE Secure Unified Research Environment TGA Therapeutic Goods Administration TIS Teratology information service UK United Kingdom UNSW University of New South Wales US United States USD United States Dollars WA Western Australia WHA Women's & Children's Healthcare Australasia WHO World Health Organization

xxi

Chapter 1 Introduction and review of literature

This chapter describes the epidemiology of smoking and smoking cessation (1) during pregnancy, and (2) after pregnancy, with a particular focus on smoking cessation pharmacotherapies as means of promoting smoking cessation. A detailed background on the pharmacological aspects of smoking cessation pharmacotherapies is provided, followed by an outline of the evidence gaps this thesis addresses.

For the purpose of this chapter, studies relating to the combustible form of tobacco are reviewed. Compared to non-combustible products and electronic cigarettes, conventional cigarettes are the most common form of tobacco product used by pregnant women [1] in most countries, especially in Australia [2]. Maternal use of non- combustible products and electronic cigarettes is associated with poor outcomes, including increased risk of stillbirths, low birth weight, preterm births, small for gestational age, and pulmonary complications [3]. However, there is little evidence on strategies for reducing exposure to these non-combustible products and electronic cigarettes. This review focuses on research carried out in high-income countries, with an emphasis on Australian data where possible, as women in Australia are the target population of this thesis.

. Smoking and smoking cessation during pregnancy

Smoking is a leading cause of preventable pregnancy complications. It is also a major risk factor for poor maternal and foetal health outcomes.

Women who smoke tobacco during pregnancy are at increased risk of developing health problems such as gestational diabetes, eclampsia, deep vein thrombosis, gastrointestinal ulcers, cardiovascular events such as myocardial infarction and stroke, and respiratory disorders such as asthma and infections [4]. Smoking during pregnancy increases the risk of developing postpartum depression by 133% (pooled odds ratio (OR) 2.33, 95% CI 1.93-2.81) [5]. A 2019 population-based study in Finland found that during the period of 1999-2015, women who smoked during pregnancy were more likely to be hospitalised due to mental health disorders (adjusted OR (aOR) 3.88, 95% CI 3.71-4.06)), respiratory disorders (aOR 1.61, 95% CI 1.52-1.71) and genitourinary disorders (aOR 1.29, 95% CI 1.23-1.35) compared to pregnant women who did not smoke [6].

In terms of pregnancy complications, smoking during pregnancy is associated with increased risk of ectopic pregnancy, miscarriage, spontaneous abortion and placental complications [7-9]. Ectopic pregnancy accounts for almost 10% of pregnancy-related deaths as well as placing the woman at increased risk for subsequent ectopic pregnancies [10]. Nicotine and other toxic chemicals in tobacco smoke can cause premature rupturing of membranes and subsequent separation of the placenta from the uterine wall [11]. Smoking during pregnancy increases the risk of placenta abruption and placenta previa by 50-300% [9].

Foetuses exposed to maternal tobacco smoking are at increased risk of intrauterine growth retardation and major birth malformations [12, 13]. Prenatal exposure to tobacco accounts for approximately 5-8% of preterm births [14], 13-19% of low birth weight in term neonates [15], 5-7% of other causes of neonatal death [16], and 22-24% of deaths due to Sudden Infant Death Syndrome (SIDS) [9]. A child born to a mother who smoked tobacco during pregnancy is more likely to develop respiratory illnesses, obesity, infantile colic, and experience adverse cognitive and behavioural effects

including attention deficit hyperactivity disorder, reduced general intellectual ability, and aggressive and antisocial behaviours [17-21]. A 2018 Australian study found that infants born to women who smoked during pregnancy had poorer health outcomes than those born to a mother who did not smoke. They had an increased risk of being admitted to the neonatal intensive care unit (aOR 1.34, 95% CI 1.27-1.43), and for having any severe neonatal outcomes (aOR 1.35, 95% CI 1.28-1.43) [22].

There are also substantial healthcare costs associated with tobacco-related maternal and foetal morbidities. To date, there is no published Australian data that consider on cost to the healthcare system as a result of maternal smoking during pregnancy. However, data from the United States (US) indicates that smoking during pregnancy costs an additional USD 279, in 2004 dollars, per maternal smoker, to treat both maternal and neonatal health complications [23]. In the United Kingdom (UK), the healthcare system spent up to GBP 64.0 million(2005), to provide medical care for smoking-related health issues among mothers (GBP 8.0-64.0 million) and infants (GBP 12.0-23.5 million) [24].

Globally, the estimated prevalence of smoking during pregnancy was estimated to be 1.7% (95% CI 0.0-4.5) in 147 countries between 1985 and 2016 [25] . The authors of this report, however, observed a wide variation between countries and World Health Organization (WHO) regions. In terms of WHO regions, the prevalence of smoking during pregnancy was the highest in the European region (8.1%, 95% CI 4.0-12.2) and the lowest in the African region (0.8%, 95% CI 0.0-2.2). In terms of countries, Ireland was found to have the highest prevalence (38.4%, 95% CI 25.4-52.4) and Tanzania had the lowest prevalence (0.2%, 95% CI 0.0-0.6) [25]. The authors estimated that for Australia, the prevalence of smoking during pregnancy was 24.4% (95% CI 19.1-30.2) [25].

In the aforementioned report, the estimate for Australia, 24.4% (95% CI 19.1-30.2) was based on a meta-analysis of 20 studies [25]. The data meta-analysed were extracted from studies conducted in selected samples between 1985 and 2010, whereby the reported prevalence estimates ranged between 2.0% and 40.1% [25]. The majority of the meta-analysed studies reported prevalence estimates from early in the study

period, and their samples were such that these studies might not represent the entire population of pregnant women in Australia. Thus, a cautious interpretation of the estimate is warranted, especially as it appears to be higher than the estimates reported by the Australian national surveillance system.

According to the Australian national surveillance system, the prevalence of smoking during pregnancy in Australia was 9.9% in 2017 [26]. This surveillance system produces annual estimates based on the annual birth data collected by each state and territory in Australia, which relies upon the mandatory reporting from midwives and other birth attendants [26]. Although the prevalence of tobacco use during pregnancy in Australia has declined over time, this decline has been slower in recent years (14.6% in 2009, 11.7% in 2013, 11.0% in 2014, 10.4% in 2015, 9.9% in 2016, and 9.9% in 2017) [26]. In New South Wales (the most populous state in Australia) (NSW), the prevalence and trend of maternal smoking mirrors the national estimates. The prevalence of smoking during pregnancy in NSW was 12.0% in 2009, 9.7% in 2013, 9.3% in 2014, 8.9% in 2015, 8.3% in 2016, and 8.9% in 2017 [27].

Disparities in the prevalence of smoking exist among sub-populations of pregnant women [28]. Of concern, the prevalence of maternal smoking is socially and environmentally patterned [29]. This is seen even in high-income countries whereby women who smoke during pregnancy are more likely to live in high-density neighbourhoods and in rural and socio-economically disadvantaged areas, to have a low level of education, to earn low incomes and have a partner who smoked [30]. By race and ethnicity, the highest prevalence of smoking during pregnancy occurs among women in indigenous and ethnic minority groups such as Native Americans and Alaskan Natives (16.7-26.0%), Canadian First Nations populations (18.0-90.0%) and New Zealand Maori (34.0%) [31, 32], with the lowest prevalence found in Asian and non-Hispanic women (0.6%) [32]. In Australia, in 2017, the prevalence of maternal smoking was higher in Indigenous women than in non-Indigenous women (44.3% vs 11.8%), and higher among those who live in remote and very remote areas compared to women who live in major cities (51.3% vs 7.2%) [26]. A higher proportion (17.8%) of maternal smokers live in socio-economically disadvantaged areas, compared with 2.9% in advantaged areas [26].

The prevalence of smoking during pregnancy is also disproportionately high among women with co-morbidities, including mental health disorders [33] and substance use disorders [34]. Women with a depression diagnosis are almost three times more likely to smoke during pregnancy than those without such diagnosis (pooled OR 2.65, 95% CI 1.62–4.30) [29]. Compared to women without alcohol, cannabis or cocaine dependence, women who had substance use disorder are twice as likely to smoke during pregnancy (pooled OR 2.03, 95% CI 1.47–2.80) [29].

Given the magnitude of health risks associated with smoking during pregnancy and the reasonably high prevalence in most high-income countries, a reduction in the prevalence of smoking during pregnancy is an important public health priority.

1.1.3.1 Health benefits of smoking cessation during pregnancy Smoking cessation, even if occurring after the first trimester, has demonstrated favourable health outcomes for both women and their children [35, 36]. This reduction in health risk occurs for all women who quit smoking regardless of the number of years they have smoked or the number of cigarettes smoked per day [35, 36].

The benefits of smoking cessation include reduced risk of preterm birth, early miscarriage, stillbirth, SIDS and improved neuro-behavioural development of the child [35-38]. In women with mental health disorders, smoking cessation improves mood; specifically, it reduces depression, anxiety, and stress-related symptoms [39]. Children born to mothers who quit smoking during pregnancy have better neuro-behavioural development than those born to mothers who smoked throughout pregnancy [40, 41].

1.1.3.2 Barriers to smoking cessation during pregnancy Pregnancy appears to be a teachable moment, with increased levels of motivation to quit smoking. Several studies reported that a higher proportion of pregnant smokers made at least one quit attempt during pregnancy compared with the non-pregnant population of smokers in their lifetime (56-%-85% [42, 43] vs 36-40% [44, 45]). A 2018 longitudinal, population-based study in the US found that the likelihood of quitting

smoking increased by 6.5 times when women were pregnant compared to when they were not pregnant (aOR 6.49, 95% CI 4.18-10.08) [46].

Despite the established benefits of smoking cessation and the high level of motivation to quit during pregnancy, the proportion of women who successfully quit smoking during pregnancy remains modest. An estimated 4-25% of women who smoke at the start of their pregnancy reported not smoking at the end of the pregnancy [47-49].

Factors associated with difficulty in quitting smoking in the general population are augmented in pregnancy. These factors include stress and high nicotine dependence [29]. Given that about half of pregnancies in maternal smokers are unplanned [50], pregnancy itself may be a stressor. Women who have low self-confidence (self- efficacy) in their ability to refrain from smoking, may continue to smoke to cope with pregnancy-related stress [51] such as concerns regarding caring for infants and birth [52]. Moreover, in pregnancy, nicotine metabolism increases and thus, it is more rapidly cleared from the body [53]. This results in greater craving for nicotine, stronger withdrawal symptoms and potentially a higher level of tobacco smoking than outside pregnancy [54].

1.1.3.3 The 5A’s approach in healthcare settings Given the amplified barriers experienced by women who smoke during pregnancy and the low rates of quitting during pregnancy, it is important that evidence-based smoking cessation interventions are available to women who smoke during pregnancy. Multiple antenatal care encounters between pregnant women and healthcare providers throughout pregnancy provide plenty of opportunities to screen for smoking and offer smoking cessation support. Moreover, about half of pregnant smokers are receptive to being offered smoking cessation support [55, 56]. This indicates that healthcare providers are well-positioned to routinely intervene with pregnant smokers.

Guidelines for healthcare providers to promote smoking cessation during pregnancy have been issued by various professional organisations including the WHO [57], the US Department of Health and Human Services [58, 59], the UK's National Health Service [60] and the Australian Royal College of General Practitioners [61]. These guidelines recommend that cessation support consistently be provided to all pregnant

women who smoke during routine antenatal care. These guidelines recommend that the smoking cessation support include the following brief 5-step process (see Figure 1): (1) Ask about smoking history and smoking behaviour, (2) Advise maternal smokers to quit, (3) Assess maternal smokers' intention to make a quit attempt, (4) Assist smokers with treatment and referrals and (5) Arrange follow-up throughout pregnancy and the postpartum period [58, 60].

Figure 1 also reports the proportion of obstetric care providers who reported ‘always’ or ‘often’ carrying out each of the five steps with women who smoke during pregnancy. These pooled proportions were extracted from a 2019 meta-analysis of 33 studies in which obstetric care providers in high-income countries completed self-reported surveys [62]. The authors of this meta-analysis found that these providers were relatively compliant to the Ask, Assess and Advise behaviours, but a much lower proportion completed the Assist and Arrange steps with only 59.1% (95% CI 56.0- 62.2%) of the obstetric care providers performing the Assist step and 33.3% (95% CI 20.4-46.2%) carrying out the Arrange step [62]. This finding points to a need to explore the components of the Assist step as the first move to improve provider-delivered smoking cessation interventions to pregnant smokers.

The Assist step consists of provision of smoking cessation treatment, which can be categorised into psychosocial and pharmacological interventions. These two treatment modalities are described in more detail in the following sections (1.1.3.4 and 1.1.3.5).

Figure 1 The 5A’s process for smoking cessation interventions and the reported compliance with each step by obstetric care providers

Reported compliance, Description of the five steps in the 5A’s process Pooled % (95% CI)*

•Screen for smoking history and behaviour among pregnant women at every clinical 91.6% encounter (88.2- 95.0%) Ask •Document smoking status at each clinical encounter

90.0% •Provide clear advice to quit smoking (72.5-99.3%) •Discuss health benefits of quitting Advise

79.2% •Assess intention to make a quit attempt (76.5-81.8%) •Assess nicotine dependence Assess

•Provide prescriptions for smoking cessation 59.1% pharmacotherapy / counselling (56.0-62.2%) •Refer patients to Quitlines / tobacco treatment Assist specialist

33.3% •Arrange follow-up contact (20.4-46.2%) Arrange

* the percentages represent the proportion of obstetric care providers who reported that they ‘often/always’ provide the nominated step of the 5A’s with women who smoke during pregnancy [62]

1.1.3.4 Psychosocial interventions In managing tobacco cessation, psychosocial intervention is defined as any non- pharmacological strategy that is designed to bring out a behavioural change, that is, quitting smoking [63]. It enhances the smoker’s motivation to quit and increases their self-efficacy to not smoke when they experience the urge to smoke [63]. It also offers support to cope with nicotine withdrawal and relapse, especially stress-induced triggers for smoking.

Types of psychosocial interventions In the most recent (2017) systematic review of 102 controlled trials (including randomised, cluster-randomised and quasi-randomised) focusing on psychosocial smoking cessation interventions for pregnancy, the evaluated interventions included counselling, health education, feedback, incentives, social support, exercise, and alternative delivery of counselling [63]. The evaluated interventions are described in Table 1.

In Australia, the types of psychosocial interventions available to the general population are counselling delivered through telephone, face-to-face or technology, health education, and social support. The Quitline services are accessible to the public at the cost of a local call [61]. Technology-based counselling comprises individualised counselling delivered online, including the Quit Coach [64] in Victoria and ‘iCanQuit’ in NSW [65]) as well as mobile-based applications (the app ‘Quit for You-Quit for Two’ [66]). Australian printed materials include the self-help booklet ‘Quit Because You Can’ containing general information to assist a smoker to quit smoking [67]. Available social support interventions include the nationally, accredited Victoria-based course ‘Fresh Start’ [68]. Led by a trained educator, this group course may take place in different settings that include healthcare settings [68]. Pregnant women can also access pregnancy-specific psychosocial interventions including the mobile-based applications (the app ‘Quit for You-Quit for Two’ [66]) as well as ‘Quit for new life’ program in NSW which comprises the provision of free NRT (up to twelve weeks) combined with psychosocial interventions including offering of advice to quit smoking, delivery of behavioural strategies, and referral to Quitline [69].

Table 1 Characteristics and efficacy of psychosocial interventions for smoking cessation that have been evaluated in pregnancy* Intervention Brief description Efficacy** (RR of smoking cessation, 95% CI) [63] Counselling Aims to increase motivation to quit and Intensive counselling and strategies to cope with nicotine (1.44, 95% CI 1.19-1.73) behavioural withdrawal and relapse. This category interventions encompasses motivational interviewing, Less intensive counselling cognitive behaviour therapy, (1.25, 95% CI 1.07-1.47) psychotherapy, relaxation, problem- solving facilitation, and other strategies.

Intensity differs according to the duration and frequency of the session. Intensive counselling may last up to an hour and include multiple sessions.

Includes telephone-based, face-to-face, and technology-based (mobile phones, computer-assisted).

Health Offers information on smoking-related 1.59, 95% CI 0.99-2.55 education risks and advice to quit without additional support such as information on how to quit.

Feedback Foetal health status and/or maternal 4.39, 95% CI 1.89-10.21 regarding the levels of tobacco by-products are (in conjunction with biologic measured and communicated to mothers. counselling) measurements

Incentive Provides financial incentives to the The pooled effect was not woman, subject to her quitting smoking. calculated due to substantial heterogeneity in effect estimates for usual care.

2.36, 95% CI 1.36-4.09 (in comparison with alternative intervention)

Social support Support provision from a peer, 1.42, 95% CI 0.98-2.07 (peer and/or professional, and/or partner to promote partner) smoking cessation.

Exercise Structured physical activity support. 1.20, 95% CI 0.72 -2.01 (in conjunction with behavioural support)

Dissemination Using different methods (active vs 1.63, 95% CI 0.62-4.32# of interventions passive) to deliver the smoking cessation intervention *adapted from Chamberlain et al. [63] **when compared against usual care #other strategy that was not within the evaluated category of psychosocial interventions, and this include one trial that addressed the dissemination of counseling program 10

Efficacy of psychosocial interventions Table 1 also summarises the pooled relative risk of quitting smoking for each category of psychosocial intervention evaluated compared to usual care. Overall, this 2017 review, that was published in 2020, found that psychosocial interventions increased the proportion of pregnant women who quit smoking by 29-40 weeks of gestation by an estimated 35% compared to usual care [63].

Compared to usual care, interventions comprising counselling, feedback, and incentives increased smoking cessation in late pregnancy by 25-44% (see Table 1). A strong effect was observed for intensive counselling (pooled risk ratio (RR) 1.44, 95% CI 1.19-1.73, based on 30 studies) [63] and a lesser effect was seen for less intensive counselling (pooled RR 1.25, 95% CI 1.07-1.47 based on 18 studies). However, the authors could not establish whether the effect is solely driven by counselling as counselling was part of a broader intervention to improve maternal health. Moreover, the effect of counselling was unclear when comparing counselling of different intensities (intensive vs less intensive).

Although the largest effect was observed for feedback interventions (pooled RR 4.39, 95% CI 1.89-10.21), this estimate should be interpreted with caution. As this estimate (pooled RR 4.39, 95% CI 1.89-10.21) was derived from two studies only, the pooled effect could not be estimated precisely. Moreover, this effect was seen only when feedback was provided with counselling.

Generally, there is moderate-quality evidence showing that psychosocial interventions designed to promote smoking cessation in pregnant women are efficacious [63]. Studies with biochemically validated smoking abstinence assessment (21 trials; RR 1.23, 95% CI 1.04-1.45) demonstrated similar efficacy compared with studies that included self-reported abstinence (30 trials; RR 1.44, 95% CI 1.19-1.73) [63]. However, whether these findings are generalisable beyond the participants of these trials is unclear [63]. This 2017 review found that information on the participation rates was absent for many of the included trials, and low participation rates were reported for those studies with such information. This suggests that the included trials may have been based on selected samples of pregnant smokers. Moreover, an older, 2016 review of 27 smoking cessation trials reported that few maternal smokers (mean proportion 13%, 95% CI 6.0-10.0%) were able to remain abstinent at the end of

pregnancy, with the majority resuming smoking after delivery [70]. When restricted to the 17 trials with biochemically validated self-reported smoking status, a slightly lower proportion of maternal smokers achieved abstinence (11%, 95% CI = 9–14%) [70]. Importantly, there is a lack of clear evidence indicating which strategy or a combination of these strategies can consistently increase smoking cessation during pregnancy and maintain it beyond pregnancy. One key reason is that the majority of the assessed psychosocial interventions encompassed a range of behavioural change strategies, tailored to address the needs of individual women. It is not known about the optimal frequency, duration, and format (including its intensity) of multicomponent psychosocial interventions. Moreover, it is unclear whether it is feasible for these psychosocial interventions to be administered by healthcare providers in routine antenatal care settings.

1.1.3.5 Pharmacological interventions Smoking cessation pharmacotherapies assist with smoking cessation by managing physiological aspects of nicotine dependence. They reduce the intensity of nicotine withdrawal symptoms and/or block the reinforcing effects of nicotine [71].

In Australia and many developed countries, three pharmaceutical products are registered as smoking cessation aids: Nicotine Replacement Therapies (NRT), bupropion, and varenicline. These are all considered first-line pharmacotherapies for smoking cessation. In Australia, it is a condition of subsidised access to these pharmacotherapies that the patient must be receiving behavioural support [72]. General information regarding each of these pharmacotherapies including its clinical pharmacology, efficacy, and safety in the general population, and product details are reviewed in greater detail in a later section of this chapter (section 1.3 Overview of smoking cessation pharmacotherapy).

Briefly, NRT is the only nicotine-based pharmacotherapy. Introduced in 1982, NRT aims to reduce the withdrawal symptoms, including craving associated with quitting tobacco use [73]. It acts on nicotine dependence without exposure to the toxic combustible constituents of tobacco smoking [73]. Both bupropion and varenicline are newer medications, with bupropion marketed in 2001 as a smoking cessation pharmacotherapy [74] and varenicline introduced in 2008 [75]. As non-nicotine

pharmacotherapies, bupropion mimics the rewarding effects of nicotine [74] whereas varenicline both mimics the rewarding effects of nicotine and blocks the reinforcing effects of nicotine from cigarettes [75].

Efficacy and effectiveness of smoking cessation pharmacotherapies

In the general population, smoking cessation pharmacotherapies increase the cessation rate by 55-124% compared to placebo [76-78].There is high-quality evidence that bupropion and varenicline increase the cessation rate by 62% (45 trials; pooled RR 1.62, 95% CI 1.49-1.76) [76] and 124% (27 trials; pooled RR 2.24, 95% CI 2.06-2.43) [77], respectively..

Pregnant women differ from the general population with regard to physiology and healthcare behaviour. Physiological changes during pregnancy affect the pharmacokinetic properties of medications [79]. Observed changes included reduced gastrointestinal motility and increased gastric pH (reducing the absorption of medications), increased total body water and plasma volume and decreased concentrations of drug-binding proteins (affecting the distribution of medications), altered activity of drug-metabolizing enzymes in the liver and increased glomerular filtration rate (affecting hepatic metabolism and renal clearance medications) [80]. The pharmacokinetic changes of medications affect the corresponding pharmacodynamic responses [81]. Little is known about how these changes manifest, with the 2016 review of 198 studies unable to ascertain the extent of change in the clinical response to the 121 examined medications that were associated with known, altered pharmacokinetics during pregnancy [82]. With respect to behaviour change, prior studies have shown that pregnant women have more safety concerns about medication use than the non-pregnant population, due to the effects in their unborn baby [83, 84]. This, in turn, may lead to reported low adherence (36.2%-48.8%) to medications [85- 87], which is likely to diminish the effectiveness of medications in pregnant women.

NRT has been studied the most extensively during pregnancy compared to bupropion and varenicline. This is likely to be due, at least in part, to the assumption that NRT is no less safe than continued smoking during pregnancy [88]. Yet systematic reviews have not been conclusive regarding the effectiveness of NRT during pregnancy as a result of small sample sizes, short duration of follow-up, poor adherence and

heterogeneity in the findings, possibly due to pooling of results relating to different formulations of NRT [71, 89].

There are nine randomised controlled trials that have evaluated the efficacy of NRT in pregnant women [90-98], six of which are controlled trials that compared NRT patches to placebo. A 2015 (most recent) meta-analysis of five of these trials found no significant difference in the biochemically validated quit rate in late pregnancy between pregnant women who used NRT and the placebo groups, despite all trials returning findings in the direction of effectiveness (pooled RR 1.28, 95% CI 0.99-1.66) [71]. The authors of this review, however, acknowledged that their capacity to detect significant findings was limited by the low statistical power [71]. These authors also cautioned against interpreting their findings as a null effect due to low adherence rates of less than 25% observed in four of the five included studies [71]. In addition, these trials differed in the treatment received by the control groups, which received placebo patches [92, 93, 96, 98] or placebo gum [97], which led to heterogeneity in the findings for the control group. When two additional, non-placebo controlled trials were included, the review found evidence that NRT increased the quit rate by 40% (pooled RR 1.41, 95% CI 1.03-1.93) [71]. However, it was unclear whether this effect could be solely attributed to NRT exposure, as NRT was administered in combination with psychosocial interventions. Since the 2015 review was published, a 2019 randomised controlled trial of NRT inhalers was published, bringing the current total to six controlled trials that investigated the effectiveness of NRT compared to placebo. In this 2019 controlled trial, women who used NRT inhalers had similar quit rates as women in the placebo group at 32-34 weeks of gestation (10% of the NRT-exposed group, n=70, vs 18% of the placebo group, n=67, p=0.22) [90].

As for the newer pharmacotherapies for smoking cessation, bupropion and varenicline, data regarding their efficacy are substantially lacking, especially for varenicline. Few trials have been conducted to assess the efficacy of bupropion during pregnancy. A 2015 US-based study examining the effectiveness of bupropion failed to recruit a sufficient number of pregnant smokers over the 2-year trial period, resulting in the inclusion of data from only 11 women [99]. Another US-based trial of 65 pregnant smokers found no significant differences in the quit rate between women who used bupropion and those who received counselling, however, the proportion of women quitting smoking appeared to be higher among bupropion-exposed women compared

to the placebo group (19% vs 2%) [100]. In addition to limited statistical power, the high rate of early withdrawal (20% retention rate) potentially introduced bias into this study due to differential attrition [100]. At the time of writing this thesis, there is one ongoing trial evaluating the efficacy of bupropion in pregnancy [101].

There are no well-designed experimental studies that have evaluated varenicline effectiveness in pregnant women. There is one ongoing trial that is evaluating the effectiveness of varenicline in pregnancy [102].

27.4% vs 14.7%), with an average difference of 12.7% (95% CI 0.8-24.6%) [103]. Despite the inherent limitations of observational research, the rigorous methods employed in this study, which included the use of propensity scores to reduce the impact of confounding, lends support to the robustness of the findings [103].

Utilisation of pharmacotherapy for smoking cessation in pregnancy For the potential of smoking cessation pharmacotherapies to promote smoking cessation to be realised, these pharmacotherapies must be prescribed by healthcare providers and used by women who smoke during pregnancy. There is considerable variation in population-based estimates of the prevalence of smoking cessation pharmacotherapy use in pregnancy. This appears to be related to the different study periods, patient populations and countries.

In the US, two studies using insurance claims data reported that a smoking cessation pharmacotherapy was used by 2.4% [104] and 9.3-11.1% [105] of pregnant smokers. A 2015 study carried out in Maryland estimated that 2.4% of smokers ever filled a

prescription for NRT during pregnancy, and almost none filled a prescription for varenicline [104]. The finding of low varenicline use was possibly due to its limited availability during the study period (2003 - 2010), as varenicline became available in 2008. In contrast, a more recent study (2017) based on aggregate data from Kansas reported a substantially higher prevalence of use for all three pharmacotherapies during pregnancy; 11.1% in 2010 and 9.3% in 2013 [105]. However, this ecological study used state and national estimates of birth rates and smoking prevalence as opposed to using individual-level data on pregnancies and smoking status [105]. Hence, cautious interpretation of the high estimate from this study is warranted.

In the two studies that have been conducted in the UK, NRT was estimated to be used in 84.8% [106] and 11.4% of pregnancies among smokers [107]. The observed disparity in the estimates could be due to the patient population studied and the time period of the data collected. The 84.8% estimate was derived from routinely collected data relating to women who attended specialist Stop Smoking Services between 2009 and 2011 [106]. These services routinely provide NRT to pregnant women [108]. The estimate of 11.4% was returned in a study of pregnant smokers who attended general practices between 2001 and 2012 [107]. The difference in these estimates is likely due to the high level of motivation among the pregnant smokers who attended the specialist services, as well as the time period over which the two studies were conducted.

In Canada’s Quebec province, a study using insurance claims data for social security beneficiaries found that 5.6% of pregnant smokers used bupropion and 24.5% used NRT between 1998 and 2009 [109]. This may not reflect the true estimate because the identification of smokers was based on self-reported smoking status before pregnancy, as opposed to during pregnancy.

In an Australian study of women who gave birth in the states of NSW and Western Australia (WA) between 2011 and 2012, 2.3-3.6% of maternal smokers used a pharmacotherapy for smoking cessation during pregnancy [110]. Varenicline was dispensed to 1.0-1.8% of maternal smokers and few maternal smokers were dispensed with bupropion 0.1-0.2% [110]. Subsidised NRT patches were dispensed to 1.3%-1.7% of maternal smokers [110]. The authors acknowledge that this was likely to be an underestimation of all NRT use during pregnancy, as at the time of the study, only NRT patches were subsidised in Australia, with both patches and other forms available over

the counter [110]. It is unlikely to be substantially underestimated, however, given the considerable cost advantage of accessing NRT under subsidy.

There were four other population-based estimates that were reported as incidental findings in studies relating to the effectiveness and safety of NRT in pregnancy. These estimates ranged between 2.0-2.5% of pregnancies in Denmark between 1996 and 2002 [111, 112], and 11-16% in Scotland between 2005 and 2007 [113, 114]. It should be noted that these estimates are reported as a proportion of all pregnancies, regardless of whether the mother smoked or not.

In brief, there is much variation in the estimates of smoking cessation pharmacotherapy use over time and between patient populations. This variation suggests that the extent to which these pharmacotherapies are used during pregnancy is dependent on the context, and may, therefore, be amenable to change by targeting the factors influencing their use. In particular, the proportion of maternal smokers using pharmacotherapies for smoking cessation is substantially lower in Australia than in other high-income countries such as the US, UK, Canada, and Denmark. Identifying factors influencing the use of smoking cessation pharmacotherapy during pregnancy in Australia will be useful to inform the design of interventions that may increase the use of these pharmacotherapies in this vulnerable population of smokers in the future.

Factors influencing prescribing of smoking cessation pharmacotherapy to pregnant women Qualitative evidence suggests that pregnant smokers are receptive to using smoking cessation pharmacotherapy [115]. In Australia, a 2008 survey in NSW found that 87% of women who smoked during pregnancy supported the provision of NRT during pregnancy [116]. Furthermore, the findings of a hospital-based 2014 survey in another Australian state revealed that 47.6% of pregnant smokers cited pharmacotherapy as their preferred smoking cessation method [117]. In qualitative studies, pregnant smokers expressed their want for smoking cessation pharmacotherapy but not receiving one, leading to the authors citing the reluctance of healthcare providers to provide these pharmacotherapies as one of the barriers to quitting smoking [116, 118]. This raises the question of whether the disinclination of healthcare providers contributes to the low uptake of pharmacotherapy during pregnancy. This hypothesis is

supported by a 2019 meta-analysis which found that few healthcare providers reported that they ever prescribed (six studies, pooled proportion=25.4%, 95% CI 12.8-38.0%) or always prescribed (four studies, pooled proportion=6.2%, 95% CI 4.9-7.4%) NRT to pregnant women [62]. If such a disclination exists, it is useful to understand the underlying reasons so that they may be addressed through strategies designed to improve prescribing of smoking cessation pharmacotherapies to pregnant women.

Concerns regarding safety of smoking cessation pharmacotherapy in pregnancy According to the WHO’s guide to good prescribing practices, the factors that need to be considered when prescribing a medication include the efficacy, safety, suitability and affordability of a medication [119]. Of these, safety is likely to be the highest order consideration, particularly for prescribing during pregnancy as the potential harms to both the women and foetuses need to be considered.

There is a lack of clear evidence regarding whether it is safe for foetuses to be exposed in utero to smoking cessation pharmacotherapies. This is reflected by the pregnancy risk categories assigned to these medications. The pregnancy risk category for NRT is D [120]. Category D medications include those suspected of causing irreversible harm to the foetus [120]. The risk categories for bupropion and varenicline are B1 and B3, respectively, where these categories indicate that limited evidence from humans exists, but animal studies have not shown evidence of harm (category B1), or evidence from animal studies is inadequate (category B3) [120].

While NRT has been studied to a greater extent than bupropion and varenicline, the evidence regarding its safety is not conclusive. It is reassuring, however, that there is no clear evidence that NRT exposure during pregnancy causes foetal harm. As for varenicline and bupropion, the existing studies suggest that use of these pharmacotherapies is not associated with an increased risk of many adverse birth outcomes, however, important outcomes such as congenital anomalies, and stillbirth remain to be examined with adequately powered studies.

In the most recent (2015) meta-analysis investigating the safety of using NRT in pregnancy, there were lower rates of preterm birth (pooled RR 0.81, 95% CI 0.60- 1.11), low birthweight (pooled RR 0.58, 95% CI 0.27-1.26), admission to neonatal 18

intensive care (pooled RR 0.90, 95% CI 0.64-1.27), and neonatal death (pooled RR 0.66, 95% CI 0.17-2.62) in the NRT-exposed group than the placebo group [71]. Although the confidence intervals were wide and the observed reductions in risk did not reach statistical significance, it appears from the direction of the odds ratios that NRT probably did not increase the risk of these adverse pregnancy outcomes. Wide confidence intervals and inconclusive findings were also returned in the analysis of the risk of miscarriage or spontaneous abortion (pooled RR 1.47, 95% CI 0.45-4.77; 4 studies), stillbirth (pooled RR 1.24, 95% CI 0.54-2.84; 4 studies), and congenital abnormalities (pooled RR 0.73, 95% CI 0.36-1.48; 2 studies) [71]. Since this meta- analysis, a 2019 placebo-controlled trial investigating the safety of NRT inhalers among pregnant smokers has been published [90]. Although this trial found a 4% improvement in the preterm birth rate for pregnant smokers who received NRT inhalers, this trial was underpowered to demonstrate statistical significance [90].

The effect of maternal NRT exposure during pregnancy on longer-term infant outcomes has been examined in only one study. In a 2012 trial examining infant outcomes at two years of age, infants whose mothers were exposed to NRT patches during pregnancy were less likely to have a disability or behavioural or developmental impairment than infants born to women who received placebo patches [121]. While it was acknowledged that the lack of harm associated with NRT use could be due to the protective effect of smoking cessation [92], a secondary analysis of this 2012 data published in 2019 found that the positive health outcomes in the NRT-exposed infants were unlikely to be mediated by NRT-induced smoking cessation [122].

Large observational population-based studies have also not returned consistent findings regarding NRT use in pregnancy and adverse birth outcomes. Although in two studies, NRT exposure was associated with reduced risk of stillbirth, the effect did not reach statistical significance [112, 123]. In contrast, NRT exposure appeared to be associated with a statistically significant increased risk of major respiratory-specific [124], and overall, congenital malformations [125]. However, these imprecise estimates (RR 4.65, 95% CI 1.76-12.05, [124] and RR 1.61, 95% CI 1.01-2.58 [125]) do not provide clear evidence that NRT might cause harm, and suggest that larger studies are needed to confirm these findings. Although NRT use was associated with an increased risk of low birth weight babies in one study [126], this effect was not observed in another study [111]. However, the increased risk observed in the Gaither et al. (2008)

study is likely due to the type of NRT used where the formulation was not specified, and lack of adjustment for nicotine dependence [126]. Transdermal NRT exposes the foetus to a steady state dosing compared to intermittent dosing of NRT, and thus, may pose increased risk to the foetus. Women with high nicotine dependence may be more likely to be prescribed NRT, and thus, had an increased difficulty to quit smoking, compared to women with low nicotine dependence. The authors also acknowledged that women with high nicotine dependence may have used NRT in addition to smoking [126]. A 2020 Australian population-based study found no evidence of poor birth outcomes, such as preterm birth and small for gestational age associated with NRT patches use during pregnancy when compared to smokers who did not use any smoking cessation pharmacotherapy during pregnancy (44.8% vs 46.3%, Hazard ratio (HR) 1.02, 95% CI 0.84-1.23)) [127].

There are fewer high-quality experimental studies that have assessed the safety of bupropion and varenicline during pregnancy. Currently, there is an ongoing trial that separately investigates the safety for bupropion [101] and varenicline [102] exposure in pregnancy. The findings of a 2018 meta-analysis of 2 trials, 11 cohort studies, 2 case- control studies, and 3 case reports revealed no evidence of harm or safety for either bupropion or varenicline, noting the poor quality of the included studies [128]. Since then, two high-quality population-based studies have been conducted, with neither finding an increased risk of adverse maternal or foetal health outcomes among pregnant women who were exposed to these pharmacotherapies. This includes a 2019 population-based cohort study of varenicline carried out in Denmark and Sweden, which found that maternal exposure was not associated with an increased risk of adverse neonatal outcomes including preterm birth and SIDS [129]. Although the risks of congenital malformation and stillbirth were also measured, the study was not sufficiently powered to return meaningful results. In Australia, a 2020 population-based cohort study found no increased risk of any adverse perinatal event associated with the use of bupropion during pregnancy (39.2% vs 39.3%, HR 0.93 (95% CI 0.73-1.19)), relative to smokers not using a pharmacotherapy for smoking cessation [127]. This study also found a significantly reduced risk of adverse perinatal outcomes among the varenicline-exposed group compared to the unexposed smoker group (36.9% vs 40.1%, HR 0.86 (95% CI 0.77-0.97)) [127]. This study also demonstrated that varenicline use was less likely to be associated with an adverse birth outcome compared to NRT use (38.7% vs 51.4%, HR 0.58 (95% CI 0.33-1.05)), though this

effect did not reach statistical significance [127]. This study was also not adequately powered to determine the risk of congenital anomalies and stillbirth associated with bupropion or varenicline exposure during pregnancy [127].

In the absence of clear data on the entire range of risks and benefits associated with the use of smoking cessation pharmacotherapies while pregnant, the recommendations made by the Royal Australian College of General Practitioners clinical guidelines are cautious regarding the use of NRT, and are opposed to the use of bupropion and varenicline in pregnancy [61]. The latest (2013) guidelines on the management of tobacco use from the WHO states ‘The panel cannot make a recommendation on use or non-use of nicotine replacement therapy (NRT) to support cessation of tobacco use in pregnancy’ [57]. The guidelines in the UK [108] and the US [130] recommend that NRT be offered to pregnant smokers who are unable to quit using psychosocial interventions or with failed, unsupported quit attempts. Moreover, these guidelines impose caveats such as statements of ‘under close supervision’ and ‘only if women are motivated’. In Australia, the smoking cessation guidelines support the use of NRT only when pregnant women are unable to quit, in conjunction with medical supervision [61]. Overall, the guidelines advise clinicians to assess, on a case- by-case basis, whether the benefits of prescribing NRT to pregnant women are likely to outweigh the risks. Almost all smoking cessation care guidelines recommend that bupropion and varenicline not be used, with the occasional exception, including the American College of Obstetricians and Gynecologists (ACOG) guidelines which recommend cautious use of bupropion and varenicline during pregnancy [130].

These cautionary guidelines and the lack of clear safety evidence of smoking cessation pharmacotherapies, may be prompting concerns and reluctance from healthcare providers to prescribe these pharmacotherapies to pregnant women. This reluctance may not be warranted, given that smoking itself poses additional risks to the mother, foetus and the child (section 1.1.1). Research examining whether healthcare providers are overestimating the risk of these pharmacotherapies in pregnant women is required. This may provide insight into whether appropriate risk communication might lead to improved prescribing of smoking cessation pharmacotherapies to pregnant women.

Other factors influencing the prescribing of smoking cessation pharmacotherapies to pregnant women In addition to safety concerns, there may be other factors operating at the level of prescribers, that influence the prescribing of smoking cessation pharmacotherapies to pregnant women. It is important to identify these potential additional factors as they may then be the focus of interventions designed to improve prescribing behaviour.

Before considering prescribing a smoking cessation pharmacotherapy, it stands to reason that obstetric care providers must first be familiar with these pharmacotherapies. In recent studies conducted in the UK (2018) and Australia (2019), a lack of available information about NRT was cited by obstetric care providers as a barrier to prescribing NRT [131, 132]. Similarly, in a qualitative study carried out in NSW in 2016 [133], obstetric care providers cited a lack of knowledge about NRT as a barrier to providing NRT as a smoking cessation aid. Although this issue has not been explored for varenicline and bupropion, it is highly possible that awareness regarding varenicline and bupropion is lacking among obstetric care providers given that these two smoking cessation pharmacotherapies were introduced more recently.

Understanding factors that contribute to the lack of familiarity with these pharmacological therapies may inform the design of knowledge-based strategies to increase the prescribing of these pharmacotherapies to pregnant women. In particular, an exploration of factors operating at the level of the health services in which healthcare providers practice (hereafter referred to as ‘facility-level factors’) is warranted given that interventions targeting such factors have led to improvements in the provision of smoking cessation care in maternity hospitals [134], community mental health services [135], and primary care settings [136, 137].

Although knowledge is necessary, it alone is inadequate to influence prescribing decisions and prescribing practices [138]. A recent trial carried out in Aboriginal Medical Services in three Australian states found that although training (including educational materials and webinars) improved knowledge regarding NRT (correct scores increased from 68% to 79%, difference of 9.9, 95% CI 3.66-16.14), it did not result in greater prescribing rates for NRT [139]. A qualitative study among midwives and gynaecologists in Belgium also found that awareness of a pharmacotherapy did not lead to it being prescribed; gynaecologists did not recommend NRT despite having

an awareness of its safety during pregnancy [140]. Collectively, this implies that, in addition to knowledge, there are possibly other factors that influence clinicians’ prescribing of smoking cessation pharmacotherapies to pregnant women.

As argued in relation to knowledge, understanding the factors that influence the prescribing of smoking cessation pharmacotherapies among obstetric care providers’ may guide the design of interventions designed to change this behaviour. Again, facility-level factors are worthy of investigation given evidence that facility-level changes can improve the provision of smoking cessation care in multiple settings. Individual-level beliefs are also likely to be important, as prescribing is the behavioural outcome that stems from the complex interplay of determinants of the decision to prescribe and the eventual prescribing [141]. Apart from knowledge, other determinants of prescribing include perceived attitudes, expectation of patients and peers, and perceived ability to prescribe a medication [141, 142].

In summary, smoking cessation is challenging for pregnant women who smoke. This calls for greater support by healthcare providers to assist women to quit smoking during pregnancy. Although psychosocial interventions are the recommended approach to increase smoking cessation in pregnant women, data that delineate which component of evaluated psychosocial interventions can consistently promote smoking cessation among pregnant women are not yet available. Furthermore, providing psychosocial interventions to support pregnant smokers to quit smoking requires time and skill, which may not be feasible in routine antenatal practice settings. As such, it is important to offer other smoking cessation support that maximises the likelihood of success. The demonstrated effectiveness of smoking cessation pharmacotherapies, when used with behavioural support in the general population, along with the promising early evidence regarding their effectiveness and safety during pregnancy, makes them an attractive option to improve smoking cessation during pregnancy.

A greater understanding of the factors that influence the prescribing of smoking cessation pharmacotherapies during pregnancy will help with the design of interventions that attempt to increase prescribing of these pharmacotherapies to

pregnant women. As safety concerns are likely to be a key factor limiting the prescribing of these pharmacotherapies to pregnant women, it makes sense to examine whether the magnitude of these safety concerns is warranted. Whether prescribers are familiar with smoking cessation pharmacotherapies is another key factor that could be limiting the prescribing of these pharmacotherapies to pregnant women. Therefore, a study investigating factors that influence this lack of familiarity is required. A focus on facility-level factors is important given evidence that implementing facility-level changes improve smoking cessation care provision in many settings. Given possession of knowledge alone is inadequate to influence prescribing of these pharmacotherapies to pregnant women, it is important to identify other factors that influencing prescribing, including those that operate at the facility-level and individual- beliefs that guide the prescribing behaviour. This research may provide insights into the barriers faced by healthcare providers that are most likely to be effective targets for improving the prescribing of smoking cessation pharmacotherapies to pregnant women.

Smoking and smoking cessation after pregnancy As summarised in the review of the literature on smoking during pregnancy, a substantial proportion of women who smoke early in pregnancy are still smoking at the end of pregnancy. Potential strategies for reducing this proportion were reviewed in section 1.1 (Smoking and smoking cessation during pregnancy). However, none of these strategies will completely overcome the problem of smoking during pregnancy. Recognising this, opportunities for tackling smoking in the period after pregnancy should also be considered.

Exposure to second-hand smoke is recognised as an important yet preventable cause of infant morbidity and mortality [143, 144]. Second-hand smoke includes the smoke from the burning cigarette and smoke exhaled by the smoker [145]. The infant is exposed to the toxic chemicals in second-hand smoke, including nicotine and carbon monoxide. Compared to adults, infants are more susceptible to the harm associated with cigarette smoke as they have higher respiratory rates and relatively immature lungs [146]. These toxic chemicals from second-hand smoke are also more likely to remain longer in infants than adults because their kidneys, liver and other organs are not fully developed to metabolise and able to excrete these chemicals [147].

Exposure to second-hand smoke increases infants’ risk of a sudden unexpected death by 144% (aOR 2.44, 95% CI 2.31-2.57) [148], asthma by 85% (aOR 1.85, 95% CI 1.35-2.53), [149], wheezing 70% (aOR 1.85, 95% CI 1.24–2.35) [149], and hospitalisation due to poor smoking hygiene and fire-related injuries by 28% (aOR 1.28, 95% CI 1.07-1.52) [150]. Infants exposed to second-hand smoke are also more likely to develop potentially life-long negative health consequences such as ischaemic heart disease [151], obesity [152] and long-term neurological morbidity [153] than those who were not exposed. The severity of these illnesses increases with increased exposure [148]. Moreover, these infants are twice as likely (pooled OR 2.19, 95% CI 1.73-2.79) to become smokers themselves, which perpetuates the transgenerational effect of smoking [154].

As for the mother, after giving birth, women who continue to smoke are at increased risk of smoking-related health complications. Table 2 shows the general health conditions that are worsened by smoking cigarettes. A systematic review is currently being undertaken to examine maternal health outcomes during the period following birth that are due to smoking [155]. A woman is also 2.85 times more likely smoke in a subsequent pregnancy if she smokes in her previous pregnancy (aOR 2.85, 95% CI 2.43-3.35) [156], further exposing her subsequent pregnancy and herself to a risk of smoking-related complications.

Table 2 Health conditions worsened by smoking* Diabetes mellitus Hypertension Blood disorders, including anaemia and clotting disorders Respiratory disorders (including acute and chronic inflammatory, and immunological) Gastrointestinal disorders (including peptic ulcer disease) Cataract (including age-related macular degeneration) Osteoporosis (including hip fractures) Infertility Thyroid disorders (hypo- and hyperthyroid) Autoimmune disorders (including rheumatoid arthritis) Acute and chronic kidney disease Cardiovascular disorders (including stroke, peripheral vascular disease, coronary heart disease) *Adapted from the 2014 US Surgeon-General report [144] and Makate et al. [157]

Few studies have directly examined the prevalence of maternal smoking after pregnancy. Nonetheless, indirect evidence can be inferred from the numerous studies showing that the proportion of women who quit smoking in late pregnancy is modest, as discussed in section 1.1.3 (Smoking cessation during pregnancy). It may, therefore, be reasonable to conclude that the majority of women smoking in pregnancy, continue to do so after giving birth. This relies upon the assumption that a woman’s smoking status does not often change from late pregnancy to the period immediately after giving birth. This assumption is supported by population-based evidence that the smoking trajectory after the second trimester of pregnancy is generally stable [42, 158]. A 2019 population-based, surveillance study in the US found that the prevalence of second 26

and third trimester smoking among women who delivered in 2016 was 6.1% and 5.7%, respectively [159].

There are fewer studies that have examined the prevalence of smoking at the time of giving birth. In England, the population-based estimates for maternal smoking at the time of delivery were 15.1% in 2006/2007 and 10.4% in the period April 2018 to March 2019 [160].

The population of women who smoke after pregnancy also include a subpopulation of women who resumed smoking soon after giving birth. Among women who quit smoking during late pregnancy, 43% (95% CI 16-72%) had returned to smoking within 6 months of giving birth [70]. This estimate was generated through a systematic review of trials of smoking cessation interventions among pregnant women [70], whose level of motivation, and likelihood of achieving and maintaining abstinence may differ from the general population [70]. Since this review was published in 2016, two other trials in the US (2016 [161] and 2019 [162]) also reported that almost two-thirds of quitters relapsed into smoking at 6 months postpartum.

1.2.3.1 Health benefits of smoking cessation after pregnancy Smoking cessation should be encouraged as early as possible after delivery because the effects on morbidity and mortality associated with smoking are reversible upon quitting. The younger a woman is when she quits smoking, the greater benefits that she could reap from not smoking, with those who quit by 30 years old avoiding 97% of excess mortality, compared to women who do so by 40 years old (avoidance of 90% of excess mortality) [163]. Although it has not been demonstrated, it also follows that quitting smoking after pregnancy helps to reduce the health risks associated with second-hand smoke exposure to newborns.

1.2.3.2 Barriers to smoking cessation after pregnancy The profile of women who are not successful with quitting smoking during pregnancy, hence likely to be smoking after pregnancy, suggests that they are likely to experience

difficulty with quitting even when not pregnant. A 2018 meta-analysis found that women with the following characteristics are less likely to quit smoking during pregnancy; low level of education, earn low income, have many children, are unmarried, diagnosed with mental health disorders (depression or stress), without a partner, and have a partner who smokes [29].

After giving birth, women who are smoking may encounter additional challenges to quit. These challenges include the typical physiological and emotional changes that occur after giving birth. In addition to recovering physically from childbirth, the women may experience low mood and emotional well-being [164], face the additional stress of managing an infant [165] and attempting to return to pre-pregnancy weight [166], lack of partners’ support for physical and emotional recovery [167]; all of which facilitate them to continue to smoke.

Importantly, even if these women want to quit smoking after giving birth, they may no longer be as motivated to quit compared to when they were pregnant. After giving birth, mothers who smoke may be driven to quit smoking to prevent their newborns from being exposed to second-hand smoke. Qualitative studies indicate that the intention to quit among maternal smokers occurs within the immediate period after giving birth [168] due to concern for their newborn [169]. However, the frequency of women intending to quit smoking dropped as the period of postpartum progressed. A 2017 longitudinal study in the UK found the proportion of women who intended to quit was almost halved at three months postpartum compared to when they were 34-36 weeks pregnant (29.7% (95% CI 23.8-35.6) vs 14.2% (95% CI 10.0-18.3)) [42].

The barriers to quitting smoking experienced by women smoking after giving birth and the lower level of motivation to quit smoking after giving birth, both point to a need to provide smoking cessation support to women who continue smoke after giving birth. A secondary analysis of data obtained from the aforementioned 2017 UK-based study found that this declining motivation to quit was associated with the corresponding reduced interest in engaging with smoking cessation support [55]. Almost half of pregnant women are receptive to smoking cessation support, while only one-third are receptive at three months postpartum (43%, 95% CI 37-49% vs 33%, 95% CI 27-39%) [55]. Thus, it is important to seize the women’s interest, while present, to engage with smoking cessation support in the immediate period after giving birth.

1.2.3.3 Interventions for smoking cessation after pregnancy in healthcare settings Regular encounters between the healthcare system and women in the postnatal period present another window of opportunity for healthcare providers to provide smoking cessation support for those who continue to smoke. In 2018, the WHO’s Maternal Morbidity Working Group recommended that the healthcare system should broaden the scope of intervening in relation to maternal health beyond the 6-week postpartum period [170]. This recommendation suggested that in some settings, there may be further opportunities for healthcare providers to provide smoking cessation support to women who smoke after pregnancy. In Australia, contact with the healthcare system for mothers and their babies is recommended to occur at least twice within the first eight weeks postpartum, thereafter every six months [171]. General practitioners (GPs), obstetricians, nurses, and midwives often provide postnatal care, especially in public healthcare settings. Frequent contact with this key group of postnatal care providers provides opportunity to offer smoking cessation support to women after giving birth. In contrast to the extensive research carried out in pregnant smokers, there are fewer studies [172-178] examining the extent to which smoking cessation support has been offered to women who smoke after giving birth. Importantly, most of these studies focused on postpartum relapse prevention strategies among those who quit during pregnancy [172-178]. This exposes the gap in evidence regarding smoking cessation interventions for postpartum maternal smokers, and points to a need to provide evidence about effective interventions within this population of maternal smokers.

There is inconsistent evidence that psychosocial interventions (comprising health education materials, motivational interviewing and counselling) are effective in promoting cessation during the postpartum period [172-177, 179]. This is likely due to the variation in the type (health education materials, motivational interviewing and counselling) and intensity of interventions evaluated as well as poor adherence to the assessed interventions [172-176]. The lack of information on the provision and/or receipt of psychosocial interventions may also contribute to the inconsistent evidence relating to the effectiveness among women after giving birth. As the effectiveness of smoking cessation pharmacotherapies, which should always be offered in conjunction with behavioural support, is well-established in the general population (detailed in

section 1.1.3.5.1), pharmacotherapy may be a promising option to promote smoking cessation postpartum.

1.2.3.4 Prevalence of pharmacotherapy use for smoking cessation after pregnancy Potential benefits of smoking cessation pharmacotherapies for women smoking in the postpartum period can only be realised if they are taken up by this population of maternal smokers. Two prior studies have addressed this question [104, 107].

In the US, a state-based population study in Maryland using pharmacy claims data from 2003 to 2010 found that 2.0% of 4,709 pregnant smokers filled a prescription for NRT within one year of giving birth [104]. The after-pregnancy use estimate was lower than the during-pregnancy use estimate (2.0% as opposed to 2.4%). In this setting, access to medications was thought to be a barrier to use of smoking cessation pharmacotherapies after pregnancy [104]. In Maryland during the study period, insurance coverage providing access to postnatal health services utilisation was limited to the first 60 days after giving birth, and this may lead to low use of subsidised medications [104]. Reduced motivation to quit smoking after pregnancy may have also contributed to lower use of smoking cessation medications. The authors found that only five women filled a prescription for varenicline in the year after delivery, although this low prevalence of use might be due to the study period, with varenicline introduced in 2008 [104]. In the UK, a population-based study that used general practice prescribing data found that about 5% of maternal smokers were prescribed NRT within nine months of delivery between 2001 and 2012 [107]. Consistent with the findings from the US study [104], the authors also found that after-pregnancy use was lower than during-pregnancy use (5.5% vs 11.0%), despite there being no difference in the subsidised access to NRT during pregnancy and after delivery.

The extent of smoking cessation pharmacotherapy use among women who continue to smoke after giving birth in Australia has not been measured. Given the extent of use varies between countries, estimates that are specific to Australia are needed. Within this investigation of utilisation of smoking cessation pharmacotherapies, it is helpful to examine the timing of use of these pharmacotherapies. Use of smoking cessation

pharmacotherapies soon after giving birth is advisable to minimise the maternal and child health risks associated with continued smoke exposure.

Information on the prevalence of smoking cessation pharmacotherapy use among women smoking postpartum will inform whether efforts to encourage and support modified uptake of pharmacologic therapy among women at this important life stage are necessary. In order to assist with assessment, a greater understanding of the factors that are, or ought to be, considered in the decision to use smoking cessation pharmacotherapies among women smoking after pregnancy, is required.

1.2.3.5 Factors influencing the use of pharmacotherapy for smoking cessation after pregnancy As explained in section 1.1.3.5.3 (Factors influencing prescribing of smoking cessation pharmacotherapy ) relating to pregnancy, factors including efficacy/effectiveness, safety, suitability, and cost of a medication systematically influence the decision to prescribe it [119]. Individual-, and system-level prescriber-related factors may also play a role; these are reviewed in greater detail in Chapter 3. As use of prescription medications is contingent on receiving a prescription, these same factors also influence whether patients use a medication. This thesis focuses on the effectiveness, and suitability of smoking cessation pharmacotherapies as factors that have the potential to influence use among women who smoke after pregnancy.

Although safety is usually the highest order consideration among the aforementioned factors, this thesis does not examine whether safety considerations drive the use of smoking cessation pharmacotherapies in women after giving birth. Using smoking cessation pharmacotherapy after giving birth may pose risk to breastfeeding infants. Only NRT is recommended for use during breastfeeding [180] as pharmacokinetic studies for bupropion and varenicline during lactation [172, 173] are lacking. Maternal use of NRT poses a low risk of nicotine exposure to the breastfeeding infant as the amount of nicotine absorbed by the breastfeeding infant is relatively safer than the exposed nicotine absorbed by inhaling the cigarette smoke from the mother [181]. The nicotine absorbed by the infant depends on the nicotine level in the milk and the amount of milk consumed. A nursing infant is exposed to low level of nicotine due to low bioavailability for oral nicotine (30-40% in adults, with no known data on infants)

[182]. This is reinforced by the low milk consumption during nursing due to low production of milk by a breastfeeding smoker [183]. Spaced smoking and/or NRT use while breastfeeding may reduce the amount of nicotine excreted via breast milk [184].

Despite there being a potential for risk when these pharmacotherapies are used among women who breastfeed, this is unlikely to account for much of the non-use of smoking cessation pharmacotherapies within one year of giving birth. In high-income countries, 45-69% of smokers do not breastfeed [185-187], and the average duration of breastfeeding among maternal smokers is 10-11 weeks [186], as opposed to the WHO recommended duration of six months [188]. Even then, breastfeeding infants consume a relatively low amount of milk due to the low production of milk by women who smoke [183]. Given the safety concern is relevant only to the population of maternal smokers who breastfeed their infants, other factors may play a larger role in influencing the use of pharmacotherapies for smoking cessation in women after giving birth.

Efficacy and effectiveness Whether a medication is effective in the population of interest ought to influence its use in the said population. To date, there is no study that evaluates the effectiveness of pharmacotherapy for smoking cessation among women smoking after pregnancy. Evidence on the ability of smoking cessation pharmacotherapies to increase smoking cessation in women after giving birth will be useful to guide smoking cessation treatment decisions for postpartum maternal smokers.

While smoking cessation pharmacotherapies are known to be effective in the non- pregnant population, their effectiveness may differ during this particular stage of life for two reasons. Similar to the reasons outlined in the section relating to pregnancy (section 1.1.3.5.1), women in postpartum period differ from female smokers in the non- pregnant state in regard to their physiology and behaviour.

For the first six weeks following childbirth, women undergo physiological changes, in which adjustments that occurred during pregnancy return to the non-pregnant state [189]. These altered physiologic characteristics may affect the pharmacokinetic properties of smoking cessation pharmacotherapies. Second, after childbirth, the associated changes in mood [190], weight [166] and stress [191], such as the

additional stress of caring for an infant [52], may affect the motivation of women to quit smoking. A 2017 study that found fewer women intended to quit at three months postpartum compared to when they were 34-36 weeks pregnancy [42].

Suitability Deciding whether to use a pharmacotherapy, and which one, is greatly influenced by whether the pharmacotherapy is suitable for an individual patient [119, 192]. WHO considers a medication as suitable based on an overall assessment of whether it is effective, safe and well-tolerated for a particular patient [119]. Whether a medication is well-tolerated by a patient is reflected by its ease of use (dosage form, dosing regimen, and duration of therapy) and its adverse effects profile (contraindications, side-effects, medication interactions, and co-existing medical conditions) [119].

While it is unlikely that the ease of using these pharmacotherapies differs substantially between individuals, the differing health profile between women at this life stage necessitates an individual assessment of the risk-benefit-risk balance.

As detailed in the following section (Section 1.3.3, in particular), use of smoking cessation pharmacotherapies is cautioned or contraindicated among individuals with specific pre-existing medical conditions. For example, varenicline and bupropion have been linked to an increased risk of neuropsychiatric adverse events, especially the tendency to worsen the course of the disease or to increase the severity of the symptoms, among individuals with a prior history of psychiatric illness [193, 194]. It is, therefore, advisable that both varenicline and bupropion are avoided among women with a history of, or current, psychiatric disorders [195]. This same reasoning applies to a number of other morbidities or concomitant use of other medications whereby under these circumstances, smoking cessation pharmacotherapies are cautioned or contraindicated.

In contrast, the suitability of smoking cessation pharmacotherapies is increased among women with health conditions worsened by smoking (outlined in Table 2), given the benefits of smoking cessation are amplified under these circumstances. For a similar reason, smoking cessation pharmacotherapies may also be more suitable for women who experience poor birth outcomes including adverse maternal and/or neonatal

events which place the mother and/or newborn at greater risk of harm from smoke exposure (as described in section 1.2.1).

An understanding of whether smoking cessation pharmacotherapies are used in postpartum smokers in whom a given pharmacotherapy is suitable and avoided by those for whom these pharmacotherapies are unsuitable, is required. This will reveal whether efforts to modify use of smoking cessation pharmacotherapies among women smoking after pregnancy are needed.

In summary, women who are not successful in quitting smoking while pregnant are likely also to experience difficulty in quitting after giving birth. Moreover, a high proportion of women who quit smoking during pregnancy resume smoking after giving birth. The postpartum period offers another opportunity for postnatal care providers to provide smoking cessation support to these women during the postpartum period. Very little research has examined which smoking cessation intervention is effective to treat postpartum maternal smoking. Studies on psychosocial interventions have returned inconsistent findings regarding their ability to increase smoking cessation rates. As pharmacological interventions, when provided concurrently with behavioural support, are the most effective smoking cessation intervention for the general population, their use and potential to benefit women smoking postpartum is worthy of investigation.

Information on the use of these pharmacotherapies will be the first step to inform whether efforts are needed and to what degree, to modify the uptake of smoking cessation pharmacotherapy in this postpartum population. To aid in addressing this question, a greater understanding of the factors that are, or ought to be, considered in the decision to use smoking cessation pharmacotherapies among women smoking after pregnancy, is required.

There are no available data on the effectiveness of smoking cessation pharmacotherapies in this population of postpartum smokers. Information on whether these pharmacotherapies are effective will help inform future decision about the extent to which these pharmacotherapies should be used by postpartum maternal smokers. Information on whether maternal morbidities and birth outcomes play a role in the

decision to use these pharmacotherapies will indicate how the uptake of these pharmacotherapies among postpartum smokers should be modified.

Overview of smoking cessation pharmacotherapy Having outlined the critical need to address smoking during and after pregnancy and the rationale for focusing on smoking cessation pharmacotherapies, this section provides background information on the general characteristics of each pharmacotherapy. These include the clinical pharmacology, efficacy in the general population, safety and tolerability, and cost of each smoking cessation pharmacotherapy.

NRT consists of purified nicotine (medicinal nicotine) that is administered to reduce the physical dependence on nicotine from cigarettes. The extent of nicotine being delivered to the body depends on the types of nicotine replacement product and the dose. Nicotine patches deliver nicotine slowly, thereby ameliorating nicotine withdrawal symptoms. Nicotine gums, lozenges, sprays and inhalers deliver nicotine more rapidly than the patches, thereby providing much of the positive effects of nicotine [73].

Bupropion hydrochloride was initially marketed as an antidepressant due to its stimulant properties. The exact mechanism of action for bupropion hydrochloride is unclear, but it acts as dopamine antagonist in the brain [196]. By blocking the neuronal reuptake of dopamine and norepinephrine, bupropion hydrochloride increases the levels of dopamine and norepinephrine, thus mimicking the rewarding effects of nicotine on these neurotransmitters. It also has a limited nicotine-receptor-antagonist action [74]. As a smoking cessation pharmacotherapy, bupropion hydrochloride is marketed as a sustained-release product. Bupropion-assisted cessation is associated with fewer depression-related symptoms [196] than during pre-treatment and weight loss [197].

Varenicline tartrate is a partial agonist of the α4β2 receptor. This α4β2 receptor is the primary receptor facilitating nicotine addiction. Unlike nicotine, a full agonist, varenicline’s partial agonist effect maintains moderate levels of dopamine to counteract withdrawal symptoms [75]. Varenicline tartrate also acts as an antagonist by blocking any further nicotine binding, which may reduce the rewarding effects of smoking and reduce reactivity to smoking cues [75].

Table 3 provides an overview of three smoking cessation pharmacotherapies including the formulations and strengths available in Australia, advantages, side-effects, contraindications, precautions, drug interactions, and costs.

Table 3 Pharmacotherapeutic properties of smoking cessation pharmacotherapies Pharmacotherapies for smoking cessation Details Nicotine Replacement Therapy, NRT [198] Bupropion [199] Varenicline [75] Prescription NRT patches was the only formulation that Bupropion is available as an authority Varenicline is available as an authority status requires prescription at the time of study. required prescription*. required prescription*. An additional Both patches and other NRT formulations are 12-week course is offered to those available over-the-counter (OTC). who managed to quit and require further smoking cessation support. Strengths and Gum 2mg, 4mg Sustained release tablet 150mg Tablet 0.5mg, 1mg formulations Patch* (16hours and 24 hours) 5mg-,10 mg-, 15mg-, 25mg/16 hours Starter pack Starter pack 7mg-, 14mg-, 21mg/24 hours 150mg (30 tablets) 0.5mg (11 tablets) and 1mg (42 Lozenges 1.5mg, 2mg, 4mg tablets) Inhaler 2mg Maintenance pack Oral strips 2.5mg 150mg (90 tablets) Maintenance pack Spray 1mg/dose 1mg (56 tablets) Advantages Reduces nicotine cravings and withdrawals Twice-daily dosing regimen facilitates Twice-daily dosing regimen facilitates by providing the therapeutic nicotine adherence adherence Can be titrated to manage withdrawal Delays weight gain associated with smoking Most effective pharmacotherapy symptoms and situational urges cessation Can be used in combination with other Can be used in combination with NRT medications Combined forms of NRT (dual therapy) is comparable in efficacy to varenicline Side effects General Dry mouth Nausea Heart palpitations, chest pains Insomnia Vomiting Nausea and vomiting Headache Chest pain, shortness of breath Gastrointestinal complaints Weight loss (symptoms of cardiovascular events) Insomnia Constipation Insomnia Specific to Anxiety/difficulty to concentrate Gums/lozenges/spray: Mouth and throat Tremor soreness, mouth ulcers, hiccups, coughing Rash Patches: Skin irritations

Pharmacotherapies for smoking cessation Details Nicotine Replacement Therapy, NRT [198] Bupropion [199] Varenicline [75] Contraindications Hypersensitivity to nicotine or other Seizure disorders Hypersensitivity to varenicline components of NRT Anorexia / bulimia disorders Severe hepatic necrosis Hypersensitivity to bupropion

Precautions Recent (less than 2 weeks) myocardial Hepatic impairment Severe renal impairment infarction Boxed warning removed in December 2016 Seizure Serious underlying arrhythmias regarding neuropsychiatric symptoms Boxed warning removed in December Serious or worsening angina pectoris including changes in behaviour, agitation, 2016 regarding neuropsychiatric Active peptic ulcer disease depressed mood, suicidal ideation, and symptoms including changes in Temporomandibular disease attempted and completed suicides [195] behaviour, agitation, depressed mood, suicidal ideation, and attempted and completed suicides [195] Interactions with No clinically meaningful interactions with Drugs that lower seizure threshold No clinically meaningful interactions other medications other medications reported (Monoamine oxidase inhibitor therapy in the with other medications reported previous 14 days, antiepileptics, theophylline, antipsychotics, antidepressants, and systemic corticosteroids) Drugs that inhibit/induce cytochrome P450 (CYP)2D6

Cost (in Australian 8-week supply (NRT patches) when 7-week supply when purchased under 12-week supply when purchased dollars) purchased under government subsidy (in government subsidy (in 2019) under government subsidy (in 2019) 2019) Concessional beneficiary: $13.00 Concessional beneficiary: $19.50 Concessional beneficiary: $13.00 General beneficiary: $82.00 General beneficiary: $123.00 General beneficiary: $82.00

For OTC products, costs vary between formulations and brands, averaging to $130- 270 /8-week supply. *An authority required prescription refers to a prescription that must be annotated with a valid authority approval number by approved prescriber [200]. 39

As explained in section1.1.3.5.1(Efficacy and effectiveness of smoking cessation pharmacotherapies), a 2018 meta-analysis of 136 trials found that any formulation of NRT increases the cessation rate by 55% (pooled RR 1.55, 95% CI 1.49-1.61) compared to placebo or no treatment [78]. There is high-quality evidence that compared to placebo, bupropion and varenicline increase the cessation rate by 62% (45 trials; pooled RR 1.62, 95% CI 1.49-1.76) [76] and 124% (27 trials; pooled RR 2.24, 95% CI 2.06-2.43) [77], respectively.This makes them the most efficacious smoking cessation interventions available, with the next largest effect size observed for advice by a physician (pooled RR 1.76, 95% CI 1.58-1.96) [89].

Varenicline is more efficacious than bupropion (pooled OR 1.59, 95% CI 1.29-1.96) and all single forms of NRT (patch (pooled OR 1.51, 95% CI 1.22-1.80), gum (pooled OR 1.72, 95% CI 1.38-2.13), and others including inhaler, spray, tablets, lozenges (pooled OR 1.42, 95% CI 1.12-1.79)) [201]. However, the efficacy of varenicline is comparable to NRT products used in combination (patch and non-patch form; pooled OR 1.06, 95% CI 0.75-1.48). Bupropion and NRT are comparable to each other in their efficacy (pooled OR 0.99, 95% CI 0.86-1.13) [201]. A 2019 meta-analysis of 63 trials indicates that different forms of NRT are associated with different levels of effectiveness with combined NRT (patch and fast-acting form) more effective than single form (pooled OR 1.25, 95% CI 1.15-1.36) [198].

Generally, NRT patches are perceived as relatively safe and are well-tolerated. The common side-effects for NRT patches are insomnia, nightmares and local skin irritation [202]. NRT is also associated with a two-fold increased risk of heart palpitations and chest pains, which may aggravate cardiovascular events [202].

Varenicline is strongly associated with nausea [75], which may not be well-tolerated during pregnancy and in physical recovery from childbirth [189], as well as insomnia and abnormal dreams [75]. An observed increased risk of neuropsychiatric events such as suicidal thoughts, depression, anxiety, aggressive reaction, agitation, emotional lability, and irritability [193, 194], led the US regulatory body, the Food and Drug Administration (FDA) to implement a black box warning for bupropion and varenicline in

2008, although this was removed in November 2016 [195]. This warning was lifted, following findings from a large clinical trial conducted across 16 countries between 2011 and 2015 that did not reveal evidence of increased risk of neuropsychiatric symptoms for varenicline and bupropion [203]. However, the FDA maintains that the increased risk remains in the population with a history of psychiatric illness [195].

The FDA and Australia’s Therapeutic Goods Administration (TGA) also placed a warning in the medication prescribing information relating to an increased risk of cardiovascular events associated with varenicline [204, 205]. This arose due to the findings of a 2011 meta-analysis that reported an increased risk of cardiovascular events among varenicline users, although this finding was refuted by the findings from a 2012 [206] and 2015 [207] meta-analyses. Concerns regarding an increased risk of cardiovascular events associated with varenicline use remain, however, due to varenicline’s action as partial agonist at the α4β2 receptor [75].

Similar to varenicline, there was an observed increased risk of neuropsychiatric events such as anxiety, agitation and jitteriness among individuals who are exposed to bupropion [199], which prompted the FDA to implement a black box warning regarding an increased risk of developing neuropsychiatric events in bupropion users; though this was later removed in 2016 [195]. A critical concern for bupropion is that it lowers the seizure threshold; thereby increasing the risk of developing seizures [199]. Thus, bupropion is contraindicated in smokers who have seizures or in smokers have conditions predisposing them to have a low seizure threshold (see Table 3) [199]. Moreover, due to bupropion being metabolised by cytochrome P450 (CYP)2D6, it interacts with many drugs by affecting the clinical effects of this enzyme’s substrates [199]. As bupropion increases the risk of developing severe adverse drug reactions and is subject to substantial interactions with other medications [208] this may result in bupropion being considered as an alternative pharmacotherapy for smoking cessation after varenicline and NRT in the general population.

In Australia, all three pharmacotherapies are subsidised under the Pharmaceutical Benefits Scheme (PBS) as a tobacco cessation aid. The PBS allows all Australian citizens, permanent residents and eligible foreign visitors residents access to subsidised prescription medications. This subsidy means that the patient only contributes a co-payment for each dispensing. The co-payment threshold is two-tiered, whereby concessional beneficiaries pay a lower co-payment towards the cost of their medication than general beneficiaries [72]. At the time of the research conducted in this thesis, only NRT patches were subsidised. Both patches and other NRT formulations are available over-the-counter (OTC), including NRT gums, inhalers, lozenges and mouth spray.

Summary of existing literature and evidence gaps Smoking during pregnancy is the leading cause of preventable pregnancy complications, and poor health outcomes for women and their newborns. Despite the overall decline in the prevalence of smoking during pregnancy in high-income countries, the prevalence of smoking remains high, with 10% of pregnant women in Australia smoking while pregnant. Although pregnancy appears to elicit a high motivation level to quit smoking, the prevalence of quitting remains modest. This implies that a substantial population of women who smoke during pregnancy need evidence-based smoking cessation support.

Although psychosocial interventions are the recommended smoking cessation intervention for pregnant women, it requires additional resources during routine antenatal care such as time and skills to deliver a tailored psychosocial intervention. Evidence is unclear as to which strategy or a combination of these strategies is consistent in increasing smoking cessation during pregnancy and maintaining it beyond pregnancy. While such studies are forthcoming, pharmacological interventions, established as the most effective interventions in the general population, hold promise for increasing the number of women who quit smoking during pregnancy.

There is much variation in the estimates of during-pregnancy use of smoking cessation pharmacotherapy over time, and between patient populations and countries. This variation suggests that the extent to which these pharmacotherapies are used during pregnancy is dependent on the context, and may, therefore, be amenable to change. Identifying factors influencing their use may inform approaches to increase the uptake of smoking cessation pharmacotherapy, especially in Australia, where the utilisation of these treatments during pregnancy is considerably lower than other high-income countries.

Although the majority of healthcare providers screen and advise pregnant smokers to quit smoking, few prescribe pharmacotherapies for smoking cessation to pregnant women. Healthcare providers, therefore, may be a key group to target when attempting to improve the use of smoking cessation pharmacotherapies in pregnancy. This broad range of healthcare providers includes obstetricians, GPs, midwives, practice nurses, pharmacists, maternal and child health nurses, all of whom provide maternity care.

Healthcare providers’ concern about whether these pharmacotherapies are safe during pregnancy is likely to be one of the important factors limiting the prescribing, and use of these pharmacotherapies. The magnitude of these concerns may not be warranted, as smoking itself poses a substantial risk in pregnancy. Research examining whether healthcare providers are overly concerned about the risk associated with the use of smoking cessation pharmacotherapies during pregnancy is required.

In addition to healthcare providers' concerns regarding the safety of these pharmacotherapies during pregnancy, other factors potentially limit the prescribing of these pharmacotherapies to pregnant women. In particular, identifying whether there are facility-level factors associated with lack of knowledge about these pharmacotherapies among obstetric care providers may help with the design of interventions that attempt to increase prescribing of these pharmacotherapies to pregnant women. For the same reason, (facility-level factors and individual-beliefs) associated with the intention to prescribe smoking cessation pharmacotherapies among obstetric care providers, should also be the focus of research.

Given the low smoking cessation rates among women smoking in pregnancy, it is likely that there is a substantial proportion of women who continue to smoke after giving birth. Compared to women who did not smoke, women who continue to smoke after pregnancy are more likely to expose their children to second-hand smoke. Second- hand smoke is an important cause of poor infant health outcomes. Therefore, women smoking after pregnancy is another key population to target with smoking cessation support.

There is little evidence to guide the provision of smoking cessation support among women who continue to smoke after delivery. However, smoking cessation pharmacotherapies have been proven effective and well-tolerated in the general population. This suggests that the potential of these pharmacotherapies should be explored among women smoking after pregnancy. First, it is important to establish the extent of smoking cessation pharmacotherapy use among women who smoke after pregnancy in Australia, including whether the opportunity to quit early during the postpartum period is seized. This information can then inform whether efforts to modify the use among women in this important life stage are required. Addressing this question requires a greater understanding of the factors that are, or ought to be, taken

into account when making the decision to use smoking cessation pharmacotherapy during the postpartum period. This includes an investigation of whether suitability (pre- existing health conditions, birth outcomes) is considered in the decision to use smoking cessation pharmacotherapy after giving birth. It is also important to obtain evidence on whether these pharmacotherapies are effective in this postpartum population, as this will indicate the extent to which smoking cessation pharmacotherapies should be used among women who smoke after pregnancy.

Thesis aims and overview The overarching premise underpinning the research in this thesis is that, the uptake of smoking cessation pharmacotherapies during pregnancy can be better supported by healthcare providers caring for maternal smokers and that the post-natal period is a further opportunity to encourage smoking cessation through pharmacotherapies. This thesis addresses the evidence gaps that impede the realisation of this over-arching premise through four original research studies.

The first two studies examine the factors that influence the prescribing of smoking cessation pharmacotherapies to pregnant women. Chapter 2 describes a study that investigates whether healthcare providers are overly concerned regarding the safety of these pharmacotherapies during pregnancy. The study in Chapter 3 examines the relationship between facility-level factors and familiarity with these pharmacotherapies among obstetricians and gynaecologists. Among obstetricians who are familiar with these pharmacotherapies, this study also examines whether the same set of facility- level factors and individual-level factors are associated with their intention to prescribe these pharmacotherapies to pregnant women.

The next two studies address use of smoking cessation pharmacotherapy among women smoking after pregnancy and attempt to improve understanding of the factors influencing that are, or ought to be, considered when deciding to use these pharmacotherapies in this population of postpartum smokers. The study in Chapter 4 aims to measure the extent of, and timing of, use of these pharmacotherapies in 12 months postpartum among women smoking at delivery. This study also examines whether the suitability of these pharmacotherapies for women with different health characteristics, specifically relating to maternal pre-existing health conditions and poor birth outcomes, are being considered in use of these pharmacotherapies after giving birth. Chapter 5 describes a study that measures the effectiveness of these pharmacotherapies when used during inter-pregnancy interval.

Chapter 6 concludes the thesis by summarising and interpreting the key findings from each study, whilst also considering the limitations of the research. This chapter also discusses the potential practice implications of the findings. This chapter recommends directions for future research to build on the work of this thesis. Overall, this thesis aims to add to the evidence base in promoting greater use of pharmacotherapies for smoking cessation during and after pregnancy. 46

More specifically, this thesis set out to address seven specific aims, as follows: 1. To measure the magnitude of concern from healthcare providers regarding smoking cessation pharmacotherapies for pregnant women, relative to other medications in the same and other risk categories (Chapter 2) 2. To examine facility-level factors associated with obstetricians’ and gynaecologists’ familiarity with smoking cessation pharmacotherapies (Chapter 3) 3. To identify facility-level factors and individual beliefs associated with obstetricians’ and gynaecologists’ intention to prescribe smoking cessation pharmacotherapies to pregnant women (Chapter 3). 4. To estimate the prevalence of utilisation of each smoking cessation pharmacotherapy in the year following delivery among women who smoked, and the timing of the first use (Chapter 4). 5. To examine the association between selected maternal morbidities and the use of each smoking cessation pharmacotherapy in the year following delivery among women who smoked (Chapter 4). 6. To investigate whether poor birth outcomes are associated with use of each smoking cessation pharmacotherapy in the year following delivery among women who smoked (Chapter 4). 7. To assess whether exposure to smoking cessation pharmacotherapies during the inter-pregnancy interval is associated with smoking cessation prior to the subsequent pregnancy (Chapter 5).

Chapter 2 Healthcare providers’ concern regarding smoking cessation pharmacotherapies during pregnancy: Calls to a teratology information service

Preface to the chapter As outlined in Chapter 1, the lack of clear evidence on the safety of smoking cessation pharmacotherapies during pregnancy may prompt concerns from healthcare providers about whether these pharmacotherapies are safe in pregnant women. These safety concerns may result in reluctance among healthcare providers to prescribe these pharmacotherapies to pregnant women.

This chapter aims to investigate whether healthcare providers are overly concerned about the safety of pharmacotherapies for smoking cessation during pregnancy compared to medications in the same and other pregnancy-risk categories.

This chapter has been published in a peer-reviewed journal, Drug and Alcohol Review [209] (see Appendix 2). The citation of the publication is:

Lee ML, Tran DT, Welsh A, Kennedy D, Havard A. Health-care providers' concern regarding smoking cessation pharmacotherapies during pregnancy: Calls to a teratology information service. Drug and Alcohol Review. 2020;39:223-31.

What follows is the Microsoft Word version of the published manuscript, with some minor revisions arising through the thesis examination process.

Published manuscript in Microsoft Word format

Introduction Smoking is one of the leading yet preventable causes of poor maternal and foetal outcomes. Maternal smoking during pregnancy adversely affects maternal and foetal health, including increased risk of ectopic pregnancy, placenta praevia, spontaneous abortion, low birth weight, preterm birth, intrauterine growth retardation, perinatal death and sudden infant death syndrome [210]. Quitting, even if occurring after the first trimester, is associated with favourable health outcomes relative to continued smoking [211]. Among 10-35% of women who smoked during pregnancy in high-income countries [25, 26], less than half spontaneously quit during pregnancy [48]. This implies that a large proportion of pregnant smokers may need assistance to quit.

Smoking cessation pharmacotherapies including nicotine replacement therapy (NRT), varenicline and bupropion significantly increase the likelihood of quitting by an estimated 82-188% in non-pregnant smokers and therefore are the recommended first- line treatments in the general population of smokers attempting to quit [201]. Although the effectiveness of smoking cessation pharmacotherapies has not been demonstrated in pregnancy, this evidence from the non-pregnant population suggests they could be a promising option for women smoking during pregnancy. This is particularly the case for women who are unable to quit without pharmacological assistance, noting that psychosocial interventions are modestly effective during pregnancy [63].

Of the three prescription pharmacotherapies available in most developed countries, NRT has the most support for its use during pregnancy. This is on the basis that it cannot be more harmful than continued smoking during pregnancy. Antenatal care guidelines in many high-income countries [108, 130] recommend NRT use during pregnancy when the likely benefits from NRT use outweigh the risk of harm. Indeed, in the United Kingdom (UK), NRT is routinely offered at specialised cessation services for pregnant women [212].

Despite widespread support for NRT use during pregnancy, population-based studies find a low prevalence of use of NRT among pregnant smokers in Australia (2.3- 3.6%)[110] and the United States (US) (1.2%)[104]. This is in comparison to the prevalence of during-pregnancy use in the UK (11%) [107] and Canada (24.5%) [109]. Up to 40% of pregnant smokers in Australia and the US are moderate and heavy smokers [213, 214] of which their high nicotine dependence would be associated with greater nicotine withdrawal in their attempt to quit. Therefore, these smokers would benefit from using NRT. Evidence regarding attitudes and behaviours of healthcare providers suggest that reluctance from healthcare providers could contribute to the low use of NRT during pregnancy. Only 2% [215-217] of surveyed providers prescribed NRT to their smoking pregnant patients in the US [216] compared to 27% in the UK [217]. Despite their pregnant patients requesting NRT, only 44% of obstetricians participating in a state-wide survey in the US would prescribe it [218]. This reluctance appears to be due, at least in part, to concerns about NRT safety during pregnancy. A population-based survey in Australia found that 56% of obstetricians perceived NRT to be unsafe during pregnancy [219]. In fact, 92% of another different state-wide survey of obstetricians in the US reported they would only prescribe NRT if more safety-related data were available [220]. Providers perceive medications, in general, to have a higher safety risk during pregnancy than their recognized risk [221, 222]. Whether there is a disparity in the magnitude of concern regarding NRT during pregnancy, and other medications of equivalent risk, has not been directly examined, hence is the subject of this study.

The other first-line cessation pharmacotherapies, bupropion and varenicline, are not recommended for use during pregnancy [71]. However, evidence that bupropion [109] and varenicline [223] use during pregnancy are not associated with an increased risk of adverse birth outcomes is emerging [127]. Given that recommendations for use of these pharmacotherapies during pregnancy may be revised in the future, this study also sought to measure the extent of concern from healthcare providers regarding bupropion and varenicline relative to other medications.

The aim of this study was to measure the magnitude of concern from healthcare providers regarding smoking cessation pharmacotherapies relative to other medications in the same and other risk categories.

Methods

Study design and data source This was a retrospective, cross-sectional study of telephone calls made to MotherSafe between 1 January 2001 and 31 December 2016. The study was based on unit record data, with each record representing a call to MotherSafe. As a Teratology Information Service (TIS), MotherSafe is a state-wide, free telephone counselling service. Established in New South Wales (NSW), the MotherSafe call centre is located in an advantaged and metropolitan area. MotherSafe is managed by trained providers who counsel callers concerned about potential exposures during pregnancy and breastfeeding. Details of MotherSafe services are described elsewhere [224].

A call to MotherSafe resulted in a database entry including pregnancy status, exposure of concern and type of caller. Due to privacy reasons, the call data released for this research did not include any identifying information on the caller. Only pregnancy- related calls were included in the analyses, defined as a call made regarding an exposure during pregnancy.

Measures This study used calls to a TIS to measure the extent of concern. TIS calls were taken as an indicator of concern regarding the medication discussed during the call on the basis that pregnancy-related safety concerns are nominated as the main reason for calling in 80-95% of TIS calls [225, 226].

Medication exposure An exposure to a medication is defined as contact, ingestion, inhalation or injection of a prescribed or over-the-counter (OTC) medication [120]. This study was based on the categorisation of exposures according to the Therapeutic Goods Administration (TGA)’s classification of risk during pregnancy [120]. The TGA classifies medications based on the risk to the foetus associated with exposure during gestation. There are seven risk categories (category A, B1, B2, B3, C, D and X). Medications categorised as Category A do not have evidence of direct or indirect foetal risk whereas those classified as Category X are of high risk to cause permanent foetal harm and thus are

contraindicated for use during pregnancy [120]. Medications classified as Category B have inadequate human data on foetal risk, with subgroups (B1, B2, and B3) reflecting whether evidence of harm from animal studies exists. Category B1 denotes no observed increase in risk of harm in animal data, and category B2 medications have no observed increase in foetal risk due to inadequate animal data, whereas for medications in Category B3 animal studies have returned increased risk of harm [120]. Category C medications pose foetal harm, without causing malformations, according to human data [120]. The pregnancy risk category for bupropion is B1, and category B3 for varenicline. Due to the presence of nicotine in NRT, NRT is classified as a category D medication as nicotine exposure is known to cause foetal harm [120]. Calls were categorised as relating to NRT, bupropion or varenicline, or one of the seven pregnancy risk-categories (category A, B1, B2, B3, C, D, X) [120]. A call could contain up to five exposures. Calls were categorised based on the first medication listed in each call record because it was likely to be the caller’s primary reason for calling, that is, the caller’s concern regarding that medication was strong enough to prompt a call. Whether additional medications mentioned were of equal concern to callers could not be assumed.

Caller type This study examined the types of callers as the primary outcome. Based on the information in the caller field, callers were categorised as healthcare providers and non-providers.

This study defined a healthcare provider as an individual who provides health care services in a legally recognised organisation [227]. The health services included the traditional and complementary services which could take place in the home and community-based settings as general and private practices, community health, local and non-government services settings such as Aboriginal Community Controlled Health Services. The traditional health service providers comprised individuals who work in private hospitals, day surgeries, medical practitioners, pharmacists and allied health professionals. The complementary health service providers comprised individuals such as naturopaths and chiropractors. In addition to the care settings, a healthcare service could also address specific health and lifestyle conditions such as mental health, sexual health, oral health and drugs and alcohol services.

Calls from healthcare providers consisted of four categories; general practitioners (GP), midwives, obstetricians, and pharmacists. Calls made by GP obstetricians, GP registrars, and GP receptionists were grouped as calls from GPs. Calls from midwives included those from midwifery students, antenatal clinics and community midwives. Calls made by obstetricians included any call from obstetricians (including general obstetricians, obstetric registrars as well as maternal-foetal medicine subspecialists), gynaecologists (including general gynaecologists and subspecialties, such as reproductive endocrinology infertility specialty, In Vitro Fertilisation, fertility clinic and reproductive medicine) and geneticists. Pharmacist calls included calls made by the pharmacy assistant and pharmacy. Together, these four broad categories encapsulated all included calls from healthcare providers.

After exclusion of other categories of healthcare provider (see Exclusions below for details), the remaining calls were grouped as non-provider calls, comprising calls from women, their partners, their family members, their friends, and other categories such as journalists, chief medical officers, pharmaceutical companies as well as translational and interpreting services interpreters.

Potential confounders The postcode nominated by the caller was mapped to an area-based measure of: (a) socioeconomic status (SES), specifically the Socioeconomic Indexes for Areas Index of Relative Socioeconomic Disadvantage (IRSD) scores [228] and (b) remoteness of residence with classification according to the Accessibility Remoteness Indicator of Australia (ARIA+) [229]. The IRSD scores were grouped as disadvantaged SES (deciles 1–3), average SES (deciles 4–7) and advantaged SES (deciles 8–10). The ARIA+ scores were grouped as remote (remote and very remote), regional (outer and inner regional) and major cities.

Exclusions This study excluded calls made relating to planning a pregnancy, general inquiry, breastfeeding and/or pregnancy as well as retrospective calls. A retrospective call was defined as a call made regarding a previous exposure during pregnancy or

breastfeeding which resulted in an adverse outcome. This study also excluded calls regarding exposure to infections, radiation, chemicals, substances used recreationally and food. Calls regarding medications with multiple therapeutic components with different risk categories and exempted medications were excluded. TGA-exempted medications lack risk classifications; they comprise products such as vitamins, herbal remedies, mineral supplements and diagnostics agents . In terms of caller types, calls from other categories of healthcare providers were excluded because they were less likely to primarily influence the decision to prescribe or use pharmacotherapy during pregnancy. These excluded calls were those made by medical students / trainees, lactation consultants, mental health team members, providers associated with Tresillian early parenting services and Chemical Use in Pregnancy Services, drugs and alcohol agencies, Aboriginal health counsellors, Australian Breastfeeding Association and Family Planning Association counsellors, antenatal care educators, as well as naturopath, radiologist, physiotherapist, dentist, social worker, genetics counsellors and counsellors. As for other categories of medical specialists such as oncologists, rheumatologists, paediatricians, cardiologists, and others, they were excluded from the healthcare provider caller group because their calls were more likely to be regarding medications specific to their specialty. This could potentially overestimate the extent of concern for those medications compared to cessation pharmacotherapies.

A total of 66,687 calls were analysed (see Figure 2 for the selection of study population). Given that varenicline was introduced in 2008 in Australia, all analyses pertaining to varenicline were restricted to calls made between 2008 and 2016 (50,396 calls).

Ethics approval This study was approved by the South East Sydney Illawarra Area Health Service Human Research Ethics Committee, reference number 10/197.

Statistical analyses For each pharmacotherapy, seven separate logistic regression models were built to assess the likelihood that calls regarding the pharmacotherapy were from healthcare providers (dependent variable), compared to medications in each of the seven risk categories. The independent variable was the risk category of the medication that was

the subject of the call. The odds ratios were adjusted for SES and remoteness of the caller. Adjusted odds ratios (OR) and 95% confidence intervals (95% CI) are reported.

All data cleaning and statistical analyses were performed using SPSS Statistics V.24 (IBM SPSS Statistics for Windows, Version 24.0, Armonk, NY: IBM Corp).

Figure 2 Selection of eligible calls made to MotherSafe between 1 January 2001 and 31 December 2016

271,704 calls were made between 2001 and 2016

133,333 calls were excluded

103,603 breastfeeding 14,024 pregnancy- planning 9,322 pregnancy status missing 4,692 general inquiries 1,108 both pregnancy and breastfeeding 584 retrospective-related

138,371 calls were pregnancy-related

62,350 calls were excluded 31,398 exempted medications 12,003 combined TGA-listed medications

9,761 unable to map to TGA list 2,211 alcohol use and substance abuse

3,789 missing entries for exposure type 1,877 queries 1,311 wrong number 76,021 pregnancy-related calls were TGA-categorised .

7,142 calls were excluded 4,547 other categories of providers 2,595 missing entries for caller types

68,879 pregnancy-related, TGA-categorised calls were from providers and non-providers

2,192 calls were excluded, missing entries for postcodes (adjusted as covariate)

66,687 calls analysed (2001-2016 cohort)

50,396 calls analysed (2008-2016 cohort) 56

Results

Table 4: Characteristics of calls to MotherSafe for 2001-2016 and 2008-2016

Characteristics 2001-2016 2008-2016 n, % (N=66,687) n, % (N=50,396) Primary medication exposure Category A 26,709 (40.1) 19,651 (39.1) Category B1 6,546 (9.8) 5,558 (11.1) Category B2 7,860 (11.8) 5,973 (11.9) Category B3 10,131 (15.2) 8,117 (16.2) Category C 10,646 (16.0) 7,926 (15.8) Category D 4,183 (6.3) 2,775 (5.5) Category X 269 (0.4) 153 (0.3) Bupropion 21 (0.03) 9 (0.0) Varenicline 47 (0.07) 47 (0.1) NRT 275 (0.4) 187 (0.4) Socioeconomic status (SES) Disadvantaged 8,243 (12.4) 6,417 (12.7) Average 22,228 (33.3) 17,051 (33.9) Advantaged 36,216 (54.3) 26,928 (53.4) Remoteness of area Remote 177(0.3) 146(0.3) Regional 7,317 (10.9) 5,599 (11.1) Major cities 59,193 (88.8) 44,651 (88.6)

Healthcare provider callers (n=13,050, 19.6%) (n=10 589, 21.0%) GP 8 022 (12.0) 6 625 (13.1) Midwife 2 209 (3.3) 1709 (3.4) Obstetrician 1 208 (1.8) 821 (1.6) Pharmacist 1 611 (2.4) 1434 (2.8) Non-provider callers (n=53 637, 80.4%) (n=39 807, 79.0%) Women 51 217 (76.8) 37 757 (74.9) Family 2 060 (3.1) 1 929 (3.8) Partner 224 (0.3) 10 (0.0) Friend 119 (0.2) 101 (0.2) Others including journalist, chief 17 (0.0) 10 (0.0) medical officer, pharmaceutical company, translational and interpreting

services interpreter

The characteristics of calls made in the 2001-2016 and 2008-2016 periods were similar in terms of the distribution of risk category for the primary medication exposure and type of caller (Table 1). Of the total 66,687 calls received in 2001-2016, exposures regarding category A and X medications accounted for 40.1% and 0.4%, respectively, of the total calls. The proportion of cessation pharmacotherapy-related calls was 0.5%, with 21 (0.03%) relating to bupropion, 47 (0.07%) regarding varenicline and 275 (0.4%) regarding NRT. Healthcare providers accounted for approximately 20% of calls, most of which (61.5%) were from GPs. Low proportions of calls were from callers in disadvantaged (12%) and remote areas (0.3%).

Compared to category A medications, calls about NRT were more likely to be from healthcare providers (adjusted odds ratio (aOR) 7.78, 95% Confidence Interval [CI] 6.08,9.97) (Figure 2). This pattern was observed for comparisons against category B1, B2, B3 and C (all p <0.001). Compared to category D medications, calls regarding NRT were equally likely to be made by healthcare providers (aOR OR 1.20, 95% CI 0.93,1.54). Compared to category X medications, calls regarding NRT were equally likely to be made by healthcare providers (aOR OR 1.31, 95% CI 0.91,1.89).

Compared to category B2 medications, the risk category to which bupropion belongs, calls about bupropion were more likely to be from healthcare providers (aOR 2.77, 95% CI 1.17,6.59). This pattern was observed for comparisons against medications in lower risk categories, including category A and B1 (aOR 10.16, 95% CI 4.26,24.23, aOR 2.74, 95% CI 1.15,6.49, respectively). Compared to medications in higher risk categories, i.e. categories B3, C, D, and X, calls regarding bupropion were equally likely to be made by healthcare providers (all p > 0.05).

Compared to category B3 (varenicline’s risk category) medications, calls regarding varenicline were more likely to be from healthcare providers (aOR 2.33, 95% CI 0.91,5.18). This pattern was observed for comparisons against medications in lower risk categories including category A, B1and B2 (all p <0.05). Compared to category C medications, calls regarding varenicline were more likely to be from healthcare providers (aOR 2.33, 95% CI 0.91,5.18). Compared to category D and X medications, calls regarding varenicline were equally likely to be made by healthcare providers (aOR 1.66, 95% CI 0.92-2.97; aOR 1.69, 95% CI 0.83,3.44, respectively).

Figure 3 Likelihood for call regarding each smoking cessation pharmacotherapy to be from providers compared to other categories of medications, 2001-2016 (and 2008-2016 for varenicline)

Discussion This study, for the first time, measured the level of concern among healthcare providers regarding smoking cessation pharmacotherapy exposure during pregnancy compared to other medications. This was explored as a potential explanation for the low prevalence of the use of smoking cessation pharmacotherapies during pregnancy.

It was found that the level of concern from healthcare providers regarding NRT exposure during pregnancy is similar to their concern regarding medications of equivalent risk (category D). Category D medications, however, may not be a fair benchmark against which to compare NRT. Category D medications are those suspected of causing irreversible harm to the foetus . In the case of nicotine, it should be acknowledged that smoking itself poses the same risks as nicotine, as well as additional risks associated with other harmful substances in cigarettes. Indeed, it is on the basis that NRT is considered safer than smoking, and that antenatal care guidelines recommend the use of NRT during pregnancy. While it is not clear what the appropriate comparison category is, a significantly greater level of concern was shown for NRT compared to all lower risk categories. These findings suggest that healthcare providers perceive the risk associated with NRT use during pregnancy to be similar to the risk associated with potentially teratogenic medications, without adjusting their perceptions to account for the risk of harm associated with the underlying smoking during pregnancy.

Healthcare providers were more likely to call about bupropion and varenicline than medications of equivalent risk during pregnancy, suggesting an overestimation of the risks associated with the use of these pharmacotherapies during pregnancy. In fact, healthcare providers had a greater level of concern regarding varenicline than other medicines in the higher risk category, category C. This overestimation of risk is not surprising given that antenatal care guidelines recommend that varenicline and bupropion not be used during pregnancy. However, current and emerging evidence indicate neither bupropion nor varenicline is associated with congenital malformations, birthweight or premature birth [128, 223, 230]. It is possible that in future, both antenatal care guidelines and regulatory bodies may revise recommendations for using bupropion and varenicline during pregnancy as further evidence emerges.

As healthcare providers’ safety concerns about NRT and bupropion influence their prescribing practices in non-pregnant patients [215, 217], the overestimation of risk regarding smoking cessation pharmacotherapies is likely to limit healthcare providers’ intention to prescribe smoking cessation pharmacotherapies to women who are pregnant. This is likely to explain, at least in part, the low rates of use of NRT among pregnant women in Australia, and perhaps in the US. For maternal and perinatal outcomes among women who smoke during pregnancy to be improved, this misperception needs to be corrected, particularly for pregnant smokers who are in greater need of assisted quit support, including those who are young, disadvantaged, have limited education and lack access to resources [30, 214, 231]. This should include more widespread communication of the existing safety data regarding these products, and the further conduct of high-quality safety studies in the pregnant population.

Strengths and limitations Strengths of this study include that it is based on state-wide telephone data from the most populous state in Australia. Records for a large number of calls were collected over 16 consecutive years, thereby reducing the likelihood that the findings represent a temporary phenomenon associated with policy changes or local adverse events.

Due to the exploratory nature of this analysis, over-interpretation of the study findings must be cautioned. Several limitations of this study must be considered. First, the calls were assumed to represent concerns. It is possible that some calls may reflect interest in the medication, with callers seeking information about the dosing and adverse effects. However, this is unlikely to be the case for a large proportion of the calls because previous studies demonstrated that 80-95% of TIS calls relate to safety concerns [225, 226]. Second, the sample of TIS callers on which this study is based may not represent all concerned individuals. Those who call a TIS are likely to differ from the population of concerned individuals on factors that influence the likelihood of calling a TIS, such as information-seeking behaviour, access to alternative resources, and personality traits [232]. While TIS callers may differ systematically from the underlying population of concerned individuals, this does not necessarily mean that selection bias is present. For selection bias to occur, selection must be related to both study factor and the outcome [233]. In this study, it is unlikely that factors associated with selection would differ systematically between the comparison groups, that is, 61

callers who were concerned about smoking cessation pharmacotherapies and callers who were concerned about other medications. The exception to this, is that providers of certain specialties probably more often call regarding medications specific to their specialty. Thus, this study excluded the calls made by other specialties including oncologists, rheumatologists, paediatricians, cardiologists and others. Third, given that TIS data are not designed for research, limited information on the characteristics of the callers was available, potentially giving rise to residual confounding. The lack of identifying information for callers meant that it was not possible to account for clustering of calls within callers. This inability to account for clustering may have led to an underestimation in the standard errors of logistic regression coefficients, and thus, potentially overestimation in the precision of the odds ratios [233]. Fourth, a separate comparison of each smoking cessation pharmacotherapy against seven risk categories increased the possibility that at least one statistically significant effect would be observed across the multiple comparisons (a Type 1 error). The fifth limitation is related to the external validity of the study findings. Similar to other TIS-based studies, a high proportion of callers were from advantaged areas as well as from major cities [232]. Although adjustment for area-based measures was made in the analyses, this selected sample limits broader generalisation to other populations of healthcare providers. The final limitation of this study is that the analyses were based on a small number of calls, particularly for bupropion and varenicline. Therefore, there is poor precision in the estimation of odds ratios. However, the effects of the main comparisons were strong enough to reach statistical significance.

Conclusion Using calls to TIS as a proxy for concern, this study found that health care providers were more concerned about bupropion and varenicline during pregnancy than same- risk category medications. Providers were equally concerned about NRT and category D risk medications, but more concerned about NRT than medications in lower risk categories, where these lower risk categories may be a fairer comparison. An overestimation of risk associated with cessation pharmacotherapies may limit their use during pregnancy. These findings indicate the need for more robust safety studies of smoking cessation pharmacotherapies in pregnant smokers.

Chapter 3 Factors associated with obstetricians’ and gynaecologists’ prescribing of smoking cessation pharmacotherapies to pregnant women

Introduction The previous chapter established that healthcare providers were more likely to call a teratology information service regarding smoking cessation pharmacotherapies than other medications of the same pregnancy risk category. The elevated concern for these pharmacotherapies is consistent with existing evidence (outlined in Chapter 1) that healthcare providers are reluctant to prescribe pharmacotherapies for smoking cessation to pregnant women, and is likely to contribute to the limited use of these pharmacotherapies during pregnancy. The study described in the current chapter identifies additional factors that influence healthcare providers’ prescribing of smoking cessation pharmacotherapies to pregnant women.

This study focuses on obstetricians and gynaecologists. In a previous Australian study, obstetricians and gynaecologists were four times less likely to prescribe NRT to pregnant smokers than general practitioners (GPs) [219]. Furthermore, in the Australian maternity care system, high-risk pregnancies (medically defined as pregnancies among women with pre-pregnancy or during-pregnancy medical complications [234, 235]), are more often managed by obstetricians and gynaecologists than GPs [236]. Among women who smoke, those who have high-risk pregnancies are likely to benefit from smoking cessation support [236]. It is, therefore, important to concentrate on this group of healthcare providers who manage the obstetric care of these high-risk pregnancies.

As indicated in the introductory chapter (Section 1.1.3.5.3.2), limited knowledge about smoking cessation pharmacotherapies potentially impacts the ability to prescribe these pharmacotherapies to pregnant women. While gaps in knowledge among obstetric providers regarding NRT have been cited as a barrier to prescribing NRT to pregnant women in the UK (2018) [131], and Australia (2019) [132, 133], this barrier may be even more apparent for the newer bupropion and varenicline. It is necessary to examine factors that contribute to the lack of familiarity with these pharmacological therapies as such information will help guide the design of education efforts to improve the prescribing of smoking cessation pharmacotherapies to pregnant women.

There is a wide range of factors that could potentially contribute to a lack of awareness of pharmacotherapies for smoking cessation among obstetricians and gynaecologists. This study focuses on factors that operate at the level of the health services in which the obstetricians and gynaecologists work (hereafter referred to as ‘facility-level factors’) as there is evidence that implementing changes at the facility-level have led to improvements in the provision of smoking cessation care in maternity hospitals [134], community mental health services [135], and primary care settings [136, 137]. By way of example, a 2017 cross-sectional survey of GPs, obstetricians and gynaecologists in Australia found that workplace routine in providing smoking cessation support was associated with an increased likelihood (aOR 1.71, 95% CI 1.03-2.80) of clinicians carrying out the ‘Ask’, ‘Advise’ and ‘Refer’ steps of the modified 5A’s process [237]. Qualitative interviews of obstetric care providers in NSW in 2018 highlighted that different facility-level characteristics served as barriers (‘systems unsupportive of implementation or monitoring of smoking cessation guidelines’) and enablers (‘systems that regulated implementation of guidelines’). in implementing the 5A’s of smoking cessation care to pregnant patients [133]. Recognising the potential of facility-level interventions in the field of smoking cessation, a randomised controlled trial is currently being carried out in 33 addiction centres across four States and Territories in Australia evaluating whether organisational changes reduce smoking rates among enrolled clients [238]. Studies [216, 237, 239-241] revealed that facility-level factors such as work setting (including community/integrated or hospital-based services) [240], service structure [241] and the presence of smoking cessation care protocols [216, 239] are associated with the extent and type of smoking cessation support provided to pregnant smokers. There is little evidence of whether facility-level factors influence obstetricians’ and gynaecologists’ awareness of pharmacotherapy for smoking cessation.

Yet the possession of knowledge alone may not be sufficient to influence prescribing decisions and practices [138], as indicated in Chapter 1 (section 1.1.3.5.3.2). There exists quantitative [139], and qualitative [140] evidence that improved knowledge does not wholly account for greater prescribing of NRT. Actual prescribing is the behavioural outcome that stems from the complex interplay of determinants of the decision to prescribe and the eventual prescribing [141], which implies that other factors guide prescribing behaviour. In particular, this study aims to identify factors associated with the intention to prescribe smoking cessation pharmacotherapies, where systematic reviews [242, 243] have demonstrated intentions to perform a behaviour to be a strong indicator of actual clinical practice. 64

In addition to identifying facility-level factors associated with the intention to prescribe smoking cessation pharmacotherapies, this study also seeks to identify factors operating at the level of individual beliefs (hereafter referred to as ‘individual-level factors’). The role of individual-level factors in influencing the intention to prescribe other medications has been established in previous research [142]. As prescribing is assumed to be a rational behaviour, several social-cognitive models including Raish’s theory of persuasion [244] Bandura’s Social Cognitive Theory [245], and Azjen’s Theory of Planned Behaviour [141] have been used to address factors influencing the decision to prescribe. A well-established conceptual framework, the Theory of Planned Behaviour is employed in the current study to inform the selection of the factors to be examined. Icek Ajzen’s Theory of Planned Behaviour (1991) is one of the most influential models used to explain healthcare providers’ behavioural change [142]. A 2008 meta-analysis reported that this theory was able to explain physicians’ behaviour to a substantial extent (frequency-weighted mean of the explained variance, R2 of 0.59) [142]. This is compared to the value of R2 of 0.42 for Raish’s theory of persuasion and Bandura’s Social Cognitive Theory [142].

Ajzen’s Theory of Planned Behaviour has been widely used in studies that seek to influence changes in prescribing of many medications (antibiotics [246, 247], statins [248], oral NSAIDs, diuretics, angiotensin-converting enzyme inhibitors, antipsychotics, oral anticoagulants [249], stimulants [250], and psychotropics [251]) among a range of different healthcare providers (physicians [250, 252, 253], psychiatrists [251], nurses [254], and GPs [255]).

In the context of smoking cessation care, cross-sectional surveys also found that this theory explained 50-74% of the intention to offer smoking cessation support among dentists and dental hygienists [256, 257], mental health professionals [258], physical therapists [259], and smoking cessation counsellors [260] and advisors [261]. Thus, this theory appears to be a useful tool to guide the identification of factors relating to the intention to prescribe pharmacotherapies for smoking cessation.

Aims The study described in this chapter has two aims. The first is to identify facility-level factors associated with obstetricians’ and gynaecologists’ familiarity with smoking cessation pharmacotherapies. Second, among obstetricians’ and gynaecologists’ who are familiar with these pharmacotherapies, the study aims to identify the facility-level factors as well as individual-level factors, associated with obstetricians’ and gynaecologists’ intention to prescribe these pharmacotherapies.

Theoretical framework Ajzen’s Theory of Planned Behaviour (see Figure 4) [141, 262, 263], is the conceptual framework driving the selection of individual-level factors potentially relating to the intention to prescribe smoking cessation pharmacotherapies to pregnant women. Ajzen’s Theory proposes that an individual is likely to perform a behaviour based on the intention to engage in the behaviour [141].

Figure 4 The Theory of Planned Behaviour*

Attitude

Perceived Benefit Perceived Risk

Subjective Norms Intention Behaviour Normative Beliefs Motivation to Comply

Perceived Behavioural Control

External Factor Internal Factor

*adapted from Ajzen (1985 and 1991) [141, 262]

This theory states that individuals process information rationally before deciding on their intention to perform a behaviour, and subsequently carrying out the behaviour [141, 262]. According to this theory, an intention to prescribe is influenced by three belief constructs: (a) Attitude regarding whether prescribing is viewed as beneficial or not; (b) Subjective Norms regarding whether prescribing is expected by others such as the patient, society and peers; and (c) Perceived Behavioural Control related to whether it is easy and feasible to prescribe.

All three constructs may not equally contribute to the actual performance of a behaviour. Generally, Subjective Norms are found to be a weak predictor of intention to perform a behaviour [264], including the intention to: (a) prescribe [248, 265, 266]; and (b) provide smoking cessation support [258, 267]. In contrast, the Perceived Behavioural Control construct has the strongest link to the intention to carry out the behaviour [253] with specific evidence of a strong link with the offering of smoking cessation support [260]. The strength of the Attitude construct was comparable to the magnitude of influence for the Perceived Behavioural Control construct in studies that examined prescribing practices [247, 266], and provision of smoking cessation care [260].

In this theory, Ajzen postulates that each of the constructs is based on underlying sub- constructs [141, 262]. The Attitude construct is manifested by two underlying sub- constructs, namely the Perceived Benefit and Perceived Risk [268]. In the current study, the attitude towards prescribing is conceptualised as the perceived benefit and risk of prescribing pharmacotherapies for smoking cessation to pregnant women. Ajzen also posits that the Perceived Behavioural Control construct refers to the perceived ability to perform the behaviour, which is driven by internal and external factors [263]. External Factors include the healthcare environment which impacts opportunities and resources that are needed by an individual to carry out the behaviour [263], represented in the current study as facility-level factors. Internal Factors driving Perceived Behavioural Control consist of inherent knowledge and perceived confidence to prescribe (i.e. self-efficacy) [263]. The construct for Subjective Norms is based on the perceived expectation of others (peers, patients or society), that is, the Normative Beliefs sub-construct, and the extent of the individual’s desire to yield to the others’ expectation, that is, the Motivation to Comply sub-construct [262]. 67

The Theory of Planned Behaviour recognises that conduct of behaviour can occur either through intention to perform that behaviour or despite the intention, that is, a pathway is independent of intention. This is reflected in the model (see Figure 4), whereby the individual’s perceived control influences intention, but also directly affects whether the behaviour is performed (characterised by the dotted arrow).

This study examined the role of both the Attitude and Perceived Behavioural Control constructs in predicting intention to prescribe. Prior research has suggested that these constructs are likely to influence obstetricians’ and gynaecologists’ intention to prescribe smoking cessation pharmacotherapies to pregnant women. Surveys among obstetricians and gynaecologists in other countries found that perceived safety and effectiveness of NRT use during pregnancy, anticipated lack of patients’ compliance and providers’ self-efficacy to prescribe NRT; all influenced the extent of prescribing NRT to pregnant women [215, 219, 269]. A 2018 cross-sectional survey of GPs and obstetricians in Australia found that prescribing NRT to pregnant women was more likely among those who perceived NRT as safer than smoking (aOR 3.24, 95% CI 2.21-4.77), effective (aOR 2.73, 95% CI 1.82-4.10), and had patients with good adherence (aOR 2.19 95% CI 1.06-4.51) [219]. Within the Perceived Behavioural Control construct, the Internal Factor of self-efficacy was examined as the aforementioned 2018 survey found that possession of confidence to prescribe NRT was associated with a strong likelihood to prescribe NRT to pregnant women (aOR 8.60, 95% CI 5.64-13.19) [219]. The Internal Factor of knowledge is removed from consideration, with the intention to prescribe only examined among obstetricians’ and gynaecologists’ familiar with these pharmacotherapies.

This study did not consider the construct for Subjective Norms as previous research has demonstrated its weak influence on the intention to prescribe. Moreover, as demonstrated through the piloting of the questionnaire (section 3.4.2), the consensus among the eight obstetricians was that the belief relating to Subject Norms was not preferred as an individual-level factor that related to their intention to prescribe pharmacotherapy for smoking cessation.

Methods This was a cross-sectional study using data collected through an online survey between November 2017 and January 2018.

3.4.1.1 Inclusion and exclusion All obstetricians and gynaecologists who, at the time of the study, were affiliated with the Royal Australian and New Zealand College of Obstetricians and Gynaecologists (RANZCOG), or members of Women's & Children's Healthcare Australasia (WHA) were invited to participate in this study. Only obstetricians and gynaecologists who were practising in Australia were eligible for inclusion in this study. From here on, this chapter uses the term ‘obstetrician’ to refer to a clinician who practised obstetrics and/or gynaecology.

RANZCOG is a professional body that represents all obstetricians in Australia and New Zealand. As of 2017, RANZCOG has a membership of approximately 1,841 Fellows and 446 trainees practising in Australia. Although all obstetricians are required to be affiliated with RANZCOG to practice in Australia, this study also recruited obstetricians affiliated with WHA to maximise survey participation. WHA is the not-for-profit, peak body for publicly funded hospitals providing maternity and women's healthcare across Australia. WHA has a membership of more than 100 maternal health services in urban and rural areas.

In order to be eligible to participate in the study, respondents were required to care for pregnant patients. This was established through the first question in the survey, which asked ‘In the last six months, have you seen pregnant patients in clinical practice?’ Respondents who selected “no” were filtered out from the survey and did not contribute further data.

3.4.1.2 Recruitment strategy All obstetricians who were affiliated with RANZCOG received an email invitation to participate (see Appendix 3A Email invitation to participate in the survey) from the RANZCOG Continuing Professional Development and Revalidation Committee (CPDRC). The email included an introductory paragraph describing the purpose of the survey, a link to the online survey (https://unsw.au1.qualtrics.com /surveys/SV_0cziKElvji6QyON), and a copy of the participant information statement and consent form as a single-scroll layout prior to the beginning of the survey. A follow- up email was sent after four weeks from the distribution date to remind affiliates of the invitation.

All principal contacts who were affiliated with the WHA collaborative network of maternity services received an email invitation to participate from the WHA liaison officer. There is one principal contact for each maternity service. Each WHA principal contact was asked by the WHA liaison officer to distribute the link to the online survey among the practising obstetricians and gynaecologists within their organisation. In addition, the survey was advertised on the WHA secure portal for members (https://women.wcha.asn.au/news/prescribing-smoking-cessation-pharmacotherapies- women-who-smoke-during-pregnancy-survey), as shown in Appendix 3B WHA website recruitment of survey participants).

The full questionnaire is presented in Appendix 3C. Briefly, the questionnaire had 16 questions, containing 47 items. The questionnaire was organised into six sections, comprising questions relating to: (1) prescribing of smoking cessation pharmacotherapies; (2) statements regarding NRT patches, mapped from constructs and sub-constructs of Ajzen’s Theory of Planned Behaviour; (3) statements regarding varenicline, mapped from constructs and sub-constructs of Ajzen’s Theory of Planned Behaviour; (4) smoking cessation care practices; (5) general questions; and (6) clinical practice setting. The questionnaire also included an item in which the survey respondents could indicate their interest in receiving self-education materials relating to smoking cessation pharmacotherapies.

This survey did not examine other formulations of NRT such as NRT gums, lozenges or inhalers because, at the time of this study, NRT patches were the only formulation of NRT available on prescription in Australia. Nor did this survey examine factors relating to bupropion prescribing practices because the use of bupropion in the general Australian population in recent years has been low, precluding it from being the focus of observational research on prescribing intention and behaviour [270].

All respondents were asked to answer questions in sections: (1) prescribing of smoking cessation pharmacotherapy; (4) smoking cessation care practices; (5) general questions; and (6) clinical practice settings. Respondents who selected “yes” to the question regarding familiarity with NRT patches and/or varenicline (‘are you sufficiently familiar with NRT patches or varenicline to answer questions regarding your beliefs or experiences around prescribing them?’), were asked to answer additional items in section: (2) statements regarding NRT patches, and/or; (3) statements regarding varenicline.

The questionnaire was piloted with eight obstetricians for content clarity. Following the pilot, several changes were made to the questionnaire such that: (a) the section pertaining to general questions was moved to the end of the questionnaire, (b) clarification of one question (reference to NRT patches), (c) removal of statement relating to the Subjective Norms (peer influence) construct and, (d) an additional question regarding smoking cessation care practices (inclusion of smoking cessation care protocol).

The final questionnaire was uploaded to an online platform, Qualtrics (Qualtrics Provo, Utah, USA) where it was hosted throughout the data collection period. This online format was tested by two non-obstetricians to obtain feedback on the flow of the survey and the ease of completion. The questionnaire took no more than 10 minutes to be completed. The survey interface allowed for single screen scrolls and displays of multiple item screens. The questionnaire could be accessed using mobile phones, tablets, laptops and all other electronic devices on commonly used operating systems and browsers. The survey was also built to generate a QR code (Quick Response

code) for offline access to the questionnaire. The online testing of the questionnaire resulted in no further changes to the survey.

3.4.4.1 Facility-level factors This study considered five facility-level factors that potentially influence the level of awareness (the first aim of this study) and intention to prescribe (the second aim of this study) among obstetricians. In this study, the first three facility-level factors examined were the presence of a smoking cessation care protocol, funding of care (public vs private), and level of training. These characteristics were selected because there is evidence that the presence of a smoking cessation care protocol [132, 216, 239], funding of care [216] and level of training [271, 272], influenced smoking cessation care delivery by obstetricians. The respondent’s level of training was considered a facility- level factor because the responsibility for clinical care in a healthcare institution, and thus smoking cessation care, is largely held by Fellow obstetricians. Thus, intervening on this factor has the potential to result in changes across the whole facility.

Although multiple response options were available to respondents on these items, some options were infrequently selected. Thus, in the analyses, several of the response options were combined for two facility-level factors. The final response categories were coded as follows: ‘agree’/ ‘disagree’ (including ‘unsure’) for the availability of smoking cessation care protocol; and ‘Fellows’/’trainees (including ‘pre-’ and ‘post-membership’) for the level of training. The original response options for funding of care were maintained in the analysis: ‘public’/ ‘private’/ ‘both public and private’.

Two additional facility-level factors, the maternity care model in which the respondent practised, and remoteness of their practice, were considered important to this study. In Australia, maternity care models influence the type of antenatal service delivery [273]. Recent systematic reviews found evidence that the provision of obstetric care for pregnant women differed across models of care according to the healthcare provider who held responsibility for the care (obstetrician-led, vs shared with other providers such as GP and/or midwives) and access to care (continuum of care, settings of care) [274, 275]. Smoking cessation support may also differ by this characteristic of care. 72

Responses to this item regarding maternity care model was coded as ‘shared care with other providers’ if the respondents indicated that they provided any form of shared care, otherwise it was coded as ‘no shared care’.

Remoteness was selected as a facility-level factor because it provides a proxy for both the level of hospital in which a respondent practices, as well as the geographical access to obstetric and alternative smoking cessation care through general practitioners, support groups, and drug and alcohol services. Remoteness was an important predictor for variation in delivery of smoking cessation support among cardiologists in a prior survey carried out in the US, with cardiologists practising in rural or suburban areas less likely to provide cessation support than those in urban areas (OR 0.92, 95% CI 0.88-0.95, for rural vs urban; OR 0.94, 95% CI 0.91-0.97, for suburban vs urban) [276].

The postcode nominated by the respondent was mapped to an area-based measure of the remoteness of the practice, classified according to the Accessibility Remoteness Indicator of Australia [229]. Remoteness of practice was categorised as ‘major cities’ and ‘regional and remote’.

Free text responses were given by less than five respondents to questions relating to facility settings and maternity care models, and, thus, were not analysed.

3.4.4.2 Individual-level factors: Attitudes and Perceived Behavioural Control constructs Respondents who responded “yes” to the question regarding familiarity with NRT patches and/or varenicline were eligible to complete subsequent questions in which they indicate their level of agreement with statements about NRT patches and varenicline. Table 5 outlines the questionnaire items and how they map to constructs of Ajzen’s Theory of Planned Behaviour [141, 262], that is, Attitudes and Perceived Behavioural Control, and their corresponding sub-constructs.

Each statement relating to the Attitude or Perceived Behavioural Control constructs was coded as ‘agree’ if the response was ‘agree’, and was coded as ‘disagree’ if the response was ‘disagree’ or ‘unsure’. The ‘unsure’ responses were combined with

‘disagree’ responses because the small sample size did not allow for the ‘unsure’ responses to be treated as a separate category for analyses.

Table 5 Organisation of survey items and response options Sections Items Response options Prescribing of smoking Familiar with NRT patches Yes, No cessation pharmacotherapy Familiar with varenicline

Intend to prescribe NRT patches to pregnant Yes, No, Unsure women Intend to prescribe bupropion to pregnant women Intend to prescribe varenicline to pregnant women

Prescribing NRT patches in the last 6 months Yes, No Prescribing bupropion in the last 6 months Prescribing varenicline in the last 6 months

Statements regarding NRT Attitude: Perceived Risk Agree, Disagree, Unsure patches, mapped from Safer than smoking constructs and sub-constructs Concerned about safety in certain patients of Ajzen’s Theory of Planned Had patients who developed adverse events Behaviour Few patients have co-occurring medical conditions Continuous dosing is NOT a barrier Patient’s compliance is lacking

Attitude: Perceived Benefit Effective May benefit heavy smokers May benefit smokers with failed quit attempts May benefit light smokers

Perceived Behavioural Control: External Factor Aware that it is PBS-listed Supportive guidelines

Sections Items Response options Perceived Behavioural Control: Internal Factor Hard to assess the benefit-risk ratio

Statements regarding Attitude: Perceived Risk Agree, Disagree, Unsure varenicline, mapped from Safer than smoking constructs and sub-constructs Concerned about safety in certain patients of Ajzen’s Theory of Planned Had patients who developed adverse events Behaviour Few patients have co-occurring medical conditions Patient’s compliance is lacking

Attitude: Perceived Benefit Effective May benefit heavy smokers May benefit smokers with failed quit attempts May benefit light smokers

Perceived Behavioural Control: External Factor Aware that it is PBS-listed Supportive guidelines

Perceived Behavioural Control: Internal Factor Hard to assess the benefit-risk ratio

Smoking cessation care Enough time to discuss cessation Yes, No, Unsure practices Smoking cessation care protocol is present Patients are interested to quit

General questions* Types of practice care activity performed within Patient care, Administrative, Academia, Research, Public health, the last six months Retired, Others

Proportion of the respondent’s weekly working 0%-100% hours which was spent in seeing patients 76

Sections Items Response options

Specialty General obstetrics, General gynaecology, General obstetrics and gynaecology, Maternal-fetal medicine, Reproductive endocrinology and infertility, Gynaecologic oncology, Urogynaecology

Level of training Pre-membership RANZCOG trainee, Post-membership RANZCOG trainee, RANZCOG Fellow

Year of awarded fellowship 1960-2017

Smoking status Never (or never regularly smoked), Former smoker, Current smoker (not trying to quit), Current smoker (trying to quit)

Proportion of pregnant patients who are smokers <5%, 5-15%, 16-30%, >30% seen in the last 6 months Clinical practice settings Postcode of facility 4-digit number

Type of facility setting Predominantly public, Predominantly private, Both public and private, Others

Maternity care models No shared care, Shared care with general practitioners, Shared care with midwives

*At the request of RANZCOG, questions relating to gender were not included.

3.4.4.3 Familiarity with and intention to prescribe smoking cessation pharmacotherapies Respondents who responded “yes” to the question regarding familiarity with smoking cessation pharmacotherapies was coded as “yes”, and “no” if they responded “no”. Respondents who responded “yes” to the question regarding their intention to prescribe smoking cessation pharmacotherapies in the next six months (in the next six months, are you likely to prescribe any of the following to your pregnant patients?) was coded as “yes”, and “no” if their response was “no” or ‘unsure’. This study used this dichotomisation because there were too few respondents who selected the ‘unsure’ and “no” responses.

3.4.4.4 Sample characteristics Questions regarding the characteristics of the sample included proportion of time spent in seeing patients, type of practice activity, primary specialty, socio-economic status of the area and state/territory in which the respondent practised, as well as smoking cessation care practices.

Responses to the question regarding the proportion of time spent seeing patients were categorised as ‘50%-100%’ and ‘fewer than 50%’. The response category to the types of practice activity was coded as ‘patient care only’ if the respondents indicated that they provided patient care only, otherwise it was coded as ‘patient care and others’. Response category to primary specialty question was coded as ‘general obstetrics (including gynaecology)’ if the respondents indicated that they specialised in general obstetrics or general gynaecology or general obstetrics and gynaecology, otherwise it was coded as ‘subspecialty’.

Socio-economic status of the area, and the state or territory in which the respondent practice, were derived from the postcode nominated by the respondent. The nominated postcode was mapped to an area-based measure of socioeconomic status (SES), using Socioeconomic Indexes for Areas Index of Relative Socioeconomic Disadvantage (IRSD) scores [228]. Based on the data from the 2006 Australian Census of Population and Housing, the IRSD scores were categorised in the following tertiles: disadvantaged SES (lower than 40th percentile), average SES (40-79th percentile) and advantaged SES (at least 80th percentile) [228]. The nominated 78

postcode was also mapped to the postal area (POAs) using postcode indices [277]. Based on the data from the 2016 Australian Statistical Geography Standard (ASGS), the POAs were grouped into states and territory of practices [277].

Responses to the questions assessing respondents’ perceptions of their patients’ interest to quit smoking and whether they consider there is enough time to address smoking cessation were categorised as “yes” for respondents who indicated “yes”, otherwise the responses were categorised as “no”, as very few selected the ‘unsure’ option. Responses to the 6-month prescribing history of smoking cessation pharmacotherapies were coded as “yes” / “no”, with the respondents who indicated “yes” to select the type of smoking cessation pharmacotherapy(s) prescribed. Respondents who indicated “yes” to the question about their interest in receiving educational materials about smoking cessation pharmacotherapies had their responses categorised as “yes”, otherwise the responses were categorised as “no”.

Free text responses to questions pertaining to practice activity and primary specialty were given by less than five respondents and, thus, were not analysed.

3.4.4.5 Potential confounders Three factors that may be associated with both the facility-level factors examined and the respondents’ familiarity with smoking cessation pharmacotherapies were considered potential confounders in the analysis examining the association between facility-level factors and familiarity with NRT patches. These included the respondent’s smoking status, the proportion of patients’ seen who smoked (past 6 months), and the year they obtained their Fellowship.

The responses to the variable ‘the proportion of patients seen who smoked (past six months)’ was categorised as ‘<5%’ / ‘5-15%’ / ‘>15%’. The responses to the question relating to the respondent’s smoking status were coded as ‘ever smoker’, if the respondent indicated that were currently smoking or had smoked in the past and ‘non- smoker’ if the respondents did not smoke at all. The variable ‘years after Fellowship’ was extracted using the difference between the year of the survey (2017/2018) and the year in which the respondent was awarded their Fellowship. This variable is taken to be an indicator of their work experience. Responses to this variable were coded as ‘fewer than 5 years’ / ‘5-10 years’ / ‘at least 15 years’. 79

The email invitation to participate in this study included an attached participant information statement and consent form (see Appendix 3A). In addition, respondents who clicked on the survey link were directed to an identical online participant information statement and consent form. After providing their consent by selecting “yes” to the question ‘Do you consent to participate’ the respondents were able to start the survey.

Each respondent was assured of voluntary participation and anonymity of their responses. The respondents did not have to answer all the questions to complete the survey. To ensure their anonymity, the survey’s settings in Qualtrics did not document their email address or their computer’s IP address.

Respondents could claim one Continuous Professional Development points from RANZCOG CPDRC upon completion. The survey was approved by the UNSW Human Research Ethics Committee, HC17750 (see Appendix 3D).

During the data collection period, survey data were stored on the Qualtrics platform. After the closure of the survey, the data were extracted into a Microsoft Excel spreadsheet and stored in a password-protected folder on the secure UNSW server. The data on the Qualtrics server was then deleted. Email addresses of the respondents who requested the results of the survey were collated and stored in a different folder on the secure UNSW server.

The Microsoft Excel spreadsheet was subsequently exported into IBM SPSS software version 24 for analyses.

The password-protected files are retained for seven years in the university data repository, in accordance with UNSW data management and storage policies.

To assess whether the survey respondents in this study were a representative sample of obstetricians in Australia, this study compared characteristics of the survey respondents to that of all RANZCOG-affiliated obstetricians who practised in Australia in 2017. The characteristics of all RANZCOG affiliates were obtained from the published RANZCOG membership data in 2017 [278]. The comparison was made with respect to two variables, state and territory of obstetric practices, and trainee/Fellow status. For Fellows only, the published data allowed for comparisons on additional facility-level factors including remoteness of practice and primary specialty. Comparison in terms of practice settings was also made using data obtained directly from the RANZCOG CPDRC. These comparisons were carried out using chi-square tests.

The examination of factors relating to familiarity with, and intention to prescribe smoking cessation pharmacotherapy was limited to NRT patches. Factors related to the intention to prescribe and the familiarity with varenicline were not examined due to few respondents indicating their intention to prescribe (n=3), and their familiarity with varenicline (n=13).

To address the first aim of the study, all survey respondents (n=163) were included in the analyses. A multivariate logistic regression model was built with familiarity with NRT patches (section 3.4.4.3) as the study outcome, and the five facility-level factors (section 3.4.4.1) as independent variables. To control for sample characteristics that could potentially influence familiarity with smoking cessation pharmacotherapies, the variables identified as potential confounders (section 3.4.4.5) were included in the model. This study could not adjust for the variable ‘number of years worked after obtaining Fellowship (a proxy for work experience)’ because this variable was highly correlated with the variable ‘level of training’.

To address the study’s second aim, only respondents who indicated their familiarity with NRT patches (n=93) were included in the analyses. Chi-square tests were used to examine whether the five facility-level factors (section 3.4.4.1) and the individual-level factors (section 3.4.4.2) were related to intention to prescribe NRT patches to pregnant women. There was no adjustment made for potential confounders due to small cell sizes (n<5) after stratification by the potential confounders.

In cases where responses were missing, the respondent was excluded from the respective analysis.

The results are presented in adherence to the Checklist for Reporting Results of Internet E-Surveys (CHERRIES) [279].

Results The survey was distributed to 2,388 recipients, of which 180 responded, which is equivalent to a response rate of 7.5%. Among respondents, one did not consent to participate, ten did not see pregnant patients, and six had missing responses on more than 50% of questions regarding facility-level factors. After exclusions, 163 obstetricians were included in this study.

3.5.1.1 Comparison of respondents and the overall population of obstetricians Survey respondents were comparable to the overall population of obstetricians in Australia in terms of the level of training (see Table 6). In contrast, they were different in terms of the state of practice, with under-representation of respondents from the Australian Capital Territory (23.9% among survey respondents against 30.9% of RANZCOG affiliates) and South Australia (3.7% vs 7.3%). In contrast, respondents from Victoria (1.2% vs 0.6%) and Western Australia (5.5% vs 2.0%) were over- represented.

Fellows comprised almost three-quarters of the survey respondents. Of these, the majority had general obstetrics as their specialty, and this did not differ significantly between survey respondents and the population of obstetrician Fellows. Fellow respondents from Tasmania and the Australian Capital Territory were over-represented among survey respondents, whereas those from South- and Western Australia were under-represented. This is consistent with the pattern observed in the above-reported comparisons with the overall population of obstetricians. In terms of funding of care, Fellows who practised predominantly in public settings were over-represented among survey respondents.

Table 6 Characteristics of survey respondents and RANZCOG-affiliated obstetricians All respondents Fellows only (Fellows and Trainees) RANZCOG p- RANZCOG p- This study 2 This study 2 Total affiliates value affiliates value N=163 N=2,338 N =163 N =2,338 State/Territory of practice* 16.47 0.02 21.00 <0.001 New South Wales 5 3.1% 51 2.1% 4 3.4% 37 2.0% Australian Capital Territory 39 23.9% 739 30.9% 27 23.1% 565 30.7% Victoria 2 1.2% 15 0.6% 2 1.7% 13 0.7% Queensland 34 20.9% 486 20.4% 27 23.1% 368 20.0% South Australia 6 3.7% 175 7.3% 4 3.4% 141 7.7% Western Australia 9 5.5% 48 2.0% 8 6.8% 41 2.2% Tasmania 46 28.2% 666 27.9% 38 32.5% 512 27.8% Northern Territory 10 6.1% 203 8.5% 5 4.3% 164 8.9% Level of training 0.03 0.92 Trainee 36 22.1% 547 22.7% Fellow 117 71.8% 1841 77.1% Remoteness of practice 6.83 0.01 Major cities 81 69.2% 1482 80.5% Regional and remote 34 29.1% 359 18.9% Funding of care 20.68 <0.001 Both public and private 20 17.1% 619 34.7% Predominantly private 35 29.9% 624 34.2% Predominantly public 59 50.4% 598 31.3% Primary specialty 0.32 0.33 General obstetrics** 97 82.9% 1562 85.0% Subspecialty 20 17.1% 279 15.0% * Percentages do not add up to 100% due to missing responses, **Including gynaecology 84

3.5.1.2 Sample characteristics Table 7 shows additional characteristics of the survey respondents. Almost one-third of the survey respondents (28.2%) were 15 years post-Fellowship. More than half of the respondents practised in public settings (58.3%), and in major cities (69.3%) while 43.6% practised in socio-economically advantaged areas. An almost equal proportion of respondents practised in Victoria (28.2%), New South Wales (23.9%) and Queensland (20.9%).

In terms of their clinical role, the majority spent at least 50% of their weekly work hours seeing patients (77.9%), 52.1% were dedicated to patient care only and 66.3% shared the care with other providers such as GPs and/or midwives.

Table 7 Characteristics of survey respondents Number, Percent* N=163 Number, Percent* State/Territory of practice N=163 Primary specialty New South Wales 39 23.9% General obstetrics** 131 80.4% Australian Capital Territory 5 3.1% Subspecialty 22 13.5% Victoria 46 28.2% Queensland 34 20.9% Level of training Fellow 117 71.8% South Australia 6 3.7% Trainee 36 22.1% Western Australia 10 6.1% Tasmania 9 5.5% Years post-Fellowship <5 38 23.3% Northern Territory 2 1.2% 5-10 33 20.2% Types of practice activity >10 46 28.2% Patient care only 85 52.1% Smoker (including ever) 24 14.7% Patient care and others 68 41.7% Proportion of time spent in seeing patients Funding of care Both public and private 20 12.3% <50% 17 10.4% Predominantly private 35 21.5% 50-100% 127 77.9% Predominantly public 95 58.3% % of pregnant patients who smoke (past 6 months) <5% 60 36.8% Remoteness of practice Major cities 113 69.3% 5-15% 46 28.2% Regional and remote 38 23.3% > 15% 47 28.8% Maternity care models Socio-economic status of practice No shared care 40 24.5% Disadvantaged 29 17.8% Shared care with GP and/or midwives 108 66.3% Average 51 31.3% * Percentages do not add up to 100% due to missing responses Advantaged 71 43.6% **Including gynaecology

3.5.1.3 Smoking cessation support practices and interest As for smoking cessation support practices (see Table 8), approximately half of the respondents reported that their patients were interested in quitting (61.3%, 95% CI 53.9-68.8) and that they had enough time to discuss cessation with their patients (47.9%, 95% CI 40.2-55.5).

Regarding prescribing, almost a quarter (23.9%, 95% CI 17.4-30.5) of the surveyed obstetricians had prescribed NRT patches within the past six months. None had prescribed bupropion or varenicline. Slightly more than half of the sample reported familiarity with NRT patches (57.1%, 95% CI 49.5-64.7), but a lower proportion of them had the intention to prescribe them (36.8%, 95% CI 29.4-44.2). A low proportion of obstetricians were aware of varenicline (8%,95% CI 3.8-12.1) and even fewer intended to prescribe varenicline (<2%, 95% CI 0.4-5.3). No assessment was made regarding familiarity with bupropion.

In terms of interest in smoking cessation pharmacotherapy, slightly more than half (57.7%, 95% CI 50.1-65.3) of the survey respondents were receptive to receiving educational materials regarding smoking cessation pharmacotherapies.

Table 8 Smoking cessation practices and interest of survey respondents Number Percent 95% CI (N=163)

Enough time to discuss cessation 78 47.9% 40.2-55.5 Smoking cessation care protocol is present 63 38.7% 31.2-46.1 Patients are interested in quitting 100 61.3% 53.9-68.8 Familiar with NRT patches 93 57.1% 49.5-64.7 Familiar with varenicline 13 8.0% 3.8-12.1 6-month history of prescribing NRT patches 39 23.9% 17.4-30.5 6-month history of prescribing bupropion 0 0.0% 0.0-0.0 6-month history of prescribing varenicline 0 0.0% 0.0-0.0 Intend to prescribe NRT patches 60 36.8% 29.4-44.2 Intend to prescribe bupropion 2 1.2% 0.3-4.4 Intend to prescribe varenicline 3 1.8% 0.4-5.3 Interested in survey results 26 16.0% 10.3-21.6 Interested in receiving educational materials 94 57.7% 50.1-65.3

Table 9 presents the unadjusted and adjusted association between five facility-level factors and familiarity with NRT patches. After adjusting for potential confounders, this study identified two facility-level factors that were statistically significant associated with familiarity with NRT patches.

Compared to those who reported not having or being unsure about having a smoking cessation care protocol at their workplace, those who reported having such protocol were almost three times more likely to be aware of NRT patches (aOR 2.85, 95% CI 1.29-6.25). Compared to those practising in major cities, those who practised in regional and remote areas were more likely to be aware of NRT patches (aOR 2.56, 95% CI 1.06-6.21).

Table 9 Association between facility-level factors and familiarity with NRT patches among obstetricians Familiar with NRT Yes No OR 95% CI aOR 95% CI (N=93) (N=70) Levels of traininga Fellow 62 72.1% 55 82.1% 1.00 ref 1.00 ref Trainee 24 14.5% 12 17.9% 1.77 0.81-3.88 2.22 0.82-5.98 Smoking cessation care protocol is presenta Disagree/Unsure 43 49.4% 49 72.1% 1.00 ref 1.00 ref Agree 44 50.6% 19 27.9% 2.64 1.34-5.19 2.85 1.29-6.25 Funding of carea Both public and private 12 14.1% 8 12.3% 1.00 ref 1.00 ref Predominantly public 60 70.6% 35 53.8% 1.14 0.43-3.07 0.58 0.14-2.37 Predominantly private 13 15.3% 22 33.8% 0.39 0.13-1.22 1.11 0.35-3.53 Maternity care modelsa No shared care 18 21.2% 22 34.9% 1.00 ref 1.00 ref Shared care with GP and/or midwives 67 78.8% 41 65.1% 2.00 0.96-4.16 0.91 0.31-2.74 Remoteness of practicea Major cities 59 69.4% 54 81.8% 1.00 ref 1.00 ref Regional and remote 26 30.6% 12 18.2% 1.98 0.91-4.31 2.56 1.06-6.21 % of pregnant patients who smoke (past 6 months) a, b <5% 28 32.6% 32 47.8% 1.00 ref 1.00 ref 5-15% 28 32.6% 18 26.9% 1.78 0.82-3.88 1.03 0.37-2.93 >15% 25 29.1% 17 25.4% 2.02 0.92-4.41 0.99 0.34-2.93 Smoking statusa, b Never 72 83.7% 57 85.1% 1.00 ref 1.00 ref Ever smoker (current & former smoker) 14 16.3% 10 14.9% 1.11 0.46-2.68 1.10 0.40-2.99 a Percentages do not add up to 100% due to missing responses b Potential confounders denoted in the grey area

Among respondents who were familiar with NRT patches (n=93, 57.1% (95% CI 49.5-64.7) of all respondents)), 61.3% (95% CI 51.4-71.2) had the intention to prescribe them. Factors related to intention to prescribe varenicline could not be examined due to small cell sizes (<5) after stratification by facility-level factors.

3.5.3.1 Facility-level factors Table 10 presents the relationship between five facility-level factors and obstetricians’ intention to prescribe NRT patches. Intention to prescribe NRT patches did not vary according to level of training, presence of smoking cessation protocol, funding of care, maternity care models or remoteness of practice. A relationship was observed between the funding status of the facility in which obstetricians practised in and intention to prescribe NRT patches, (66.7% predominantly public vs 30.8% predominantly private vs 58.3% both public and private) however, this effect did not reach statistical significance (2=5.75, p=0.06). A difference was also observed between level of training (54.8% Fellows vs 75.0% Trainees) and intention to prescribe NRT patches (2=2.94, p=0.09). This effect, however, did not reach statistical significance.

Table 10 Association between facility-level factors and intention to prescribe NRT patches among obstetricians who were familiar with NRT patches Intend to Do not intend / p- Facility-level factors Unsurea 2 value (N=57) (N=36) n row % n row % Level of training Fellow 34 54.8% 28 45.2% 2.94 0.09 Trainee 18 75.0% 6 25.0% Smoking cessation care protocol is present Disagree/Unsure 25 58.1% 18 41.9% 0.09 0.76 Agree 27 61.4% 17 38.6% Funding of care Both public and private 7 58.3% 5 41.7% 5.75 0.06 Predominantly public 40 66.7% 20 33.3% Predominantly private 4 30.8% 9 69.2% Maternity care models No shared care 9 50.0% 9 50.0% 0.95 0.33 Shared care with GP and/or 62.7% 37.3% midwives 42 25 Remoteness of practice Major cities 33 55.9% 26 44.1% 1.33 0.25 Regional and remote 18 69.2% 8 30.8% a These include 24 unsure responses

3.5.3.2 Individual-level factors Table 11 shows that obstetricians who reported agreement with the following four attitudes were statistically significant more likely to intend to prescribe NRT patches than those who did not hold these beliefs; NRT patches are PBS-listed (69.5% vs 41.4%, 2 =6.42, p=0.01), NRT patches would benefit patients who are light smokers (71.7% vs 47.6%, 2 =6.42, p=0.01), NRT patches would benefit those who had prior failed quit attempts (64.2% vs 14.3%, 2 =6.70, p=0.02), patients are not likely to be compliant with NRT patches (75.7% vs 48.0%, 2 =6.77, p=0.01). In contrast, respondents who found it hard to assess the benefit- risk ratio were statistically significant less likely to intend to prescribe NRT patches (34.7% vs 62.5%,2 =4.22, p=0.04).

Table 11 Relationship between individual-level factors and intention to prescribe NRT patches among obstetricians who were familiar with NRT patches

Do not intend / Intend to Unsure p- Statements about NRT* (N=57) 2 (N=36) value n* n* Attitudes Safer than smoking 48 87.30% 27 77.10% 1.58 0.21 Concerned about safety in certain 12 21.80% 13 37.10% 2.5 0.11 patients Had patients who developed 16 29.10% 5 14.30% 2.62 0.11 adverse events Few patients had co-occurring 54 98.20% 35 100% 0.64 1 medical conditions Continuous dosing is not a barrier 50 96.20% 29 82.90% 4.43 0.06 Patient’s compliance is lacking 28 53.80% 9 25.70% 6.77 0.01 Effective 18 34.60% 10 28.60% 0.35 0.55 May benefit heavy smokers 53 96.40% 32 91.40% 4.7 0.06 May benefit smokers who had prior 52 98.10% 29 82.90% 6.7 0.02 failed quit attempts May benefit light smokers 33 62.30% 13 37.10% 5.33 0.02

Perceived Behavioural Aware that it is PBS-listed 41 77.4% 18 51.4% 6.42 0.01 Hard to assess benefit-risk ratio 6 11.3% 10 28.6% 4.22 0.04 Supportive guidelines 33 63.5% 23 65.7% 0.05 0.83 *Frequency of those who agree (n, %), missing responses were <5% for each statement

Discussion

In this selective sample of 163 obstetricians and gynaecologists who were affiliated with the Australian branch of RANZCOG and providing antenatal care, slightly more than half (57.1%, 95% CI 49.5-64.7) were familiar with NRT patches. As familiarity with medication is essential for any healthcare provider to prescribe or recommend the medication, this estimate indicates that increasing knowledge of NRT among obstetricians practising in Australia is required in order to improve prescribing of NRT to pregnant women.

To inform the design of interventions that address this lack of familiarity, this study examined the factors associated with being familiar with NRT patches. Having a smoking cessation protocol in one’s workplace, and practising in remote and regional areas were identified as being strongly related to familiarity with NRT patches.

Among obstetricians who were familiar with NRT patches, less than two-thirds (61.1%, 95% CI 51.4-71.2) intended to prescribe them. This suboptimal level of intention to prescribe indicates that lack of knowledge is not the only factor that contributes to the limited prescribing of NRT patches to pregnant women. This study had limited power to provide clear evidence of whether the five examined facility-level factors are related to the intention to prescribe NRT patches. While there appeared to be relationships between some facility-level factors and the intention to prescribe NRT patches, the effects were not statistically significant, hence they should be explored further in future studies. Individual-level factors including beliefs relating to a perceived benefit for patients (e.g. those with previous failed quit attempts and light smokers), perceived lack of compliance among patients and perceived behavioural control (e.g. high self- efficacy, believed that it is PBS-listed) were associated with intention to prescribe NRT patches. Understanding how these beliefs are related to prescribing intentions, causally or otherwise, may inform strategies for increasing prescribing among obstetricians who are familiar with NRT patches but do not intend to prescribe them to pregnant women.

Few obstetricians were familiar with varenicline (8.0%), and even fewer intended to prescribe varenicline (1.8%) or bupropion (1.6%). These low estimates indicate that in

order to increase the uptake of these pharmacotherapies by obstetricians, improving the knowledge base regarding these pharmacotherapies is likely to be needed first.

3.6.1.1 Implications of the findings regarding familiarity with NRT patches The finding that the presence of a smoking cessation care protocol was related to obstetricians’ familiarity with NRT patches is consistent with prior studies finding that the presence of smoking cessation care protocol improved the delivery of smoking cessation interventions among obstetric care providers (physicians, midwives, physician assistants) [216, 280] and improved the likelihood of prescribing NRT among obstetricians and GPs [219]. Further research is now needed to examine the effect of having a protocolised smoking cessation care on prescribing behaviour for smoking cessation pharmacotherapy during pregnancy. Research that examines the important elements of such protocols may also be required. Due consideration will also need to be given to the possibility that obstetric care providers may be held legally responsible should adverse outcomes arise when they prescribe smoking cessation pharmacotherapies to pregnant women.

Obstetricians practising in regional and remote areas were more likely to be familiar with NRT patches. This is not accounted for by the model of maternity care and funding of care associated with practising in these locations, because adjustment for these features was made in the analysis. This finding may, thus reflect other underlying factors that drive the observed relationship. In rural and remote areas, 51.3% of pregnant women smoke relative to 7.2% of pregnant women in major cities [26]. Access to healthcare is also reduced for women who live in rural and remote areas in Australia [281, 282]. Therefore, the most likely explanation is that obstetricians who practised in rural and remote areas tend to acquire more generalist skills than their counterparts in urban and major cities [283]. Pregnant smokers who live in rural and remote areas have limited access to other smoking cessation support services, including those provided by the local hospital and community support [284] As such, obstetricians who provide care for these pregnant women may be driven to equip themselves with the necessary skills and knowledge to provide smoking cessation support that their patients are less able to obtain elsewhere.

As such, although the study observed that remoteness of practice was associated with familiarity with NRT patches, remoteness of practice is not a factor that is amenable to modification through an intervention. This calls for further research to establish the underlying mechanism that drives the observed differences between obstetricians practising in rural and urban areas, so that these may be the focus of future interventions.

3.6.1.2 Implications of the findings regarding the intention to prescribe NRT patches Among obstetricians who were familiar with NRT patches, the findings from this study are suggestive of potential relationships between three of the studied facility-level factors (level of training, funding of care) and the intention to prescribe NRT patches. Larger scale studies are needed to confirm whether such relationships exist, especially as studies carried out among RANZCOG affiliates have returned inconsistent results regarding the influence of facility funding (public vs private) towards intention to carry out clinical practices [285, 286].

Among obstetricians who were familiar with NRT patches, individual-level factors associated with the intention to prescribe NRT patches included the beliefs that NRT patches are likely to benefit patients who are light smokers and those with prior failed quit attempts. An effect, although not statistically significant, was also observed for obstetricians who perceived that NRT patches would benefit their pregnant patients who are heavy smokers. Together, the findings fit with evidence that believing that NRT is effective influences the likelihood to prescribe NRT [215, 219]. Should this finding be confirmed, it suggests that dissemination of information regarding the profile of pregnant women who may benefit from using NRT patches should be evaluated as a strategy to improve the prescribing of NRT patches by obstetricians.

In contrast, safety-related beliefs did not appear to be related to obstetricians’ intention to prescribe NRT patches. This finding is inconsistent with research that found safety concerns were the main barrier for not prescribing NRT among GPs and obstetricians [132, 215, 217, 220]. This disparity with the aforementioned studies may be explained by the fact that the current study was restricted to obstetricians who were familiar with NRT. It may be that safety concerns are cited as a barrier to prescribing among those

who do not possess adequate knowledge of NRT. Further exploration of this possibility will provide insight whether interventions to increase intentions to prescribe NRT will need to address the safety of NRT, or if this measure may no longer be necessary once obstetricians have been informed about NRT.

The finding that self-confidence was related to intention to prescribe NRT patches is consistent with the findings from other studies that self-efficacy has a strong influence on prescribing [257]. However, this finding needs to be confirmed in future studies with adjustment made for potential confounders to establish whether it represents a causal relationship. Future research should also explore the role of prior prescribing and training in smoking cessation, as these are likely to influence self-efficacy to prescribe smoking cessation medications. If this finding is confirmed, it suggests future interventions that have a focus on boosting obstetricians’ self-efficacy to prescribe NRT to pregnant women, might be successful.

3.6.2.1 Intention to prescribe among those who are familiar with smoking cessation pharmacotherapy It appears that familiarity with NRT patches did not always translate to the intention to prescribe NRT patches. Of those who were familiar with NRT patches, slightly more than half (61.1%, 95% CI 51.4-71.2) intended to prescribe it to pregnant women. This confirms the argument made in the introductory section of this chapter that enhancing knowledge alone may not be adequate to influence a change in the prescribing behaviour of smoking cessation pharmacotherapies to pregnant women. This then raises the question of what obstetricians know and how the knowledge is used to guide obstetricians’ prescribing behavior. These questions should be the focus of future studies.

3.6.2.2 The proportion of obstetricians’ prescribing smoking cessation pharmacotherapies This study did not set out to measure the prevalence of awareness of, intention to prescribe, and actual prescribing of smoking cessation pharmacotherapies to pregnant women. Nonetheless, the findings in relation to these measures are informative.

Within this highly selected population of surveyed obstetricians, slightly more than half were aware of NRT patches (57.1%, 95% CI 49.5-64.7), one-third had the intention to prescribe them (36.8%, 95% CI 29.4-44.2), and one-quarter prescribed them in the past six months (23.9%, 95% CI 17.4-30.5). These figures are likely to overestimate the population prevalence of these behaviours because compared to the broader population of obstetricians, the respondents to this survey appear to have greater knowledge of, and are more supportive of, pharmacotherapies for smoking cessation. This supposition is made based on the high observed proportion reporting favourable tobacco cessation practices (patients’ interest to quit and sufficient time to address cessation) as well as the high level of interest in receiving educational materials on smoking cessation pharmacotherapy. Nonetheless, with respect to NRT, the findings are in the realm of prior studies, all of which suggest that awareness of, intention to prescribe, and actual prescribing of NRT, are suboptimal. Price and colleagues (2006) reported that 6% of Ohio’s obstetricians prescribed NRT in the past 6 months in a state-wide survey (response rate of 44%) [215]. In a 2000 survey of obstetricians in Boston (response rate 50%), 44% of respondents were likely to prescribe NRT [220]. In a 2018 survey of Australian obstetricians (inclusive of GPs), 62.1% prescribed NRT (response rate 6.2%) [219].

This study, for the first time, provides evidence on obstetricians and gynaecologists prescribing of varenicline and bupropion to pregnant women. In contrast to the figures reported for NRT patches, a lower proportion of obstetricians were familiar with varenicline (8.0%), intended to prescribe varenicline (1.8%) or intended to prescribe bupropion (1.6%). On the one hand, the lack of knowledge regarding both pharmacotherapies may limit their intention to prescribe (<2%) and actual prescribing to pregnant women (0%). This potentially explains the low smoking cessation pharmacotherapy utilisation rates among pregnant women in Australia [110]. On the other hand, the study findings are reassuring as bupropion and varenicline are currently not recommended for use during pregnancy. While the lack of awareness

towards these pharmacotherapies occurred within the context of prohibitory guidelines, these guidelines may be revised in the future to reflect emerging evidence regarding the safety and effectiveness of these pharmacotherapies during pregnancy (as explained in Section 1.1.3.5.3). Further research is, therefore, necessary to evaluate the best strategies to increase the knowledge base for bupropion and varenicline among obstetricians.

The low response rate of 7.5%, and the observed differences in the characteristics of survey respondents and the general population of obstetricians practising in Australia are suggestive of potential non-response bias in this study. The likely non- representativeness of this sample of obstetricians practising in Australia limits the generalisability of the study findings.

As the survey was designed to allow ease of access and ease of completion, both of which aimed to maximise the response rate [287, 288], this survey achieved a slightly higher response rate than that for the paper-based survey of RANZCOG affiliates and GP obstetricians that investigated factors associated with prescribing NRT in pregnancy (6.2%) in 2015 [219]. Yet, it is lower than other published surveys carried out among RANZCOG affiliates (15-35%) [289-291] and WHA affiliates (46.6-67%) [292-294]. There are several likely explanations for the survey’s low response rate. Its implementation including the lack of personalised contact and one-time survey reminder, and the saliency of the survey topic (tobacco cessation) could all potentially contribute to poor response rates, all of which are reported in a review of best practice strategies to employ in designing surveys of healthcare providers [295]. That the saliency of the survey topic (tobacco cessation) is a likely explanation for the survey’s low response is supported by the response rates to other recent surveys carried out among RANZCOG affiliates. The following survey topics reported at least three-fold higher response rates than this study’s response rate; vaginal delivery (response rate 31.7%) [285], labour management (response rate of 29%), gynaecological surgery (26%) [296], abortion (response rate of 25%) [291], infertility (response rate of 25.7%) [297] and adherence to antenatal corticosteroid guidelines (response rates of 20%) [298].

In terms of internal validity, there were five main limitations. First, the cross-sectional design of the study did not allow for establishing neither temporal nor causal relationships between the study factors and outcomes. Second, while the self-report intention to prescribe is a good predictor of the prescribing behaviour, it may not reflect actual prescribing.

Third, this study had a limited sample size of 163 survey respondents. Wide confidence intervals around the measures of association between facility-, and individual-level factors, and the intention to prescribe smoking cessation pharmacotherapies, indicate that the study was not able to measure these relationships precisely. A priori sample size calculations may have provided insight into the feasibility of achieving this aim, and/or may have prompted the implementation of recruitment boosting techniques. This led to inconclusive findings regarding some relationships that appeared to exist but did not reach statistical significance. The non-significant effects were observed for relationships between facility-level factors (facility settings and level of training) and individual-level factors (beliefs regarding NRT efficacy), and the intention to prescribe NRT patches. The limited sample size also prevented an examination of one of the study’s initial aims; to examine facility- and individual-, level factors associated with the familiarity with, and intention to prescribe, varenicline. This small sample size did not allow for adjustment of potential confounders in the analyses of factors relating to intention to prescribe NRT patches. In particular, gender is likely to be important confounder as many physician-based studies have observed that gender is related to knowledge and attitude regarding clinical practices [219, 299]. However, questions relating to gender was not included in the survey at RANZCOG’s request (see footnote in Table 5). An additional implication of the low response rate is that a relatively small number of respondents selected the ‘unsure’ category on many survey items, which did not allow for a separate investigation of this response. In the analyses, these respondents were pooled with those who selected “no”. The combined responses may have introduced heterogeneity into this comparison group, resulting in difficulty to observe any differences between those who intend to, and those who do not intend to prescribe smoking cessation pharmacotherapies.

Fourth, this study could not adjust for the potential effects of clustering of obstetricians who practised in the same institution. To ensure the confidentiality and anonymity of each survey respondent, information on their institution details was not included in the

survey. This may lead to the over- or under-estimation of the effect of the presence of a smoking cessation care protocol for obstetricians who worked within the same facility.

Fifth, not all potentially relevant questions were included in the survey. The ideal method to inform the selection of survey items is to conduct qualitative interviews and focus groups [300], and thatwas not carried out in this study. In particular, information on obstetricians’ training in delivering smoking cessation was not collected. The most contemporary review of nine randomised trials did not find evidence that training health professionals in smoking cessation influences the provision of NRT (pooled OR 1.57, 95% CI 0.87-2.84) [301].

Obstetricians and gynaecologists are a key group of obstetric care providers to target for improving smoking cessation pharmacotherapy prescribing practices. This study revealed several factors associated with familiarity with, and intention to prescribe, NRT patches to pregnant women. The finding that having a smoking cessation protocol was associated with being familiar with NRT patches indicates a potential need to institute a protocolised smoking cessation care in facilities to improve the knowledge base regarding pharmacotherapies for smoking cessation. This may not, however, result in a corresponding improvement in the prescribing of these pharmacotherapies among obstetricians, as this study also observed a suboptimal level of intention to prescribe NRT patches among obstetricians who were familiar with them. Future interventions that focus on the important elements of such protocols are required. Individual-level factors related to the intention to prescribe NRT patches included perceived benefits for pregnant patients who are light smokers and those with previously failed quit attempts, being aware that NRT patches is PBS-listed and having the self-confidence to prescribe NRT. If these findings are confirmed in future studies that allow for causal inference to be made, they suggest that interventions that focus on improving the perceived benefits of NRT patches and boosting obstetricians’ self- efficacy to prescribe NRT to pregnant women, might be successful in increasing the prescribing of NRT patches by obstetricians. This study also identified other potential factors to explore in future research that aims to identify areas of focus for improving the prescribing of smoking cessation pharmacotherapies among obstetricians. In particular, the findings were suggestive of a relationship between the obstetricians’

100

level of training, funding of care and remoteness of practice, however, the limited sample size prevented clear conclusions being drawn.

101

Chapter 4 Use of smoking cessation pharmacotherapies in the 12 months after giving birth Introduction The studies described in Chapters 2 and 3 addressed the first aim of the thesis, which is to examine factors that influence the prescribing of smoking cessation pharmacotherapies to pregnant women. The previous chapter found a range of additional factors that influence obstetricians’ prescribing of smoking cessation pharmacotherapies to pregnant women. However, addressing these prescribing-related factors will not completely overcome the problem of smoking during pregnancy. Section 1.2.2 (Prevalence of smoking after pregnancy) illustrated that many of women who smoke late in pregnancy are likely to smoke after giving birth. This calls for seizing opportunities to address smoking after giving birth.

Worldwide, 40% of infants and children are exposed to second-hand smoke [152]. Exposure to second-hand smoke is known to be an important cause of preventable infant morbidity and mortality [143, 302]. The health risks of infants’ exposure to second-hand smoke are described in section 1.2.1 (Health risks of smoking after pregnancy). Briefly, they include an increased risk of sudden unexpected death [148], respiratory infections (wheezing, asthma, bronchitis, pneumonia) [149], and hospitalisation due to fire-related injuries [150]. Infants exposed to second-hand smoke face an increased risk of developing potentially life-long negative health consequences such as ischaemic heart disease [151], obesity [152] and long-term neurological morbidity [153]. Moreover, these infants are twice as likely to take up smoking in their later years [154]. Therefore, reducing second-hand smoke exposure is a global health priority.

Maternal smoking is one of the primary sources of infant exposure to tobacco smoke. Women who smoke at the time of delivery are more likely to smoke during their child’s infanthood and during a subsequent pregnancy [303, 304]. As described in section 1.2.3.2 (Barriers to smoking cessation after pregnancy), smoking cessation after pregnancy is likely to be challenging for women who smoke throughout their pregnancy [29]. Given the inconsistent evidence on whether psychosocial interventions are effective for women smoking during the postpartum period (section 1.2.3.3 Interventions for smoking cessation after pregnancy in healthcare settings) and the known effectiveness of smoking cessation pharmacotherapies in the general

102

population (section 1.3.2 Efficacy in the general population), there is value in exploring the use and potential benefit of smoking cessation pharmacotherapies for women smoking postpartum.

Studies in the US [104] and the UK [107] found that 2% and 5% of women smoking after pregnancy, respectively, used smoking cessation pharmacotherapies. Obtaining information on the extent to which smoking cessation pharmacotherapies are used by women who smoke after pregnancy in Australia will be useful in informing the extent to which efforts to modify smoking cessation pharmacotherapies uptake among postpartum women are required.

It is also important to understand how soon after delivery women use smoking cessation pharmacotherapy. Early postpartum use of smoking cessation pharmacotherapy should be supported because it potentially minimises infant exposure to second-hand smoke during a vulnerable stage of development [305]. Moreover, regular contacts between women and health services for postpartum and postnatal care offers an opportunistic window for maternal and infant care providers to intervene [306]. Australian postnatal care guidelines recommend that a woman and her newborn see health professionals at least twice within the first 8 weeks postpartum, and every six months thereafter, until the child reaches two years of age [171, 307].

Information on the maternal health conditions associated with the use of smoking cessation pharmacotherapies will be valuable. As summarised in section 1.2.3.1 outlining the health risks of smoking after pregnancy, the following health conditions are worsened by continuing smoking: psychiatric, cardiovascular and pulmonary disorders [157, 302]. The suitability of smoking cessation pharmacotherapies is increased in women with these morbidities, as the potential benefits of using the pharmacotherapy are substantially magnified. In contrast, among women with pre- existing medical conditions in which pharmacotherapy use is cautioned or contraindicated (as detailed in section 1.3.2), the risks associated with pharmacotherapy use are likely to outweigh the benefits. Thus, the suitability of these pharmacotherapies is reduced among women with pre-existing medical conditions. Likewise, for women with mental health conditions, nicotine withdrawal may exacerbate mental health symptoms [308], and therefore, these pharmacotherapies are less suitable for women with mental health conditions. Understanding whether the use of

103

smoking cessation pharmacotherapies differ according to the presence of each of these health conditions will provide insight into whether the suitability of smoking cessation pharmacotherapies is considered among women smoking after pregnancy.

It is also useful to understand the extent of smoking cessation pharmacotherapy use in women who experienced poor birth outcomes. This will inform whether poor birth outcomes are used as a motivator to engage maternal smokers in quitting after pregnancy. Prior studies found that maternal smokers who gave birth to infants with smoking-related complications were equally as likely to smoke during their subsequent pregnancy [303, 309, 310]. Whether the same pattern applies to attempts to quit smoking using a pharmacotherapy after delivery has not been examined.

Aim This study has three aims. The first aim is to estimate the prevalence of utilisation of each smoking cessation pharmacotherapy in the 12 months after giving birth among women who smoke after pregnancy, as well as the timing of the utilisation. Second, this study examines the relationship between selected maternal morbidities and the use of each cessation pharmacotherapy in the 12 months after giving birth. Third, this study investigates whether poor birth outcomes are associated with the use of each pharmacotherapy in the 12 months after giving birth.

Methods

This study was based on the data linked for the Smoking MUMS (Maternal Use of Medications Safety) Study, which measured the utilisation, effectiveness and safety of smoking cessation pharmacotherapies in pregnancy [311]. Details of the data sources and linkages for the Smoking MUMS study are described elsewhere [312]. Briefly, the Smoking MUMS study linked perinatal records, hospital admission records, death records and pharmaceutical dispensing records for all women who gave birth in NSW and Western Australia (WA) between 2003 and 2012.

104

The perinatal data comprises records from the NSW Perinatal Data Collection (PDC) and the WA Midwives Notification Scheme (MNS), and include all births between 2003 and 2012. These are legislated data collections covering all live and stillbirths of at least 20 weeks gestation or at least 400g birth weight. The perinatal data do not contain information on pregnancy losses occurring earlier than 20 weeks gestation. Data are collected at birth by the attending midwife or doctor, and they include information relating to maternal demographics, smoking status, medical conditions, obstetric, and neonatal outcomes.

The hospital data consists of the NSW Admitted Patient Data Collection (APDC) and WA Hospital Morbidity Data Collection (HMDC). All admissions and discharges between 2001 and 2012 from all public, private and repatriation hospitals, and private day procedures are included. The APDC and HDMC are statutory data collections. The diagnoses are coded according to the International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Australian Modification (ICD- 10-AM) [313]. The hospital procedures are coded according to the Australian Classification of Health Interventions, Eighth Edition (ACHI) [314].

A delivery of a singleton birth generates three records: (1) one birth record in the perinatal data; (2) one admission record in the hospital data for mother; and (3) one admission record in the hospital data for the baby. This study refers to the maternal admission record associated with the delivery as the delivery admission record.

The death data included all registered deaths in NSW and WA between 2003 and 2014.

105

The pharmaceutical dispensing data contain records of prescription pharmaceutical products subsidised through the PBS and dispensed between 2003 and 2013. The PBS data include records for dispensed pharmaceuticals in community pharmacies and private hospitals. Each PBS record contains the following information: product name, Anatomical Therapeutic Chemical (ATC) code, strength of therapeutic ingredient, quantity dispensed, prescribed and dispensed dates, and the patient’s age, sex and beneficiary status. The dispensing data do not have information on pharmaceuticals dispensed during public hospital stays. As mentioned in Chapter 1 (section 1.3.4 Cost), the PBS allows all eligible residents in Australia to have access to subsidised prescription medications. Approximately 80% of dispensed medications in Australia are subsidised by the PBS [315]. Patients who hold concession cards, ‘concessional beneficiaries’ have lower annual co-payment threshold than the rest of the population, ‘general beneficiaries’, that is $6.50 vs $40.30 (as at 2019) [72]. Concessional beneficiaries are individuals who receive income support, supplements or discounts from the Australian government [316]. The majority of concessional beneficiaries aged 16-44 years receive this support due to unemployment, and parenting or carer responsibilities [316]. If the dispensed price of a medication falls below the co-payment threshold, the beneficiaries pay the full price and are not entitled to a subsidy. Until the Australian government started collecting under-copayment data in July 2012, many medications dispensed to general beneficiaries were not recorded in the PBS data, resulting in the PBS data used in this study containing complete dispensing data for concessional beneficiaries only [317]. The cost of varenicline, bupropion, and NRT patches exceeded the co-payment threshold for both concessional and general beneficiaries. As such, all dispensings of subsidised smoking cessation pharmacotherapies are captured in the PBS data, irrespective of the patient’s beneficiary status. As detailed later (section 4.3.4), dispensing records of other medications apart from varenicline, bupropion and NRT patches, were used to identify maternal morbidities. As the cost of some of these medications was lower than the co- payment threshold for general beneficiaries, complete data capture for these medications was available for concessional beneficiaries only. Due to data custodian- imposed restrictions, the dispensing data approved for use in the Smoking MUMS Study did not include all, but a comprehensive range of medications.

106

The perinatal data were linked to hospital and death data by the NSW Centre for Health Record Linkage (CHeReL) and the WA Linkage Branch. Linkage of the perinatal and dispensing data was performed by the Australian Institute of Health and Welfare (AIHW). Probabilistic linkage methods and a best practice protocol for preserving privacy were applied during the linkage process [318]. False positive and false negative rates for the NSW linkage were <0.3% and <0.1% [319], respectively, whereas for WA linkage, both were estimated as 0.11% WA [320].

Data preparation and cleaning for the Smoking MUMS study have been described elsewhere [312]. Briefly, the authors combined data across two jurisdictions and used stepwise methods to identify and correct inconsistent data values [312]. The majority of these errors were due to linkages carried out by different linkage units, and women who had records in both states. The data cleaning methods ensured that each linked record was unique, and that values were consistent within and between records relating to each pregnancy and woman [312].

The cohort of women who delivered between 1 February 2011 and 31 December 2012 in NSW and WA, and who were current smokers at the time of delivery were identified using information recorded in the perinatal, hospital, and death datasets. The start date of the cohort entry was chosen because all subsidised smoking cessation pharmacotherapies were available to the whole population from 1 February 2011. Inclusion of deliveries until December 2012 allowed for one year of follow up in death data and dispensing data.

Figure 5 presents the selection of the study cohort with the steps outlined in the following text.

First, all delivery admission records of a woman were sorted in ascending order of dates of delivery.

Second, delivery admission records containing information indicating that a woman smoked at the time of delivery were identified. This was based on the presence of the

107

ICD-10-AM diagnosis code Z72.0 [313] (any use of tobacco in the past 28 days) in any diagnosis field of the delivery admission record.

Third, women with no smoking diagnosis recorded in any delivery admission records and prior to index smoking pregnancy were excluded.

Fourth, pregnancies were excluded if the woman died or became pregnant within 365 days of delivery. This required the ascertainment of date of conception for each pregnancy. Information on the date of delivery and weeks of gestation are recorded in the perinatal data. The date of conception was calculated as: date of delivery- gestational age at delivery x 7 + 14 days [311, 321]. The death date was ascertained from the death dataset.

Fifth, for women who had more than one delivery admission record during the study period indicating she smoked at the time of delivery, the earliest pregnancy (referred to as the index pregnancy), was selected for the analyses.

To address the second and third aims of this study, the study population was restricted to concessional beneficiaries who gave birth in NSW. As described earlier (section 4.3.1), concessional beneficiaries had complete capture for all dispensed medications, thus allowing for potentially complete ascertainment of maternal morbidities. Due to data-custodian imposed limits, the dispensing data for women giving birth in NSW included a wider range of medications than the dispensing data for women giving birth in WA. This study defined a woman as a concessional beneficiary if she had at least one dispensing record as a concessional beneficiary, and no records as a general beneficiary in the one year prior to conception and throughout pregnancy to birth.

108

Figure 5 Selection of women who smoked at delivery, NSW and WA

1,495,401 pregnancies between 1 January 2003 and 31 December 2012

247,080 pregnancies between 1 February 2011 and 31 December 2012

225,909 women (232,234 pregnancies) were excluded due to no smoking diagnosis

14,846 pregnancies belonging to 14,171 women who had at least one pregnancy with a smoking diagnosis

344 pregnancies with a smoking diagnosis were excluded 11 maternal death < 365 days after delivery 334 subsequent pregnancy < 365 days after delivery

14,502 pregnancies belonging to 14,160 women who had a pregnancy with a smoking diagnosis, and at least 365 days follow-up

341 subsequent pregnancies were excluded after first smoking diagnosis 23 with smoking diagnosis 318 with no smoking diagnosis

14,160 women with smoking diagnosis in their index pregnancy and at least 365-day follow-up 5,377 concessional beneficiaries who gave birth in NSW

109

A woman was defined as using smoking cessation pharmacotherapy in the 12 months after delivery if she had at least one dispensing of NRT patches, bupropion or varenicline within 365 days after the date of delivery for the index pregnancy.

Anatomical Therapeutic Chemical (ATC) codes were used to identify dispensed medications [322]. The ATC code for varenicline is N07BA03, for NRT patches is N07BA01 and for bupropion is the older N07BA02 because, in 2009, bupropion’s ATC code changed from N07BA02 to N06AX12 [323, 324]. In Australia, bupropion is not registered for use as an antidepressant [322].

In women who were dispensed with more than one smoking cessation pharmacotherapy in the 365 days after delivery, the first smoking cessation pharmacotherapy dispensed after the index pregnancy was identified. The date of the first dispensing was referred to as the date of use.

Maternal morbidities were selected from a list of health conditions described by the WHO Maternal Morbidity Working group [325], guided by three criteria: (1) the condition is worsened by smoking (benefit associated with pharmacotherapy use assumed to outweigh the risk); (2) use of one of the smoking cessation pharmacotherapies is contraindicated or cautioned against (risk assumed to outweigh the benefit); and (3) the condition is exacerbated by nicotine withdrawal (risk assumed to outweigh the benefit). Two conditions, tuberculosis and cancer, were not examined due to low prevalence in the study population. Information on obesity, body mass index and dermatoses aggravated by pregnancy was not available. Whether the remaining maternal morbidities met one of the three abovementioned criteria were identified from the literature on health complications due to smoking (as described in section 1.2.1 Health risks of smoking after pregnancy).

110

Table 12 summarises the health conditions worsened by smoking and the conditions for which smoking cessation pharmacotherapies are cautioned or contraindicated (as detailed in Table 3 of section 1.3.1 Clinical pharmacology). Table 12 also includes the health conditions potentially exacerbated by nicotine withdrawal.

111

Table 12 Maternal morbidities based on reasons for selection Health conditions Health conditions exacerbated by nicotine withdrawal or associated with worsen by adverse effects of smoking cessation pharmacotherapy smoking Exacerbated by nicotine Smoking cessation Smoking cessation Maternal morbidities withdrawal pharmacotherapy pharmacotherapy cautioned contraindicated

Substance use disorder (drugs and √ alcohol) [326] Diabetes mellitus [107] √ Hypertension [107] √ Blood disorders, including anaemia √ and clotting disorders [327] Mood disorders [328-330] √ Postpartum depression Varenicline use is [331, 332] and suicidal cautioned in patients with behaviour [330] a history of suicidal ideation and behaviour [195] Anxiety disorders [328, 333] √ Anxiety symptoms [54] Varenicline use is cautioned in patients with a history of anxiety [195] Psychosis disorders [329, 334] √ Postpartum psychosis Varenicline use is [334] cautioned in patients with a history of hallucination and psychotic episodes [195] Respiratory disorders (Inflammatory& √ immunological) [107, 335] GORD [336-338] √ Varenicline is cautioned in patients with GORD due to nausea-associated side effects of varenicline [205]

112

Health conditions Health conditions exacerbated by nicotine withdrawal or associated with worsen by adverse effects of smoking cessation pharmacotherapy smoking Exacerbated by nicotine Smoking cessation Smoking cessation Maternal morbidities withdrawal pharmacotherapy pharmacotherapy cautioned contraindicated

Musculoskeletal & connective tissue √ Varenicline and NRT are disorders, treated by NSAIDS [339] cautioned among NSAIDS users due to increased, reversible risk of cardiovascular events Epilepsy [340] √ Varenicline is cautioned in patients with a history of seizures [205] Steroid-responsive disorders [339] √ Thyroid disorders (hypo- and √ hyperthyroid) [341] Autoimmune disorders [339] √ Acute and chronic kidney disease √ Varenicline is cautioned in [342] patients with impaired renal function [205] Cardiovascular diseases [343] √ Varenicline is cautioned in patients with a history of cardiovascular disease [205] NRT is cautioned in patients with a recent (less than 2 weeks) acute cardiovascular event, unstable or worsening angina, severe cardiac arrhythmia [202]. GORD: Gastro-intestinal oesophageal reflux disease NSAIDs: Non-steroidal anti-inflammatory drugs

113

To identify maternal morbidities, this study used hospital, perinatal and PBS data. A binary variable indicating the presence of each maternal morbidity (‘yes’ or ‘no’) was created.. When using hospital admission records, this study carried out a search for relevant diagnoses and procedures recorded in all available fields. A 365-day lookback period (that is, 365 days prior to the date of conception) was used to increase the ascertainment of maternal chronic conditions [344]. There are several chronic conditions (including chronic bronchitis, chronic congestive heart failure) which are commonly managed in non-hospital settings, and thus they were possibly not recorded in the hospital data until the women delivered or were admitted during pregnancy. For these conditions, this study used admission records occurring in both the lookback and gestation periods. Perinatal data were used to supplement the identification of pre- existing diabetes [345] and hypertension [346, 347] because these two morbidities were also recorded in the perinatal records (in checkboxes). Based on the validated RxRisk-V tool [348], this study used the ATC codes of medications to the study cohort to further identify maternal morbidities. All the codes for diagnoses and procedures that were looked up in the hospital data as well as the ATC codes searched for in the dispensing data are listed in Appendix 4A Ascertainment of pre-existing maternal morbidities from hospital and PBS dispensing data).

In this study, poor birth outcomes comprised maternal adverse events and neonatal adverse events.

Maternal adverse events comprised severe morbidity outcomes at delivery (MMOI), preterm premature rupture of membrane (PPROM), placental abruption and intrapartum Caesarean delivery. The MMOI is a composite morbidity indicator developed for routinely collected datasets [349] derived from the individual measures of 14 morbid complications and 11 procedures occurring at delivery (Appendix 4B Components and ICD10 codes for Maternal Morbidity Outcome Indicator (MMOI), adapted from Roberts et al. (2008), [349]). The measures were obtained using the information in the delivery admission records. This study examined PPROM and placental abruption because they are smoking-related complications [350] and therefore, they might motivate a woman to attempting quit smoking after giving birth. These complications were not included in the MMOI. PPROM was identified using the ICD-10-AM code O42 recorded in the delivery admission records, subject to the gestational age < 37 weeks. Placental abruption was identified using ICD-10-AM code

114

O45 recorded in delivery admission records. This study considered emergency caesarean section delivery as an adverse event because of the increased risk conferred [351, 352]. Planned caesarean delivery and instrumental delivery were not considered as such. For NSW, emergency Caesarean section delivery was derived from mode of delivery (that is, caesarean section) and onset of labour (that is, spontaneous or induced) in perinatal records. For WA, emergency caesarean section delivery was recorded as it is in perinatal records.

Neonatal adverse events included severe neonatal adverse outcomes indicator (NAOI) [353], small for gestational age (SGA)[15], preterm birth [14], Apgar score less than seven at five minutes [354], admission to neonatal special care [26] and stillbirth [9]. Severe NAOI is a composite measure based on 15 morbid complications and seven procedures carried out at delivery (Appendix 4C Components and ICD10 codes for Neonatal Adverse Outcome Indicator (NAOI), adapted from Lain et al. (2012)) [353]. This information was derived using records from perinatal data, and the admission records in the hospital for the baby, and subsequent transfers to other hospitals before first discharge home. Preterm birth is defined as birth prior to 37 weeks of gestation, whether medically indicated or spontaneous [355]. Births prior to 32 weeks’ gestation are included in the NAOI [353]. Preterm birth was estimated using the gestational age at delivery and onset of labour (both recorded in the perinatal data). Using the Australian birthweight percentile standards [356], SGA is defined as birthweight below the 10th percentile, adjusted for gestational age and sex for a singleton neonate. Stillbirth was defined as neonatal death occurring after 20 weeks of gestation or of at least 400g of birthweight whether intrapartum or antepartum death [357]. Stillbirth was derived from the variable ‘status of baby discharge’ (that is, stillbirth) recorded in the perinatal data. Apgar score at five minutes indicates the clinical condition of a baby five minutes after giving birth [358]. It is an aggregate score of observed neonatal characteristics: skin colour, heart rate, respiratory effort, muscle tone and reflexes [358]. A 5-minute Apgar score of less than seven indicates health-related complications [354]. Admission to neonatal special care is defined as receipt of specialised medical care for neonates than is available on the postnatal ward [26]. Information on both Apgar score and admission to neonatal special nursery care were obtained from the perinatal data.

115

Figure 6 Graphical representation of maternal morbidities and poor birth outcomes that could potentially motivate a woman who smoked at delivery to using smoking cessation pharmacotherapy in the 12 months postpartum

Potential confounders Maternal demographics Smoking characteristics History of substance use Socio-economic determinants Reproductive health

Poor birth outcomes Medical morbidities in maternal smokers Maternal adverse events Attempting to quit using Pre-existing, worsened by smoking cessation pharmacotherapy Medication-related (caution/contraindicated) Neonatal adverse outcomes

The arrows indicate the main, but not exclusive, direction of potential causality, with red arrows denote the potential of bidirectional causality

*adapted from the 2016 WHO maternal morbidity model [325]

116

This study adjusted for characteristics that could potentially influence the postpartum use of smoking cessation pharmacotherapy postpartum. They included maternal socio- demographics (maternal age at delivery, maternal country of origin, marital status, country of birth, Aboriginal or Torres Strait Islander status, socioeconomic status of residence, remoteness of residence, parity and nicotine dependence as shown in Figure 6. The number of cigarettes smoked after the 20th week of gestation was used as a proxy for nicotine dependence, as the number of cigarettes has been shown to be a reliable predictor for nicotine dependence [359].

Maternal age at delivery was numerically recorded as it is (in years) in the perinatal records. This was later grouped into < 25, 25-<35 and > 35 years. A woman was identified as having Indigenous status if a “yes” was recorded for Aboriginal and Torres Strait Islander status in any of her perinatal or hospital records. Based on her perinatal record, a woman’s country of birth was identified. If this information was missing or inadequately described in this perinatal record, it was supplemented by information recorded in the woman’s delivery admission record. This variable was categorised into Australia-born and overseas-born. In NSW, information on women's marital status was available in the hospital data only. For WA, although this information was available for both hospital and perinatal data, perinatal records was used instead because of minor inconsistency between these two data sources. This variable was subsequently categorised as living with a partner (married, de-facto relationship) or not living with a partner (never married, widowed, divorced, separated). The perinatal data defined parity as the number of previous pregnancies greater than 20 weeks and numerically coded parity as 0, 1, 2, 3 and higher. This study grouped parity into nulliparous (parity=0), multiparous (parity 1 to 4) and grand multiparous (parity >=5) [360].

117

The statistical local area of residence (SLA) identified from the perinatal data was mapped to an area-based measure of: (a) socioeconomic status (SES), using Socioeconomic Indexes for Areas Index of Relative Socioeconomic Disadvantage (IRSD) scores [228] and (b) remoteness of residence, classified according to the Accessibility Remoteness Indicator of Australia [361]. Similar to the approach described in Chapter 2 and Chapter 3, based on the data from the 2006 Australian Census of Population and Housing, the IRSD scores were grouped into the following categories: disadvantaged SES (<40th percentile), average SES (40-79th percentile), and advantaged SES (>=80th percentile) [228]. Based on the data from the 2006 Australian Census of Population and Housing, the remoteness of residence was grouped as major cities, inner regional, outer regional, remote and very remote [361]. This study grouped remoteness of residence into major cities and, regional and remote (inner regional, outer regional, remote and very remote) because of the low-frequency count for each category.

Women were the unit of analysis, with only one pregnancy per woman included in the analyses.

To address the first aim, this study estimated the proportion of maternal smokers who used a smoking cessation pharmacotherapy in the 365 days after their delivery dates, using the total number women who smoked at the time of birth as the denominator. Time-to-dispensing was calculated as the number of days between the date of delivery for the index pregnancy and the date of dispensing. Time-to-dispensing values were grouped into: (1) 7-day intervals (weeks) for calculating the cumulative prevalence of smoking cessation pharmacotherapy use and; (2) 8-week intervals (referred to as 2- month intervals in tables for ease of reporting) for measuring the distribution of pharmacotherapy use across the 12 months after delivery. All analyses were conducted separately for each state because the provision of postnatal health services could potentially be different [362].

118

To address the second and third aims, the analyses were restricted to concessional beneficiaries who gave birth in NSW because complete dispensing data for all medications of interest was available only for concessional beneficiaries who gave birth in NSW during the study period.

Multivariable logistic regression models were built to examine whether each study factor was independently associated with the use of pharmacotherapy in the 12 months postpartum. Only variables with at least five observations in each category were included in the multivariable models [363]. Women with missing or unreported values were excluded from the analyses. The comparator group were women who smoked at the time of delivery but did not use any pharmacotherapy in the 12 months after delivery. Adjusted odds ratios (OR) and 95% Confidence Interval (95% CI) are reported.

The number of women who were dispensed with bupropion were too few to examine (n=6), therefore logistic regression models were performed for varenicline and NRT patches. For each pharmacotherapy, three multivariable logistic regression models were built. Model 1 examined the relationship between maternal morbidities and use of each pharmacotherapy, adjusted for potential confounders. No adjustment was made for poor birth outcomes in this model because birth outcomes could potentially lie on the causal pathway between maternal morbidities and pharmacotherapy use, as shown in Figure 6. To examine the relationship between poor birth outcomes and dispensing of pharmacotherapy, the models were adjusted for maternal morbidities. This relationship was examined separately for maternal outcomes (Model 2) and neonatal outcomes (Model 3). Maternal morbidities were treated as a potential confounder as they may independently affect birth outcomes and attempts to quit using pharmacotherapy.

Spearman correlation coefficients were calculated to test for potential collinearity between each covariate. Except for maternal age and parity (r2 =0.62), none of the potential confounders were highly correlated (r2 <0.5) as shown in Appendix 4D Correlation matrix for each potential confounder). Therefore, parity was excluded as a covariate in all multivariable models.

119

All statistical tests were 2-sided at 5% level of significance. Data management and analyses were performed using SAS statistical software version 9.4 (SAS Institute, Inc., Cary, NC, USA).

Ethical approval for this study was granted by the NSW Population and Health Services Research Ethics Committee (2012.06.397), the Australian Institute of Health and Welfare Ethics Committee (EC2012.2.22), the Department of Health WA Human Research Ethics Committee (2013/38), the NSW Aboriginal Health and Medical Research Council Ethics Committee (871/12) and the WA Aboriginal Health Ethics Committee (460). There was no interaction between women and researchers and no identifying information such as names or addresses were provided to the researchers.

The study data were stored and accessed through the Secure Unified Research Environment (SURE) [364]. This is a remote-access computing environment that allows researchers to analyse linked-data for approved studies in Australia. To access this virtual workspace, it requires a username, password and one-time access code provided by a soft token on a smartphone (Mi-Token). The comprehensive security features maximises the safety and protection of research data stored within this environment.

120

Results There were 240,080 women with 247,080 pregnancies between 1 February 2011 and 31 December 2012. Of these, 14,160 women met the inclusion criteria and were included in the analyses. Table 13 presents characteristics of the study population. Of these, 5.9% (n=14,160) had a current smoking diagnosis (ICD-10-AM diagnosis code Z72.0) in the delivery admission record for their index pregnancy and therefore were identified as smoking at the time of delivery.

Overall, the majority of women who smoked at the time of delivery gave birth in NSW (66.4%), were 25-34 years old (47.9%), lived with a partner (56.2%), had prior pregnancy (s) (74.2%), were Australia-born (88.3%), non-Indigenous (78.2%), and smoked less than 10 cigarettes daily (75.7%). The proportion of women who lived in advantaged and disadvantaged areas was approximately equal, as was the proportion living in regional/remote areas and major cities. Among those who had poor birth outcomes, the most common outcomes were emergency caesarean birth (11.2%) and delivering a small for gestational age neonate (17.9%).

Approximately one third (38.0%) of maternal smokers who gave birth in NSW were concessional beneficiaries. With the exception of the socio-economic and remoteness variables, the distribution of maternal socio-demographics and poor birth outcomes among concessional beneficiaries who gave birth in NSW was approximately similar to that of the entire cohort. A higher proportion of concessional beneficiaries who gave birth in NSW lived in a disadvantaged area (46.9%) and in major cities (52.2%). Among women with a relevant morbidity, the most common morbidities were pre-existing mood disorders (22.2%) and respiratory disorders (15.9%).

121

Table 13 Characteristics of women who smoked at delivery, 2011-2012 All women who gave Concessional birth in beneficiaries who gave NSW and birth in NSW WA Total 14,160 Percent 5,377 Percent

State

NSW 9,401 66.4%

WA 4,759 33.6% Potential confounders Maternal age at delivery, years < 25 5,426 38.3% 2,221 41.3% 25-<35 6,779 47.9% 2,504 46.6% >35 1,955 13.8% 652 12.1% Living with partner* 7,958 56.2% 2,378 44.2% Parity* Primiparous 3,645 25.7% 1,078 20.0% Multiparous 9,273 65.5% 3,886 72.3% Grand multiparous 1,235 8.7% 409 7.6% Australian-born 12,500 88.3% 4,999 93.0% Remoteness of residence* Major cities 6,341 44.8% 2,808 52.2% Regional & remote 7,630 53.9% 2,542 47.3% SES of residence* Disadvantaged 5,574 39.4% 2,524 46.9% Average 3,355 23.7% 1,523 28.3% Advantaged 5,042 35.6% 1,303 24.2% Indigenous status* 3,084 21.8% 1,338 24.9% Quantity of cigarettes smoked after 20 weeks of pregnancy* < 10 10,718 75.7% 3,977 74.0% >= 10 2,894 20.4% 1,310 24.4% Maternal morbidities Hypertension (including gestational) 223 4.1% Diabetes (including gestational) 185 3.4% Cardiovascular 47 0.9% Substance use disorder 362 6.7% Psychosis 266 4.9% Renal 45 0.8% Blood 117 2.2% Anxiety disorder 358 6.7% Mood disorder 1,196 22.2% Airway 857 15.9% Epilepsy 100 1.9% Autoimmune disorder 13 0.2% GORD 268 5.0% Thyroid disorder 51 0.9% Use of NSAIDs 412 7.7% 122

All women who gave Concessional birth in beneficiaries who gave NSW and birth in NSW WA Use of steroids 222 4.1% Eating disorder 6 0.1% Hepatitis C 9 0.2% Maternal adverse events PPROM 363 2.6% 90 1.7% Placenta abruption 101 0.7% 24 0.4% Emergency caesarean 1,588 11.2% 370 6.9% MMOI 183 1.3% 40 0.7% Composite of any 2,061 14.6% 515 9.6% Neonatal adverse events Preterm 1,175 8.3% 319 5.9% SGA 2,528 17.9% 1,020 19.0% Admission to NSC 2,111 14.9% 944 17.6% Low Apgar score 354 2.5% 99 1.8% Stillbirth 120 0.8% 28 0.5% Severe NAOI 1,196 8.4% 315 5.9% Composite of any 4,712 33.3% 1,838 34.2%

NSW: New South Wales, WA: Western Australia SES: Socio-economic status GORD: Gastro-intestinal oesophageal reflux disease NSAIDs: Non-steroidal anti-inflammatory drugs PPROM: preterm premature rupture of membrane MMOI: maternal morbidity outcomes indicator SGA: small for gestational age NSC: neonatal special care NAOI: neonatal adverse outcomes indicator

*there were some missing data on variables ‘living with a partner’ (0.8%), ‘parity’ (<0.1%), both ‘remoteness’ and ‘socio-economic status’ (1.3%), ‘Indigenous’ (<0.1%), and ‘number of cigarettes smoked’ (3.9%).

123

Among 14,160 women who smoked at the time of delivery, the overall prevalence of smoking cessation pharmacotherapy use in the 12 months postpartum was 8.7% (95% CI 8.2-9.1), (1,233 women). Varenicline (6.3%, 95% CI 5.9-6.7; 890 women) was the most frequently used smoking cessation pharmacotherapy and bupropion was the least frequently used smoking cessation pharmacotherapy (0.2%, 95% CI 0.1-0.3; 26 women). NRT patches was used by 2.2% 95% CI 2.0-2.5), (317 women).

As shown in Table 14, the median number of days to the first use of varenicline was 195 days (inter-quartile range (IQR): 93-262 days)) after giving birth. The average time to first NRT patches use was shorter with a median of 141 days (IQR: 58-245days). The average time to first bupropion use was 195 days (IQR: 84-249days).

For varenicline, the pattern of use across 2-month intervals within the 12 months after delivery was similar in both States. This was not the case for NRT patches. In WA, the use of NRT patches in the first 2 months postpartum (0.12%, 17 women) and in the 7th and 8th month interval (0.11%%, 16 women) was relatively high. The pattern of bupropion use could not be reported due to cells counts <5 across each 2-month interval.

124

Table 14 Distribution of NRT patches and varenicline use in the 12 months postpartum among women who smoked at the time of delivery, NSW and WA, 2011-2012

Varenicline NRT Time to use* Total NSW WA Total NSW WA (N=890/14,160, (N=631/14,160, (N=259/14,160, (N=317/14,160, (N=244/14,160, (N=73/14,160, 6.3%) 4.5%) 1.8%) 2.2%) 1.7%) 0.1%)

<2 months 125 0.9% 95 0.7% 30 0.2% 82 0.6% 244 0.5% 17 0.1% Month 3 and 4 164 1.2% 114 0.8% 50 0.4% 55 0.4% 65 0.3% 11 0.1% Month 5 and 6 156 1.1% 106 0.7% 50 0.4% 52 0.4% 44 0.3% 9 0.1% Month 7 and 8 159 1.1% 118 0.8% 41 0.3% 47 0.3% 43 0.2% 16 0.1% Month 9 and 10 149 1.1% 109 0.8% 40 0.3% 42 0.3% 31 0.2% 9 0.1% Month 11 and 12 137 1.0% 89 0.6% 48 0.3% 39 0.3% 33 0.2% 11 0.1%

Median, days 195 181 179 141 136 179 IQR, days (93,262) (91,259) (93,272) (58,245) (55.5,239.5) (65,245) *8-week intervals referred to as 2-month intervals IQR: Inter-quartile range

125

Figure 7 shows cumulative prevalence of varenicline and NRT patches use in the 12 months postpartum, calculated across all 7-day intervals. Figure 7 and Table 14 indicated that the proportion of women who used varenicline was the lowest (0.9%, 125 women) in the first 8 weeks. After this immediate 8-week interval, the proportion increased and remained approximately consistent across subsequent intervals.

In contrast, the proportion of women who used NRT patches was the highest (0.58%, 82 women) in the first 8 weeks postpartum before decreasing to the lowest (0.28%, 39 women) in the last interval (after 48 weeks), as shown in Figure 7 and Table 14.

126

Figure 7 Cumulative prevalence of varenicline and NRT patches use in the 12 months postpartum among women who smoked at delivery, NSW and WA, 2011-2012

Prevalence Varenicline 5%

0% Week Week 4 Week 8 Week 12 Week 16 Week20 Week24 Week28 Week32 Week36 Week40 Week44 Week48 Week52 Varenicline NSW Varenicline WA

NRT patches 5%

0% Week 4 Week 8 Week 12 Week 16 Week20 Week24 Week28 Week32 Week36 Week40 Week44 Week48 Week52 NRT NSW NRT WA Weeks 127

Table 15 presents the unadjusted and adjusted association between maternal morbidities, poor birth outcomes and varenicline use, while Table 16 shows the relationship between these factors and NRT patches use. Only adjusted ORs (aOR) for potential confounders and maternal morbidities from Model 1 are shown because the estimates did not differ substantially across three models. The aORs from Model 2 and Model 3 for Table 15 are presented in Appendix 4E and Appendix 4F. Appendix 4G and Appendix 4H contained the aORs from Model 2 and Model 3 for Table 16.

After adjusting for potential confounders, mood disorders was the only maternal morbidity associated with both NRT patches and varenicline use postpartum. Compared to women without mood disorders, women who had mood disorders were more likely to use varenicline (aOR 1.99, 95% CI 1.64-3.13) and NRT patches (aOR 1.40, 95% CI 1.09-1.79). Prior use of NSAIDs was associated with small increase in the odds of using varenicline postpartum (aOR 1.47, 95% CI 1.03-2.10).

This study did not find any relationship between adverse maternal events at birth and postpartum use of varenicline (aOR 1.36, 95% CI 0.98-1.88) or NRT patches (aOR 0.82, 95% CI 0.46-1.45). However, if the study had greater statistical power, it may have been concluded that women with adverse maternal events were more likely to use varenicline postpartum relative to those without, as the lower bound of the confidence interval for the observed effect was close to 1.00. Similarly, no association was found for adverse neonatal events for postpartum use of varenicline (aOR 0.94, 95% CI 0.75-1.18) or NRT patches (aOR 0.96, 95% CI 0.64-1.34).

128

Table 15 Association between maternal morbidities and poor birth outcomes in relation to varenicline use in the 12 months postpartum, 2011-2012 Characteristics No pharmaco- Varenicline OR 95% CI aOR 95% CI aOR 95% CI aOR 95% CI therapy Total 4,806 % 392 % Model 1 (adjusted Model 2 (adjusted Model 3 (adjusted for potential for maternal for maternal confounders) morbidities and morbidities and potential potential confounders ) confounders ) Maternal age < 25 2,009 41.8% 159 40.6% 1.00 ref 1.00 ref 25-<35 2,219 46.2% 203 51.8% 1.16 0.93 1.44 1.12 0.90 1.41 >35 578 12.0% 30 7.7% 0.66 0.44 0.98 0.60 0.39 0.92 Country of birth Australia 4,452 92.6% 379 96.7% 1.00 ref 1.00 ref Overseas 354 7.4% 13 3.3% 0.43 0.25 0.76 0.35 0.19 0.65 Living with a partner Yes 2,129 44.3% 170 43.4% 1.00 ref 1.00 ref No 2,658 55.3% 222 56.6% 0.96 0.78 1.18 0.96 0.78 1.20 Remoteness of residence Major cities 2,490 51.8% 221 56.4% 1.00 ref 1.00 ref Regional & 2,290 47.6% 170 43.4% 0.84 0.68 1.03 0.87 0.70 1.09 remote SES of residence Disadvantaged 2,290 47.6% 160 40.8% 1.00 ref 1.00 ref Advantaged 1,341 27.9% 127 32.4% 1.30 1.00 1.68 1.24 0.95 1.63 Average 1,149 23.9% 104 26.5% 1.36 1.06 1.73 1.28 1.00 1.64 Indigenous Status No 3,586 74.6% 313 79.8% 1.00 ref 1.00 ref Yes 1,219 25.4% 79 20.2% 0.74 0.58 0.96 0.73 0.56 0.95

129

Characteristics No pharmaco- Varenicline OR 95% CI aOR 95% CI aOR 95% CI aOR 95% CI therapy Total 4,806 % 392 % Model 1 (adjusted Model 2 (adjusted Model 3 (adjusted for potential for maternal for maternal confounders) morbidities and morbidities and potential potential confounders ) confounders ) Quantity smoked after 20 weeks of pregnancy < 10 sticks 3,579 74.5% 277 70.7% 1.00 ref 1.00 ref >= 10 sticks 1,145 23.8% 108 27.6% 1.22 0.97 1.54 1.23 0.97 1.55 Maternal morbidity** Hypertension 193 4.0% 22 5.6% 1.42 0.90 2.24 1.42 0.89 2.24 (including gestational) Diabetes 163 3.4% 16 4.1% 1.21 0.72 2.05 1.12 0.64 1.98 (including gestational) GORD 231 4.8% 26 6.6% 1.41 0.93 2.14 1.37 0.88 2.12 Mood 1,027 21.4% 103 26.3% 1.31 1.04 1.66 1.32 1.03 1.69 Anxiety 315 6.6% 26 6.6% 1.01 0.67 1.53 0.99 0.63 1.54 Psychosis 234 4.9% 16 4.1% 0.83 0.50 1.40 0.78 0.46 1.35 Drugs and 333 6.9% 19 4.8% 0.68 0.43 1.10 0.67 0.41 1.11 alcohol Respiratory 754 15.7% 67 17.1% 1.11 0.84 1.46 1.02 0.76 1.37 Use of NSAIDs 353 7.3% 42 10.7% 1.81 1.32 2.49 1.48 1.04 2.10 Epilepsy 88 1.8% 5 1.3% 0.69 0.28 1.72 0.69 0.28 1.73 Use of steroids 195 4.1% 14 3.6% 0.88 0.50 1.52 0.83 0.47 1.46 Blood coagulation 109 2.3% 6 1.5% 0.67 0.29 1.53 0.72 0.31 1.67 Maternal adverse 456 9.5% 45 11.5% 1.24 0.89 1.71 1.21 0.86 1.69 events, composite** Neonatal adverse 1,649 34.3% 130 33.2% 0.95 0.76 1.18 0.97 0.78 1.22 events, composite** *aOR: adjusted odds ratio ** Reference category: Women who did not have the morbidity *** Percentages do not add to 100% due to missing data 130

Table 16 Association between maternal morbidities and poor birth outcomes in relation to NRT patches use in the 12 months postpartum, 2011-2012 Characteristics No pharmaco- NRT OR 95% CI aOR 95% CI aOR 95% CI aOR 95% CI therapy Total 4,806 % 173 % Model 1 (adjusted Model 2 (adjusted Model 3 (adjusted for potential for maternal for maternal confounders) morbidities and morbidities and potential potential confounders ) confounders ) Maternal age < 25 2,009 41.8% 52 30.1% 1.00 ref 1.00 ref 25-<35 2,219 46.2% 79 45.7% 1.38 0.96 1.96 1.28 0.89 1.85 >35 578 12.0% 42 24.3% 2.81 1.85 4.26 2.55 1.63 3.97 Country of birth Australia 4,452 92.6% 162 93.6% 1.00 ref 1.00 ref Overseas 354 7.4% 11 6.4% 0.85 0.46 1.59 0.73 0.38 1.41 Living with a partner Yes 2,129 44.3% 76 43.9% 1.00 ref 1.00 ref No 2,658 55.3% 97 56.1% 0.98 0.72 1.33 0.95 0.70 1.30 Remoteness of residence Major cities 2,490 51.8% 95 54.9% 1.00 ref 1.00 ref Regional & 2,290 47.6% 78 45.1% 0.89 0.66 1.21 0.97 0.69 1.34 remote SES of residence Disadvantaged 2,290 47.6% 71 41.0% 1.00 ref 1.00 ref Advantaged 1,341 27.9% 53 30.6% 1.38 0.95 1.99 1.32 0.89 1.94 Average 1,149 23.9% 49 28.3% 1.28 0.89 1.83 1.24 0.86 1.80 Indigenous Status No 3,586 74.6% 136 78.6% 1.00 ref 1.00 ref Yes 1,219 25.4% 37 21.4% 0.80 0.55 1.16 0.89 0.61 1.31 Quantity smoked after 20 weeks of pregnancy 131

Characteristics No pharmaco- NRT OR 95% CI aOR 95% CI aOR 95% CI aOR 95% CI therapy Total 4,806 % 173 % Model 1 (adjusted Model 2 (adjusted Model 3 (adjusted for potential for maternal for maternal confounders) morbidities and morbidities and potential potential confounders ) confounders ) < 10 sticks 3,579 74.5% 118 68.2% 1.00 ref 1.00 ref >= 10 sticks 1,145 23.8% 54 31.2% 1.43 1.03 1.99 1.26 0.90 1.77 Maternal morbidities Hypertension 193 4.0% 6 3.5% 0.86 0.38 1.96 0.75 0.32 1.74 (including gestational) Diabetes 163 3.4% 6 3.5% 1.02 0.45 2.35 0.85 0.37 1.98 (including gestational) GORD 231 4.8% 9 5.2% 1.09 0.55 2.15 0.82 0.40 1.66 Mood 1,027 21.4% 64 37.0% 2.16 1.58 2.96 1.95 1.39 2.74 Anxiety 315 6.6% 17 9.8% 1.55 0.93 2.60 1.03 0.60 1.79 Psychosis 234 4.9% 15 8.7% 1.86 1.08 3.20 1.28 0.71 2.30 Drugs and 333 6.9% 10 5.8% 0.82 0.43 1.58 0.62 0.31 1.20 alcohol Respiratory 754 15.7% 35 20.2% 1.36 0.93 1.99 1.15 0.77 1.73 Use of NSAIDs 353 7.3% 17 9.8% 1.93 1.17 3.18 1.20 0.71 2.03 Epilepsy 88 1.8% 7 4.0% 2.26 1.03 4.96 1.84 0.81 4.19 Use of steroids 195 4.1% 12 6.9% 1.76 0.96 3.22 1.38 0.72 2.62 Blood 109 2.3% ------coagulation# Maternal adverse 456 9.5% 45 12.4 0.78 0.44 1.38 0.73 0.40 1.33 events, composite Neonatal adverse 1,649 34.3% 130 33.2 0.97 0.70 1.33 0.88 0.63 1.23 events, composite *aOR: adjusted odds ratio ** Reference category: Women who did not have the morbidity *** Percentages do not add to 100% due to missing data # Not modelled due to insufficient events

132

Discussion

In this study of women who smoked at the time of delivery, the proportion of women who used a smoking cessation pharmacotherapy in the 12 months postpartum was 8.7% (95% CI 8.2-9.1%). Varenicline was used by most women, 6.3% (95% CI 5.9- 6.7%) compared to 2.2% (95% CI 2.0-2.5%) of maternal smokers using NRT patches. This study observed a contrasting pattern of varenicline and NRT patches use in the 12 months postpartum. The proportion of women using varenicline was lowest in the first 8 weeks interval immediately following birth (0.88%), and thereafter increased to 1-1.20% of women using it in each 8-week interval. The prevalence of NRT patches use was highest in the first interval (0.58%), thereafter decreasing to about 0.30-0.40%.

In the analyses examining potentially influential health conditions, this study observed very few maternal morbidities were associated with the use of smoking cessation pharmacotherapy. Women with mood disorders were more likely to use either varenicline or NRT patches postpartum, and women who previously used NSAIDs were significantly more likely to use varenicline postpartum. This study did not find that experiencing poor birth outcomes was significantly associated with the likelihood of using smoking cessation pharmacotherapy postpartum. Although there appeared to be a relationship between adverse maternal events and the postpartum use of varenicline, the effect did not reach statistical significance.

These findings regarding the use of smoking cessation pharmacotherapy after delivery are the first population-based estimates from Australia. The overall estimate is higher than the estimate from the population-based study in the US (2.0% of pregnant smokers were dispensed with a smoking cessation pharmacotherapy in the one year postpartum [104]). The estimate for NRT use, however is lower than the population- based estimate for the UK (about 5% of women who smoked during pregnancy were prescribed NRT in the 9 months postpartum) [107]. However, neither of these studies was designed to accurately measure the proportion of postpartum smokers using pharmacotherapy; they were based on women smoking during pregnancy, of which 4-

133

25% would have quit smoking during pregnancy and thus, no longer would be smoking postpartum [47-49, 70].

In Australia, the 12-month postpartum use (8.7%) of a smoking cessation pharmacotherapy was considerably higher than the use during pregnancy (2.3-3.6%) [110]. Nonetheless, this finding revealed that the period after giving birth is still an under-used time window to encourage quit attempts using smoking cessation pharmacotheraoy.

The finding that varenicline use was more prevalent in the year after delivery than NRT is unsurprising given that varenicline is more effective than NRT in the general population [77]. The observation that the prevalence of varenicline use was the lowest in the first 2 months after giving birth may be due, in part, to concerns regarding the safety of using varenicline during lactation; without information on whether women were breastfeeding, this possibility cannot be confirmed. Nonetheless, studies regarding the safety of varenicline use during lactation are needed, given the apparent preference for varenicline over other pharmacotherapies. Unless indicated otherwise, greater NRT use immediately after delivery should be encouraged given that NRT is not contraindicated in women who are breastfeeding.

With the exception of women with a diagnosis of mood disorders and prior use of NSAIDs, this study did not observe any relationship between pre-existing maternal morbidities and postpartum use of varenicline or NRT patches. The findings are inconsistent with those of a population-based study in the US; among five examined maternal morbidities, substance use disorder was associated with a greater use of a smoking cessation pharmacotherapy in the year after giving birth (OR 1.65, 95% CI 1.01-2.76). However, whether this is genuine difference in the patterns of pharmacotherapy use postpartum, or due to the methodological limitations of both studies (women in the studies’ cohorts were not necessarily smoking postpartum), is unclear.

The finding that smoking cessation pharmacotherapies were not used to a greater extent in women at higher risk of harm from their continued smoking is concerning, highlighting the possibility that the suitability for smoking cessation pharmacotherapies

134

may not be adequately considered in women smoking postpartum. This calls for strategies to more selectively encourage use of smoking cessation pharmacotherapies among women whose co-morbidities increase their suitability for the use of these pharmacotherapies.

In contrast, the use of these pharmacotherapies was greater in women who were likely less suitable for these pharmacotherapies due to an increased risk of adverse events from these pharmacotherapies. While smoking cessation pharmacotherapies may be less suitable for women with a history of mood disorders, a likely explanation for their greater use is that women with a history of mood disorders including depression are more likely to be screened for mood disorders during the postpartum period and thus be treated for smoking cessation.

The positive association between prior history of NSAIDs use and postpartum use of varenicline is a concern because there may be an increased risk for cardiovascular events associated with use of varenicline among those who had a recent use of NSAIDs. A 2006 meta-analysis of 138 randomised trials found that NSAIDs increase the risk of cardiovascular events [365] by 42% (rate ratio 1.42, 95% CI 1.13-1.78). Meanwhile, evidence from a 2012 meta-analysis of 22 randomised trials suggests that varenicline might do the same, although the effect did not reach statistical significance (0.63% in the varenicline group vs 0.47% in the placebo group, with a risk difference of 0.27%, 95% CI -0.10-0.63) [206]. As NSAIDs that are subsidised by PBS included the higher strength formulations [72], and these are associated with greater COX-1 inhibition activity, all of which may indicate a higher risk of bleeding if there is a recent use of these high-strength form of NSAIDs [366]. An increase risk of bleeding potentially increase the risk of an adverse cardiovascular event [367].

It was reassuring, however, that this study did not observe any association between varenicline use and other diagnosed disorders that placed them at risk of developing varenicline side-effects such as history of GORD and epilepsy disorders. However, the relatively small numbers of women with these comorbidities limits the power of the study to detect relationships that might be present. Therefore, this suggests future studies explore such associations with larger samples to support or refute the relationship.

135

The finding that poor birth outcomes were an inadequate motivator to induce quit attempts using smoking cessation pharmacotherapy fits with evidence that a high proportion of women continue smoking during their next pregnancy despite poor pregnancy outcomes [303, 309, 310, 368]. This suggests that current cessation efforts are failing to capitalise on the opportunity presented by the postpartum period, especially since maternal and infant healthcare providers already have access to the recuperating mother and child. Although smoking mothers of hospitalised infants in the US and UK [369, 370] were receptive to NRT and tobacco cessation support, evidence suggests postnatal providers’ were reluctant to prescribe NRT, although the extent of this reluctance varied across obstetricians, midwives and paediatricians [241, 371-373]. Further research is needed to investigate the facilitators and barriers to prescribing smoking cessation pharmacotherapies among postnatal healthcare providers.

In terms of external validity, first, there is a high likelihood of under-ascertainment of smoking status at delivery. Previous validation studies reported that the sensitivity of identifying smoking from perinatal data was 89.6% and 66.3% from hospital data, whereas the specificities are generally found to be higher (>93%) [374, 375]. The impact of this imperfect ascertainment of smoking is that the cohort of women smoking at the time of birth was not complete, and that the women identified may differ from the entire population of women who smoke at the time of delivery. Therefore, the findings regarding the prevalence of correlates of pharmacotherapy utilisation generated here may not generalise to the entire population of women smoking at the time of birth.

Second, the finding regarding the relationship between maternal morbidities, poor birth outcomes, and the use of smoking cessation pharmacotherapy may not be generalizable to the entire population of postpartum smoking. This is partly due to the restriction of these analyses to concessional beneficiaries who gave birth in NSW. Concessional beneficiaries are recipients of income support from the Australian government, with most concessional beneficiaries aged 16-44 receiving this support due to low-income earning, unemployment, and parenting or carer responsibilities [316], suggesting that concessional beneficiaries tend to be of lower socio-economic

136

status than the general population of postpartum smokers. This may limit the generalisation of this study finding.

However, this study benefitted from using routinely collected data sources. First, selection bias is minimised as this study included the entire population of women giving birth in NSW and WA within the study period. This included home births, as well as deliveries in private hospitals [26]. Second, it allowed the temporal relationship between a wide range of pre-existing maternal comorbidities and the use of smoking cessation pharmacotherapies within 12-month postpartum to be examined in a timely and cost-efficient manner.

Regarding internal validity, in the absence of measures of the actual use of smoking cessation pharmacotherapy, this study used dispensings of subsidised pharmacotherapies as a proxy for utilisation. Moreover, a single dispensing of the medication was assumed to represent use, when two or more dispensings is sometimes taken as a more conservative indicator of use [376]. However, requiring two dispensings is unlikely to be appropriate for smoking cessation pharmacotherapies because they are short-term therapies (usually 2 dispensings in total), and the average duration of use is considerably shorter than the recommended course (see Table 3). Pregnant women enrolled in smoking cessation trials used NRT patches for an average 2 weeks [95, 377], instead of the recommended 8 -12 weeks of NRT. Moreover, many utilisation studies relating to smoking cessation pharmacotherapies in the general population [129, 378, 379], and pregnancy [110, 127], defined use of these pharmacotherapies as at least one dispensing. As such, NRT use could be underestimated because the PBS data did not capture supplies of NRT to hospitalised patients, nor OTC purchases such as NRT gums, lozenges and inhalers. The prevalence of NRT use would potentially be higher if all formulations of NRT were accounted for. However, it is anticipated that the OTC purchases of NRT would have a negligible impact on the overall use of NRT due to the significant price differences between subsidised and recommended retail price. As described in section 1.3.4 Cost (see Table 3), as of 1st January 2019, a concessional beneficiary (whose co-payment threshold is $6.50) will pay $13.00 for 8-week supply for NRT patches, while a general beneficiary will pay $82.00 (whose co-payment threshold is $41.00) [72]. This is in contrast to at least $336 for OTC purchase of NRT gums or lozenges [72]. Therefore, it

137

is highly unlikely that the majority of maternal smokers would choose to access the unsubsidised form of NRT.

Likewise, in the absence of true measures of maternal morbidities, this study relied on the diagnoses and dispensed medications documented in the perinatal, hospital and dispensing data as surrogate measures for the presence of maternal morbidities. Not all medications were available in the dispensing data extract for this project. This may result in under-ascertainment of maternal morbidities that use medications as surrogate measures. However, the selected maternal morbidities (see Table 12) are sufficiently identified using the available dispensing data. One exception is the lack of information on beta-blockers, medications indicated primarily for treatment of hypertension [120]. However, the diagnosis of hypertension is adequately identified from the perinatal and hospital data [312, 380, 381]. It is also possible that the diagnoses were under-coded because they did not influence the patient’s treatment. Moreover, many of the morbidities do not warrant hospitalisation or treatment with medications and therefore, may be missed, resulting in under-estimation of the morbidity. This is more so for disorders including substance use and alcohol use disorder as well as GORD, whereby only severe episodes necessitating hospitalisation and prescribed medications were identified. This possible misclassification means that the observed association between maternal morbidity and use of pharmacotherapy is possibly attenuated towards null findings. However, the breadth of information available in the linked administrative datasets allowed for investigation of a wide range of study factors and adjustment for potential confounders. This study was able to examine 15 maternal morbidities as opposed to five morbidities examined in the US study [104]. This study was also able to explore the effect of individual morbidities, including psychosis, mood and anxiety disorders as opposed to a general category of mental health disorders, as done in the aforementioned research [104].

Another limitation relates to unmeasured confounders. This study did not have information on the following independent risk factors for smoking behaviour change: body mass index, living with a partner who smokes, level of education, and level of nicotine addiction in terms of duration of years smoked, Fagerstrom Test for Nicotine Dependence scores or biochemical markers [29, 382]. Duration of years smoked is an important confounder as many of co-morbidities associated with smoking are dose-

138

dependent, primarily due to cumulative exposure to tobacco smoke [383]. In the absence of validated measures, nicotine dependence was ascertained from the self- reported number of cigarettes smoked. This could be subject to recall bias.

This first, large Australian population-based study found low use of smoking cessation pharmacotherapies in the 12 months postpartum among women who smoked at delivery. The prevalence of varenicline use was higher than the prevalence of NRT patches use; however, this pattern was reversed the first 8 weeks after giving birth. Importantly, except for diagnosed mood disorders, this study did not find an association between morbidities worsened by smoking or poor birth outcomes and the use of smoking cessation pharmacotherapy. It appears that suitability for smoking cessation pharmacotherapies may not be adequately considered among these women with comorbidities in the decision to use these pharmacotherapies. To minimise infants’ exposure to maternal second-hand smoke, strategies to increase the use of smoking cessation pharmacotherapies after giving birth are needed to capitalise on the immediate period after delivery. In particular, the opportunity for use among women most likely to benefit from quitting smoking postpartum due to experiencing poor birth outcomes should be seized. This could include potentially targeting postnatal care providers who are in contact with women and infants. Future research is needed to examine the safety of varenicline use after giving birth.

139

Chapter 5 The effectiveness of smoking cessation pharmacotherapy during the inter-pregnancy interval

Introduction Chapter 1 (section 1.2.2 Prevalence of smoking after pregnancy) indicates that most women who smoke up to their third trimester are likely to smoke after giving birth, and as described in section 1.2.3.2 (Barriers to smoking cessation after pregnancy) may experienced difficulty with quitting smoking unassisted. These women could potentially benefit from using smoking cessation pharmacotherapy after giving birth, given that these pharmacotherapies are the most effective smoking cessation intervention in the general population (section 1.3.2 Efficacy in the general population). However, decisions about the use of smoking cessation pharmacotherapies among women smoking after giving birth ought to be based on evidence of their effectiveness in this postpartum population.

The effectiveness of smoking cessation pharmacotherapies may differ among women in the postpartum period due to changes in the physiology that occurred during pregnancy returning to the prepregnant state [189]. These altered physiologic characteristics may affect the pharmacokinetic properties of smoking cessation pharmacotherapies, thus evidence regarding their effectiveness in the general population cannot be assumed to apply in postpartum smokers.

To date, three randomised controlled trials have evaluated the effectiveness of using smoking cessation pharmacotherapy in the postpartum period. Inconsistent findings were returned from three trials of NRT, with two studies showing an increased proportion of participants stopping smoking in the intervention groups at 6 months; a 2008 study [14% vs 5% (n=21)] [384] and a 2019 trial (reduction of 2.7% smoking prevalence) [385]. Another 2014 trial did not find a significant difference between intervention and control group (4.3% vs 4.1%, n=999) [386]. Importantly, the results from these studies do not provide insight into the effectiveness of NRT among women smoking in the postpartum period as NRT was provided to parents as part of a broader intervention to treat parental quitting [385, 386]. There was no information on whether mothers specifically received the intervention, and limited information on whether the mothers quit smoking. It is also uncertain whether the effect on maternal quitting

140

observed in the 2014 trial [386] can be attributed to NRT, or whether it is a result of influence by their partner’s quitting as partner smoking is a strong predictor of maternal smoking cessation [29]. There is also no evaluation of newer smoking cessation pharmacotherapies including varenicline, despite it being the most effective smoking cessation pharmacotherapy in the general population (section 1.3.2 Efficacy in the general population).

In measuring the effectiveness of smoking cessation pharmacotherapies after giving birth, this study considers only those maternal smokers who had subsequent pregnancies during the study period. In women who smoke in one pregnancy, a high proportion (50.1%-72.2%) continued smoking in their next pregnancy [303, 368, 387, 388]. In women who are current smokers at the time of delivery, the inter-pregnancy period should be taken as an opportunity to address quitting after delivery and before the subsequent pregnancy. Indeed, obstetric care guidelines in the US, UK and Australia strongly recommend smoking cessation care support as an integral component of pre- and inter-pregnancy care provision [389-391]. All women of reproductive age are recommended to receive inter-pregnancy care as a continuum of her postpartum care regardless of whether they conceive subsequently [389]. In addition to improving the likelihood that women do not smoke through their next pregnancy, quitting after giving birth helps to reduce the harm associated with second- hand smoke exposure to her newborn. The health risks associated with infants’ exposure to second-hand smoke are discussed in Chapter 1. Briefly, infants exposed to second-hand smoke are at increased risk of SIDS (deaths less than one year) [148], worsening respiratory infections [149], increased risk of hospitalisation due to fire- related injuries, [150] and increased likelihood of developing life-long negative health consequences such as ischaemic heart disease, obesity [152] as well as long-term neurological morbidity [153].

Importantly, in assessing the effectiveness of smoking cessation pharmacotherapy, this study controls for nicotine dependence. Adequate adjustment for nicotine dependence is important because systematic differences exist between smoking cessation pharmacotherapy users and non-users, where nicotine dependence is usually higher among these pharmacotherapy users [392-394], making them less likely to quit smoking. Inadequate adjustment for nicotine dependence is an important

141

methodological concern that was raised as a limitation in many population-based studies examining the real-world effectiveness of pharmacotherapies in the general population [395-397].

Aim This study aims to assess whether exposure to smoking cessation pharmacotherapy during the inter-pregnancy interval is associated with smoking cessation prior to the subsequent pregnancy.

Methods

This study was based on the data linked for the Smoking MUMS Study, which measured during-pregnancy utilisation, effectiveness and safety of smoking cessation pharmacotherapies. Details of the data sources and linkages for the Smoking MUMS study are described elsewhere [29] and in Chapter 4. Briefly, the Smoking MUMS Study used six linked datasets, including perinatal records, hospital admission records, death records and pharmaceutical dispensing records for all women who gave birth in NSW and WA between 2003 and 2012.

This retrospective population-based cohort included women who had consecutive pregnancies, both of which resulted in a delivery in NSW between 1 January 2008 and 31 December 2012. Consecutive pregnancies were eligible for inclusion in the study if the mother was smoking at the time of first delivery in the pregnancy pair.

This study population was identified using the following steps. First, all pregnancy records in the NSW perinatal data with a delivery date between 1 January 2008 and 31 December 2012 were ordered according to the date of delivery.

142

Pregnancies in which the mother was recorded as a smoker at the time of the delivery were identified through the presence of the ICD-10-AM diagnosis code Z72.0 in any diagnosis field of the delivery admission records. This diagnosis code represents the presence of any tobacco use in the last 28 days.

For each pregnancy in which the mothers smoked at the time of delivery, subsequent pregnancies of the same mother during the study period were identified. As this study focuses on consecutive pregnancies, the earliest subsequent pregnancy was identified based on the woman’s parity, specifically with an increment of 1 for the perinatal record item ‘number of previous pregnancies’. This resulted in an initial population of eligible pregnancy pairs, defined as two consecutive pregnancies in which the first delivery had an ICD-10-AM diagnosis code Z72.0. In each pregnancy pair, this study defined the first delivery as the ‘first pregnancy’ and the subsequent pregnancy as the ‘second pregnancy’.

Next, the interval between pregnancy in each pair was calculated, referred to as the inter-pregnancy interval. The inter-pregnancy interval was calculated as the time in days between the date of delivery of the first pregnancy and the estimated conception date of the second pregnancy. As described in Chapter 4, the date of conception was calculated as: date of delivery- gestational age at delivery x 7 + 14 days [311, 321]. This study excluded pregnancy pairs with an inter-pregnancy interval shorter than 180 days because women with short inter-pregnancy intervals (less than 6 months) have different risk profiles. Women with short inter-pregnancy intervals are more likely to have poor maternal, foetal health and pregnancy outcomes [398-400]. These different risk profiles may influence the likelihood of quitting smoking after giving birth, as the motivation to quit smoking may differ in women with short inter-pregnancy intervals.

For women with more than one eligible pregnancy pair during the study period, only the earliest pair was analysed. This earliest pair is defined as the index pregnancy pair for the woman.

Although data on pregnancies resulting in births in WA were available through the Smoking MUMS Study, the main analyses in the current study were limited to pregnancies in NSW. A high proportion of women who gave birth in WA between 2008

143

and 2012, had missing information on the potential confounder, nicotine dependence. The proportion of women with missing information on the variable ‘quantity of cigarettes smoked’, was 81.8% for all women who delivered in WA during the 2008-2012 period.

As described in Chapter 4, pharmaceutical dispensing records were used to identify the dispensing of any smoking cessation pharmacotherapies during the inter-pregnancy interval. Varenicline dispensings were identified using ATC code N07BA03, and NRT patches dispensings were based on the presence of ATC code N07BA01. Bupropion dispensing were identified using ATC codes N07BA02 and N06AX12 because in 2009, bupropion’s ATC code changed from N07BA02 to N06AX12 [323, 324]. In Australia, bupropion is not registered for use as an antidepressant [322].

Exploratory analyses revealed that few women were dispensed bupropion or NRT patches during the inter-pregnancy period. Varenicline was the only smoking cessation pharmacotherapy that could be examined in this study.

Women were defined as exposed to varenicline in the inter-pregnancy period if they had one or more dispensing of varenicline, with the date of dispensing falling during the inter-pregnancy period. Women who were dispensed varenicline as well as bupropion and /or NRT patches during the inter-pregnancy intervals were excluded. This allowed the study to measure the independent effect of exposure to varenicline on promoting smoking cessation. In the general population, there is some evidence to suggest that varenicline used in combination with other pharmacotherapies may be more efficacious than varenicline monotherapy [401, 402]. The first varenicline dispensing during this period was referred to as the index varenicline dispensing. Women who received their index varenicline dispensing less than 7 days before the conception date for the second pregnancy, were also excluded. Generally, varenicline is recommended to be used 7 days before the planned quit date.

This study considered the unexposed group as women who were NOT dispensed with varenicline, bupropion or NRT patches during the inter-pregnancy interval.

144

The outcome of interest in this study was smoking cessation prior to the second pregnancy. Women’s smoking status during the second pregnancy was used as a proxy for this smoking cessation outcome. Perinatal data provided information on women’s smoking status during pregnancy.

The smoking information recorded in the NSW perinatal data was revised during the study period. For deliveries between 2008 and 2010, the smoking items comprised whether the mother smoked at all during pregnancy and the number of cigarettes smoked in the second half of pregnancy. For deliveries after 1 January 2011, the smoking items were altered to include whether a woman smoked in the first and second half of pregnancy and the quantity smoked in the first and second half of pregnancy. Table 17 shows items included in the perinatal data and the derivation of smoking cessation outcome.

145

Table 17 Description of the smoking information available and methods used to define smoking cessation outcome Time Item description Responses Definition of period smoking cessation 2008-2010 Did the mother 0 = No Based on field smoke at all during 1 = Yes values “No” pregnancy? 9 = Not stated AND

How many cigarettes 0 = Did not smoke Based on field each day on average 1 = Less than 1 per day values “0” were smoked in the 2 = 1-10 per day second half of 3 = More than 10 per day pregnancy? 4 = Unknown 9 = Not stated

2011-2012 Did the mother 0 = No Based on field smoke during the first 1 = Yes values “No” half of pregnancy? 9 = Not stated AND

Quantity smoked 0 = Did not smoke Based on field during the second 1 = Less than 1 per day values “0” half of pregnancy 2 = 1-10 per day 3 = More than 10 per day 4 = Unknown 9 = Not stated

Did the mother 0 = No IF DATA FOR smoke during the 1 = Yes FIRST HALF OF second half of 9 = Not stated PREGNANCY pregnancy? MISSING Based on field value “0”

146

Given the changes to the smoking information available, different methods were used to define smoking cessation for deliveries during the period of 2008-2010 and 2011- 2012, as summarised in Table 17. For deliveries between 2008 and 2010, women were considered non-smokers in the second pregnancy if a “no” was recorded against ‘Did the mother smoke at all during pregnancy?’ AND “did not smoke” was recorded for the question ‘How many cigarettes each day on average were smoked in the second half of pregnancy?’. Women were identified as smokers in the second pregnancy if a “yes” was recorded against ‘Did the mother smoke at all during pregnancy?’ OR a “1”, “2” or “3” was recorded in the field ‘How many cigarettes each day on average were smoked in the second half of pregnancy?’. Values of ‘1”, “2”, and “3” correspond to “Less than 1 per day”, “1-10 per day”, and “More than 10 per day”, respectively.

For 2011-2012 deliveries, women were identified as non-smokers in the second pregnancy if “no” was recorded in response to the item ‘Did the mother smoke during the first half of pregnancy?’ AND “0” for the question ‘How many cigarettes each day on average were smoked in the first half of pregnancy?’. If a “yes” response was recorded to ‘Did the mother during the first half of pregnancy?’ OR values of “1”, “2” or “3” were recorded in the field ‘How many cigarettes each day on average were smoked in the second half of pregnancy?’, these women were considered smokers.

There were 1,108 women who had missing information for items relating to smoking in the first half of pregnancy (2011-2012) and quantity smoked in the second half of pregnancy (2008-2010). In these women who delivered between 2011-2012, this study relied on responses for smoking in the second half of pregnancy “no” for ‘Did the mother smoke during the second half of pregnancy?’ AND “0” for ‘How many cigarettes each day on average were smoked in the second half of pregnancy?’. As for deliveries in 2008-2010 with missing information on the quantity smoked, this study relied upon responses to the item ‘Did the mother smoke at all during pregnancy?’. Eight women were excluded from the analyses because data on smoking in the second pregnancy were missing.

147

Thirteen maternal characteristics measured at the time of first pregnancy were included in the analyses as potential confounders in the association between exposure to smoking cessation pharmacotherapy during the inter-pregnancy interval and quitting prior to the second pregnancy. They included maternal socio-demographics (nine variables), the number of cigarettes smoked during the second half of pregnancy, poor birth outcomes (two variables) and maternal morbidities (two variables).

Nine of these potential confounders were included in the analyses described in Chapter 4. This study applied the same categorisation of these variables. Poor birth outcomes for the first pregnancy included a composite measure of any maternal adverse events, and a composite of neonatal-related adverse events, using the measures described in Chapter 4.

As described in the previous chapter, this study also considered the number of cigarettes smoked during the second half of pregnancy as a proxy for nicotine dependence [359]. Nicotine dependence is a key potential confounder in the relationship between varenicline exposure and the likelihood to quit [395, 403], as mentioned in the introduction of this chapter. This information was ascertained from the perinatal record associated with the first pregnancy, specifically from the variable ‘How many cigarettes each day on average were smoked in the second half of pregnancy?’. This study excluded 895 women who had missing data on this potential confounder.

Information on maternal morbidities was limited to pre-existing diabetes (including gestational) and pre-existing hypertension (including gestational), obtained from the perinatal record. The rest of the maternal morbidities examined in Chapter 4 were not included as potential confounders in the main analyses for this study because their ascertainment would require the restriction of the study population to concessional beneficiaries. As explained in Chapter 4, some maternal morbidities are identified using medication dispensing records, where until July 2012, the PBS data completely captured medications dispensed to concessional beneficiaries only. Therefore, the restriction to concessional beneficiaries would reduce the study’s statistical power to detect an association between varenicline exposure during the inter-pregnancy interval

148

and smoking cessation prior to the second pregnancy, if such a relationship does exist. Moreover, the study described in Chapter 4 found that few maternal morbidities (mood disorders and prior dispensing of non-steroidal anti-inflammatory drugs prescription) were associated with varenicline dispensing postpartum, thus these maternal morbidities are unlikely to be important confounders in this study. Nevertheless, in sensitivity analyses based on concessional beneficiaries (detailed in section 5.3.6), adjustment was made for the entire range of maternal morbidities examined in Chapter 4.

Three additional potential confounders included parity, inter-pregnancy interval and year of first pregnancy. Parity was grouped into 0 (nulliparous), 1-4 (multiparous) and >=5 (grand multiparous). Parity was identified based on the numerical responses recorded for the variable ‘number of previous pregnancies greater than 20 weeks gestation’. Inter-pregnancy interval was calculated in days and grouped into < 1 year (180-<366 days), 1-2 years (366-<730 days), 2-3 years (730-<1095 days) and > 3years (>1095 days). The year of the first pregnancy delivery was derived from the date of delivery for the first pregnancy. All three variables were obtained from perinatal data.

Descriptive data regarding the characteristics of the study population at the time of the first pregnancy are reported. In order to assess the representativeness of the women included in the analyses, the characteristics of women who were included and excluded due to missing information on the number of cigarettes smoked during the second half of pregnancy were compared.

A multivariable logistic regression model was built to examine whether varenicline exposure during the inter-pregnancy interval was associated with smoking cessation prior to the second pregnancy, after adjusting for potential confounders. Spearman’s correlation coefficient (r2) was calculated to test for potential collinearity between each covariate. None of the potential confounders were highly correlated (r2 <0.5). Adjusted odds ratios (OR) with 95% confidence intervals (95% CI) are reported.

149

To examine whether additional maternal morbidities could significantly alter the effect estimates, a sensitivity analysis was carried out. This sensitivity analysis restricted the study population to concessional beneficiaries who gave birth in NSW (55.9% of the overall study population). Using the same method to identify concessional beneficiaries as used in the previous chapter, this study defined concessional beneficiaries as women who had at least one dispensing record as a concessional beneficiary and no records as a general beneficiary in the one year prior to conception and throughout pregnancy to birth. Eight additional maternal morbidities comprising a 1-year history of disorders relating to drugs and alcohol, mood, psychosis, anxiety, respiratory, gastro- oesophageal reflux, dispensing of non-steroidal anti-inflammatory medications, and dispensing of steroids (measured as described in Chapter 4), were included in the multivariable model.

All statistical tests were two-sided and based on a 5% level of significance. Data management and analyses were performed using SAS statistical software version 9.4 (SAS Institute, Inc., Cary, NC, USA).

Ethical approval for this study was granted by the NSW Population and Health Services Research Ethics Committee (2012.06.397), the Australian Institute of Health and Welfare Ethics Committee (EC2012.2.22), the Department of Health WA Human Research Ethics Committee (2013/38), the NSW Aboriginal Health and Medical Research Council Ethics Committee (871/12) and the WA Aboriginal Health Ethics Committee (460). The study data were stored and accessed through the Secure Unified Research Environment (SURE) [364].

150

Results

Figure 8 shows the cohort selection procedure as described in section 5.3.2. In the population of eligible pregnancy pairs of women who gave birth in NSW between 2008 and 2012, 4,395 women had complete information on the ‘number of cigarettes smoked during the second half of pregnancy’ for the first pregnancy. Appendix 5 shows that women who were excluded due to missing information on this variable were different compared to those who were included in this study. Compared to women who were included in the study, the excluded women were more likely to be older than 25 years, have a partner, be overseas-born and live in major cities as well as in advantaged areas (see Appendix 5).

The final cohort comprised 4,330 women, with their index pregnancy pair the unit of analysis.

151

Figure 8 Selection of a cohort of women smoking at delivery and had a subsequent pregnancy in NSW, 2008-2012

475,732 pregnancies of 372,385 women resulting in deliveries in NSW between 2008 and 2012

346,893 women excluded no record of smoking at the time of delivery in any pregnancy

36,585 pregnancies of 25,492 women who smoked at the time of delivery

24,108 pregnancies (19,292 women) excluded

18,746 no subsequent pregnancy 664 non-consecutive pregnancies 4,698 prior pregnancies to eligible pair

6,200 women with 6,616 consecutive pregnancy pairs

1,318 pregnancies (902 women) excluded 1,097 short inter-pregnancy interval Identifypregnancy pair 221 pairs after index pair

5,298 women with 5,298 index pregnancy pair

903 women excluded 8 unknown smoking outcome 895 missing data on potential confounder for quantity smoked during second half of first pregnancy

4,395 women with complete information on exposure, outcome and potential confounders

18 women excluded 15 varenicline and/or bupropion/NRT 3 varenicline 47 women excluded Furtherexclusions dispensed < 7 days 17 bupropion before conception 29 NRT date of second 1 bupropion and pregnancy NRT

547 varenicline- 3,783 unexposed exposed women women women 152

Table 18 presents the characteristics of the 4,330 included women at the time of their first pregnancy. Most maternal smokers delivered their first pregnancy in the pair prior to 2010 (77.6%), were younger than 35 years old (93.9%) and conceived their subsequent pregnancy less than 2 years after delivery of the first pregnancy (75.1%), with an average inter-pregnancy interval of 1.30 years (inter-quartile range: 0.88-1.99 years) (see Table 18). The majority were Australia-born (91.6%), smoked less than 10 cigarettes daily (62.4%) and had previous births (58.9%). Approximately half of the women lived with a partner and in disadvantaged areas. In the first pregnancy, 9.8% of women experienced an adverse event, 4.1% had hypertension, and 3.4% had diabetes while 32.8% of their newborns experienced an adverse event.

There were 547 women (12.6%) who were exposed to varenicline during the inter- pregnancy interval. Women exposed to varenicline more often lived in advantaged areas (34.6%) than in women who were unexposed to any pharmacotherapy (26.6%). A higher proportion of varenicline-exposed women conceived their second pregnancy at least two years after delivery of their first pregnancy (43.0% vs 22.3%) than those who did not use any pharmacotherapy.

In women exposed to varenicline, 64.5% received their first dispensing at about 9 months prior to conceiving their subsequent pregnancy. Almost two-thirds (63.8%) had one dispensing of varenicline only, equivalent to a 4-week supply. The recommended course for varenicline is 12 weeks.

153

Table 18 Characteristics of women who were current smokers at the time of delivery for the first pregnancy in the pair according to varenicline exposure status in the inter-pregnancy interval, NSW, 2008-2012 Women who were varenicline Characteristics Total Exposed Unexposed (N=4,330) (N=547) (N=3,783) n n Maternal age, years < 25 2,237 51.7% 263 48.1% 1,974 52.2% 25-<35 1,829 42.2% 259 47.3% 1,570 41.5% >= 35 264 6.1% 25 4.6% 239 6.3%

median 24 25 24

mean ± S.D. 25.0 ± 5.4 25.3 ± 5.0 25.0 ± 5.5 a IQR 21-29 22-29 21-29 Living with a partnerb 2,318 53.5% 328 52.6% 1,990 60.0% Australia-born 3,968 91.6% 517 94.5% 3,451 91.2% Remoteness of

residenceb Major cities 2,241 51.8% 302 55.2% 1,939 51.3% Regional & remote 2,030 46.9% 242 44.2% 1,788 47.3% b SES of residence disadvantaged 1,932 44.6% 223 40.8% 1,709 45.2% average 1,144 26.4% 132 24.1% 1,012 26.8% advantaged 1,195 27.6% 189 34.6% 1,006 26.6% Indigenous statusb 964 22.3% 77 14.1% 887 23.4% Quantity of cigarettes smoked in the second half of pregnancy < 10 2,704 62.4% 323 59.0% 2,381 62.9% >= 10 1,626 37.6 224 41.0% 1,402 37.1%

Parity Nulliparous 1,779 41.1% 239 43.7% 1,540 40.7% Multiparous 2,366 54.6% 293 53.6% 2,073 54.8% Grand multiparous 185 4.3% 15 2.7% 170 4.5% Year of first delivery 2008 1,968 45.5% 273 49.9% 1,695 44.8% 2009 1,391 32.1% 176 32.2% 1,215 32.1% 2010 753 17.4% 77 14.1% 676 17.9% 2011 218 5.0% 21 3.8% 197 5.2% Inter-pregnancy interval,

years 0.5-<1 1,420 32.8% 91 16.6% 1,329 35.1% 1-<2 1,831 42.3% 221 40.4% 1,610 42.6% 2-<3 805 18.6% 170 31.1% 635 16.8%

>=3 274 6.3% 65 11.9% 209 5.5%

median 1.30 1.81 1.25

mean ± S.D. 1.51±0.81 1.88 ± 0.84 1.46 ± 0.77 IQR 0.87-1.99 1.16-2.49 0.85-1.90

154

Women who were varenicline Characteristics Total Exposed Unexposed (N=4,330) (N=547) (N=3,783) n n Interval between dispensing and subsequent pregnancy, months <6 187 34.2% 6-<12 166 30.3% 12-<18 96 17.6% 18-<24 53 9.7% >=24 45 8.2% Median, days 263 IQR 137-456 Number of varenicline

dispensing 1 349 63.8% 2-3 187 34.2% >3 11 2.0% Any maternal adverse 423 9.8% 64 11.7% 359 9.5% events Any neonatal adverse 1,422 32.8% 159 29.1% 1,263 33.4% events Hypertension 176 4.1% 27 4.9% 149 3.9% Diabetes 104 2.4% 10 1.8% 94 2.5% aIQR: Inter-quartile range SES: Socio-economic status bthere were some missing data for variables ‘having a partner’ (0.6%), both ‘remoteness’ and ‘socio-economic status’ (1.4%), ‘Indigenous status’ (<0.1%).

155

Table 19 presents the relationship between exposure to varenicline during the inter- pregnancy interval and smoking cessation prior to the second pregnancy. Among the 4,330 women who smoked at the delivery of their first pregnancy, 23.0% (995 women) quit prior to their second pregnancy. Among the 547 women who were exposed to varenicline, 26.5% (145 women) quit prior to their second pregnancy.

Table 19 also presents the results of the main analyses, in which the only maternal morbidities adjusted for were pre-existing diabetes and pre-existing hypertension. After adjusting for potential confounders, women who were exposed to varenicline were statistically significant more likely to quit prior to their second pregnancy than unexposed maternal smokers (aOR 1.29, 95% CI 1.03-1.60).

156

Table 19 Association between varenicline exposure during the inter-pregnancy interval and smoking cessation prior to the second pregnancy Proportion quit smoking (N=4,330) OR 95% CI aORa 95% CI n Varenicline exposed No 850 / 3,783 22.5% 1.00 ref 1.00 ref Yes 145 / 547 26.5% 1.25 1.01-1.53 1.29 1.03-1.60 Maternal age, years < 25 492 / 2,237 22.0% 1.00 ref 1.00 ref 25-<35 437 / 1,829 23.9% 1.11 0.96-1.29 1.19 1.01-1.40 >=35 66 / 264 25.0% 1.18 0.88-1.59 1.36 0.97-1.91 Living with a partner No 388 / 1,987 19.5% 1.00 ref 1.00 ref Yes 603 / 2,318 26.0% 1.45 1.25-1.67 1.29 1.10-1.51 Australia-born No 114 / 362 31.5% 1.00 ref 1.00 ref Yes 881 / 3,968 22.2% 0.62 0.49-0.78 0.74 0.57-0.95 Indigenous status No 842 / 3,366 25.0% 1.00 ref 1.00 ref Yes 153 / 964 15.9% 0.57 0.47-0.68 0.77 0.63-0.95 Remoteness of residence* Major cities 557 / 2,241 24.9% 1.00 ref 1.00 ref Regional & remote 417 / 2,030 20.5% 0.93 0.79-1.09 0.96 0.86-1.08 SES of residence* disadvantaged 425 / 1,932 22.0% 1.00 ref 1.00 ref average 235 / 1,144 20.5% 0.92 0.77-1.10 0.93 0.77-1.12 advantaged 314 / 1,195 26.3% 1.26 1.07-1.50 1.11 0.93-1.33 Quantity of cigarettes smoked after 20 weeks of pregnancy <10 759 / 2,704 28.1% 1.00 ref 1.00 ref >= 10 236/1,626 14.5% 0.44 0.37-0.51 0.47 0.40-0.56 Parity Nulliparous 516 / 1,779 29.0% 1.00 ref 1.00 ref Multiparous 452 / 2,366 19.1% 0.58 0.50-0.67 0.59 0.50-0.70 Grand multiparous 27 / 185 14.6% 0.42 0.28-0.64 0.41 0.26-0.65 Year of first pregnancy delivery 2008 502 / 1,968 25.5% 1.00 ref 1.00 ref 2009 286 / 1,391 20.6% 0.76 0.64-0.89 0.72 0.60-0.85 2010 150 / 753 19.9% 0.73 0.59-0.89 0.70 0.56-0.87 2011 57 / 218 26.1% 1.03 0.75-1.42 0.75 0.53-1.06 157

Proportion quit smoking (N=4,330) OR 95% CI aORa 95% CI n

Inter-pregnancy interval, years <1 344 / 1,420 24.2% 1.00 ref 1.00 ref 1-2 409 / 1,831 22.3% 0.90 0.76-1.06 0.86 0.72-1.03 2-3 187 / 805 23.2% 0.95 0.77-1.16 0.81 0.65-1.02 >3 55 / 274 20.1% 0.79 0.57-1.08 0.62 0.44-0.88 Any maternal adverse event No 877 / 3,905 22.4% 1.00 ref 1.00 ref Yes 118 / 423 27.9% 1.34 1.07-1.68 1.09 0.86-1.38 Any neonatal adverse event No 713 / 2,908 24.5% 1.00 ref 1.00 ref Yes 282 / 1,422 19.8% 0.76 0.65-0.89 0.80 0.68-0.94 Hypertension No 949 / 4,154 22.8% 1.00 ref 1.00 ref Yes 46 / 176 26.1% 1.20 0.85-1.69 1.06 0.74-1.52 Diabetes No 965 / 4,226 22.8% 1.00 ref 1.00 ref Yes 30 / 104 28.8% 1.37 0.89-2.11 1.40 0.89-2.20 aaOR: adjusted odds ratio SES: Socio-economic status *Percentages do not add up to 100% due to missing data Potential confounders denoted in the grey area

158

Table 20 reports the results of the sensitivity analysis based on concessional beneficiaries. The effect estimate increased by more than 20% when adjustments was made for eight additional maternal morbidities (see Table 20). This change in the effect estimates suggests a strong possibility of residual confounding in the main analysis.

Table 20 Sensitivity analysis in examining association between varenicline dispensing during inter-pregnancy interval and smoking cessation prior to the second pregnancy, concessional beneficiaries

Number of women who quit smoking prior to their second pregnancy (%) Total Exposed to Unexposed to aORa (95% CI) Varenicline varenicline

Main analysis: All women who had two 145 / 547 850 / 3,783 4,330 1.29 (1.03-1.60) consecutive pregnancies (26.5) (22.5) in NSW, 2008-2012

All concessional beneficiaries who had two consecutive 71 / 317 2,421 1.55 (1.14-2.10) pregnancies in NSW (18.3) 371 / 2104 between 2008-2012, (17.6) adjustment for additional maternal morbiditiesb aaOR: adjusted odds ratio badditional adjustment for eight maternal morbidities: disorders relating to drugs and alcohol, mood, psychosis, anxiety, respiratory, gastro-oesophageal reflux, dispensing of non-steroidal anti-inflammatory medications, and dispensing of steroids

159

Discussion

This population-based cohort study found that among women who smoked at the time of their first delivery, those who were dispensed varenicline within their inter-pregnancy interval were more likely to quit smoking prior to their subsequent pregnancy compared to those who did not use any subsidised smoking cessation pharmacotherapy. This occurred despite 63.8% of women dispensed varenicline receiving only a single dispensing (equivalent to four weeks supply), and on average receiving their first varenicline nine months prior to conceiving their next pregnancy. However, as the study did not have information on all relevant variables including motivation to quit and receipt of behavioural support, this observed effect must be considered in comparison with other studies in the general population, and the possibility of residual confounding.

To my knowledge, this is a first study that addresses the inter-pregnancy effectiveness of smoking cessation pharmacotherapy among women smoking after giving birth. The findings suggest that varenicline is more effective than receiving no smoking cessation pharmacotherapy in the postpartum period.

Although there are several important methodological limitations to this study, the conclusion drawn is plausible given that the findings are consistent with the results of studies conducted in the general populations in the UK, Canada, Australia and the US [395, 396, 404]. In particular, varenicline has been found to be even more effective among female smokers than male smokers (women’s pooled OR 4.96, 95% CI 3.78- 5.49 vs men’s pooled OR 3.43, 95% CI 2.87- 4.11; significant interaction between gender and varenicline) [405].

Moreover, in the general population, varenicline is found to be the most effective smoking cessation intervention available, as detailed in Chapter 1 (section 1.3.2 Efficacy in the general population). Briefly, it is more effective than psychosocial interventions [201] and placebo [77]. Varenicline has superior effectiveness to bupropion [77], and all single forms of NRT [77].

160

This study has several notable limitations. As a result of using linked administrative data sources, this study relied upon the assumption that entries of ‘non-smoking’ in the perinatal records represented smoking cessation. This imperfect identification of smoking cessation was based on self-reported smoking status, where biochemically validated smoking status would be preferable [406]. As discussed in Section 4.5.3, previous validation studies demonstrated that while sensitivity of smoking information in the perinatal data was 89.6% and 66.3% in hospital data, the specificity was high (>93%) [374, 375]. This suggests that smoking may be underreported but is likely to be accurate when present. In addition, the misclassification in the smoking cessation outcome was minimised by ascertaining smoking status from two linked sources, the perinatal and hospital data, as outlined in Section 5.3.4. This use of a proxy in place of direct measurement of abstinence, may have resulted in some differential misclassification in the study outcome. The lack of information on other potential quit attempts, including both assisted (such as behavioural support) and unassisted ones, means that the comparison group is relatively undefined. Similarly, the lack of information on the provision and/or receipt of behavioural support among varenicline- treated women may have introduced some heterogeneity in the measurement of the treatment of interest, reducing the precision with which varenicline’s effectiveness was measured. Exposure to over the counter formulations of NRT or NRT patches obtained through sources apart from the PBS could not be identified in this study. Moreover, the duration of follow-up varied between women, as it was defined by the duration of their inter-pregnancy interval. Therefore, the opportunity to quit smoking, and the risk of relapsing after a quit attempt is not equal for all maternal smokers.

The findings of this study are generalisable only to women who disclosed the number of cigarettes smoked in their first pregnancy. This study only included women with complete data on this variable because the quantity of cigarettes smoked daily is likely to be an important confounder in the relationship between attempting to quit smoking using pharmacotherapy and smoking cessation. A comparison of the characteristics of included and excluded women indicated that excluded women were more likely to be older than 25 years old, have a partner, be overseas-born and live in major cities as well as in advantaged areas (see Appendix 5). These characteristics are consistent

161

with those reported by other studies which found that women who disclosed their smoking status were different than those who did not [303, 407]. Given these systematic differences between the women included in the analysis and those excluded, the study findings may not be generalisable to all women who smoked at the time of delivery of their first pregnancy and who subsequently conceived again.

Another limitation was the use of a single dispensing record to indicate varenicline use. It may be possible that a woman who received a single dispensing may not have used the varenicline that she was dispensed. This misclassification of exposure to varenicline would shift the estimate of effectiveness towards the null. This study could not rule out the possibility of residual confounding. In particular, this study was unable to use women dispensed bupropion or NRT patches as a comparator group due to a small number of women, thereby giving rise to the possibility of confounding by indication (severity of nicotine dependence). However, this study adjusted for the number of cigarettes smoked to reduce the likelihood of confounding by nicotine dependence. Although this study did not adjust for additional maternal morbidities in the main analysis because it would substantially reduce the size of the cohort, additional potential confounders were included in sensitivity analyses based on women who were concessional beneficiaries. The results of the sensitivity analyses (see Table 20) show that the effect size increased by more than 20% when adjustment was made for the additional eight maternal morbidities. The increase in the estimate as a result of additional statistical adjustment suggests a strong possibility of residual confounding in the main analysis. The magnitude and direction of the estimate derived in the sensitivity analysis implies that adjustment for this residual confounding results in an enhanced effect of varenicline. The estimates derived from the main analyses may, thus, reflect a conservative estimate of varenicline’s effect on smoking cessation.

The findings from this study are indicative of varenicline effectiveness when used by women smoking in the inter-pregnancy interval. Given the study’s methodological limitations, further investigation is warranted. Should these findings be confirmed, this

162

research has several important implications for managing postpartum smoking cessation among women who smoke up to the end of their pregnancy.

First, further research investigating varenicline’s safety among maternal smokers is warranted. This is important as concerns for varenicline’s safety while breastfeeding may limit its use among postpartum maternal smokers who otherwise may have higher success with quitting smoking.

Next, further examination of whether NRT is effective during the postpartum period is necessary, with a focus on whether NRT is more effective than varenicline. This is important as during the postpartum period, NRT is not contraindicated for use, whereas the safety of varenicline has yet to be established, and thus varenicline is not recommended for use in the current lactation guidelines [180]. The findings regarding postpartum effectiveness of NRT will be useful as it will inform decisions about which smoking cessation pharmacotherapy to use in promoting smoking cessation after delivery.

The final implication points to an overall need for greater efforts to promote smoking cessation among women who continue to smoke after giving birth. This is important as this study revealed that many maternal smokers did not use any smoking cessation pharmacotherapies during the inter-pregnancy interval. There remains a large population of women who smoke after giving birth and subsequently went on to have another pregnancy. In practice, these smokers could be potentially targeted by increasing their uptake of effective interventions to assist them to quit smoking, especially using effective pharmacotherapy, including varenicline.

This population-based study provides initial evidence of varenicline’s effectiveness when dispensed during the inter-pregnancy interval among women who smoked at delivery. This result holds after adjusting for key potential confounders, particularly,

163

nicotine dependence. As the results of this study were likely to be subject to residual confounding, further research with greater capacity to measure exposure to other smoking cessation interventions in both the exposed and unexposed women is necessary to confirm these results. Data on the safety of varenicline among women who breastfeed are also required. If the effectiveness and safety of varenicline are demonstrated, these results could inform substantial changes to the management of postpartum maternal smoking.

*For studies described in Chapter 4 and 5, acknowledgements are made to the NSW Centre for Health Record Linkage, the Australian Institute for Health and Welfare, the Western Australia Data Linkage Branch, as well as data custodians of the NSW Perinatal Data Collection, WA Midwife Notification System, NSW Admitted Patient Data Collection, WA Hospital Morbidity Data Collection, NSW Registry of Births, Deaths and Marriages, WA Registry of Births, Deaths and Marriages, and Pharmaceutical Benefits Scheme data for allowing access to the data and conducting the linkage of records.

164

Chapter 6 Discussion

This chapter discusses the thesis as a collective body of research that aimed to identify factors influencing the use of pharmacotherapy for smoking cessation during and after pregnancy. This chapter discusses the key findings of the thesis within the context of the strengths and limitations of the research undertaken. This chapter also draws attention to the implications of the thesis findings and makes recommendations for future studies that would further advance knowledge in the field of smoking cessation pharmacotherapy use during and after pregnancy.

Review of the thesis aims This thesis sought to examine factors that influence the prescribing of smoking cessation medications to pregnant women who smoke. The examined factors included healthcare providers’ safety concerns regarding these pharmacotherapies in pregnant women, and facility-level and individual-level factors associated with familiarity with, and the intention to prescribe, smoking cessation pharmacotherapies to pregnant women.

As many women who smoke during pregnancy continue to smoke after giving birth, this thesis also attempted to improve understanding of the factors that are, or ought to be, considered when deciding whether smoking cessation pharmacotherapies should be used in this population of postpartum smokers. This thesis investigated whether the suitability of these pharmacotherapies for individual women, in particular among women with pre-existing health conditions and women who experienced poor birth outcomes, is considered in the decision to use smoking cessation pharmacotherapy after giving birth. This thesis also examined whether smoking cessation pharmacotherapies are effective in this population.

165

166

Summary and interpretation of the main findings of the thesis The findings of each study are summarised in its respective chapter, along with an interpretation of those findings. This section, therefore, focuses on reviewing the findings that address the two overarching thesis aims.

The first two studies in this thesis (Chapters 2 and 3) identified factors that could be the focus of interventions designed to encourage more prescribing of, and thus, greater uptake of pharmacotherapy among pregnant women who smoke. The focus was on factors that influence healthcare providers’ receptiveness to prescribe these medications to women who are pregnant.

The study described in Chapter 2 attempted to measure the extent of concern that the healthcare providers had regarding smoking cessation pharmacotherapies compared to other medications in the same and other risk categories during pregnancy. This study was based on calls to a teratology information service regarding medications during pregnancy; these calls were taken as a surrogate indicator of concern regarding the safety of the medication when used during pregnancy.

The study found that healthcare providers were more concerned about bupropion than other medications of the same risk category, B2 (aOR 2.77, 95% CI 1.17-6.59). Medications categorised as risk category B2 comprise those with inadequate safety evidence from both human and animal studies [120]. Similarly, a greater level of concern was shown towards varenicline compared to medications in risk category B3 (varenicline’s risk category), (aOR 2.33, 95% CI 1.30-4.17). Medications in risk category B3 have an increased frequency of foetal harm in animal data but human data are lacking [120].

Healthcare providers were equally concerned about NRT and category D medications (aOR OR 1.20, 95% CI 0.93-1.54). However, they were more concerned about NRT than lower-risk category medications (risk category A, B1, B2, B3, and C). As maintained in Chapter 2, category D may not be an appropriate comparison category of

167

risk for NRT, as the general consensus [408] and emerging research [88] agree that NRT is safer for pregnancy than continued smoking. Thus, the finding that NRT is accorded the same level of concern as medications that are thought to cause irreversible harm to the foetus implies that healthcare providers may not account for the established risk of harm associated with smoking during pregnancy.

These findings suggest that healthcare providers may be overestimating the risk associated with smoking cessation pharmacotherapies during pregnancy. In addition to the cautions in clinical practice guidelines,this may contribute to reluctance to prescribe these pharmacotherapies to pregnant women. This reluctance offers a potential partial explanation of the low rates of NRT patches, bupropion and varenicline use among pregnant women in Australia [110].

There are also a range of other factors that potentially limit healthcare providers’ receptiveness to prescribe smoking cessation pharmacotherapies to pregnant women. This was explored in the study described in Chapter 3. A survey was undertaken with a key population of obstetric care providers, obstetricians and gynaecologists, in order to examine factors associated with their knowledge of, and their intention to prescribe, pharmacotherapies for smoking cessation to pregnant women. The study first examined whether facility-level factors were associated with obstetricians’ familiarity with these pharmacotherapies. Among those who were familiar with these pharmacotherapies, the study also determined whether these same facility-level factors as well as individual-level factors influenced the intention to prescribe these medications to pregnant women.

The findings of this study may not be generalisable to the entire population of obstetricians and gynaecologists in Australia due to the low response rate (7.5%), and the observed differences in the characteristics of survey respondents and the general population of obstetricians and gynaecologists practising in Australia. In this selected sample, slightly more than half of the sample reported familiarity with NRT patches (57.1%, 95% CI 49.5-64.7), and very few were familiar with varenicline (8%, 95% CI 3.8-12.1). The suboptimal familiarity is likely to limit the prescribing of pharmacotherapies for smoking cessation among obstetricians and gynaecologists. This study found that obstetricians practising in a facility with a smoking cessation

168

protocol (aOR 2.85, 95% CI 1.29-6.25), and facilities located in the rural and regional areas (aOR 2.56, 95% CI 1.06-6.21) were more likely to be familiar with NRT patches. Factors related to bupropion were not investigated as the use of bupropion in the Australian general population in recent years has been low, precluding it from being the subject of observational research on prescribing behaviour. This study was unable to explore factors pertaining to varenicline due to few respondents indicating their intention to prescribe it (n=3), and their familiarity with varenicline (n=13).

This study observed a suboptimal level of intention to prescribe NRT patches among obstetricians who were familiar with this medication. Among obstetricians who were familiar with NRT patches, 61.1% (95% CI 51.4-71.2) had intention to prescribe these patches. This finding reaffirms prior evidence that enhancing the knowledge of obstetricians alone is insufficient to increase their prescribing of these pharmacotherapies. The limited intention to prescribe among obstetricians who were familiar with NRT patches indicates that there are factors apart from knowledge that influence the intention to prescribe NRT patches to pregnant women.

Among obstetricians who were familiar with NRT patches, this study identified potential relationships between facility-level factors including the obstetricians’ level of training, and funding status of the facility in which obstetricians practised, and their intention to prescribe NRT patches. The imprecise estimates, as indicated by the wide confidence intervals, however, mean that this study is unable to be definitive about its findings.. Among obstetricians who were familiar with NRT patches, those who believed that NRT patches would benefit the light smokers among their patients, as well as smokers who had previous failed quit attempts, were more likely to intend to prescribe NRT patches. Obstetricians who were aware that NRT patches are PBS-listed and those that did not find it hard to assess the risk-benefit profile of NRT patches were also more likely to prescribe NRT patches. Thus, obstetricians’ beliefs regarding the perceived benefits for their patients, and their perceived self-efficacy to prescribe NRT patches appear to be important individual-level factors relating to the intention to prescribe NRT patches.

169

The study reported in Chapter 4 measured the extent of smoking cessation pharmacotherapies use among women who smoke after pregnancy in Australia, including whether the opportunity to quit early during the postpartum period is seized. It also investigated whether suitability (pre-existing health conditions, birth outcomes) is considered in the decision to use smoking cessation pharmacotherapies after giving birth.

The study described in Chapter 4 was based on linked data comprising perinatal records, hospital admission records, death records and pharmaceutical dispensing records for all women who gave birth in NSW and WA between 2011 and 2012. The study assumed that women who had a current smoking status documented in their hospital records at the time of delivery were likely to smoke beyond birth.

The findings from the study in Chapter 4 revealed that among women who were current smokers at the time of delivery, 8.7% were dispensed with a smoking cessation pharmacotherapy in the 12 months postpartum. Varenicline was the most frequently used pharmacotherapy compared to NRT patches, 6.3% vs 2.2%. Although the use of pharmacotherapy during the postpartum period was higher than that of during pregnancy (8.7% vs 2.3-3.6% [110]) this postpartum estimate remained low. This finding also suggests that the opportunity to use pharmacotherapy to assist with smoking cessation in the postpartum period is not seized by enough maternal smokers in Australia.

Fewer women used varenicline in the first 8 weeks after giving birth (0.88%), compared to later periods of equivalent durations. NRT use was higher during the immediate postpartum period (0.58%) than other periods of equivalent duration, likely because it is the only recommended pharmacotherapy during breastfeeding. Both findings are reassuring, given the higher likelihood of breastfeeding at this immediate postpartum period and that varenicline is not recommended during breastfeeding.

170

While the use of smoking cessation pharmacotherapies should be encouraged and supported in this postpartum population, the use may not be justified in all women during the postpartum period. This is in part, due to comorbid health conditions that may predispose the women to an increased risk of adverse events associated with contraindications, side-effects, and/or medication interactions with smoking cessation pharmacotherapies. For this reason, Chapter 4 also examined the relationship between pharmacotherapy use and a range of maternal co-morbidities. This was done to provide insight into whether suitability of these pharmacotherapies is taken into account when making decision to use or not use a pharmacotherapy in the postpartum period. This study revealed that, for most of the co-morbidity/ies examined, the likelihood of using pharmacotherapy did not differ between maternal smokers with and without the co-morbidity(s). The only exceptions were that maternal smokers with mood disorders were more likely to use varenicline (aOR 1.99, 95% CI 1.64-3.13) and NRT (aOR 1.40, 95% CI 1.09-1.79), and women who previously used NSAIDs (aOR 1.47, 95% CI 1.03- 2.10) were more likely to use varenicline in the 12 months postpartum. The reason for the observed relationship between previous NSAIDs use and varenicline use in the 12 months postpartum is unclear. Without information on the indication for which NSAIDs were prescribed, any likely explanation to the observed relationship is speculative and thus requires further investigation.

In contrast, smoking cessation pharmacotherapies may be considered more suitable for women who experience poor birth outcomes including adverse maternal and/or neonatal events, given the potential benefits of smoking cessation pharmacotherapies are amplified in such women. Chapter 4 found no evidence that experiencing poor birth outcomes, whether maternal or neonatal events, was associated with the likelihood of using pharmacotherapy after delivery. However, there may be evidence that adverse maternal events increased the likelihood to use varenicline after giving birth (aOR 1.36, 95% CI 0.98-1.88).These findings suggest that the opportunity to support women who could benefit from quitting smoking, due to their morbidities (including respiratory disorders, diabetes, hypertension and others) and their experience of poor birth outcomes (maternal and neonatal), is being missed in the postpartum period.

A key factor likely to influence the decision to use smoking cessation pharmacotherapies is their effectiveness in the population of interest. An investigation

171

of whether pharmacotherapy is effective after giving birth was carried out in the study described in Chapter 5. This study used the same linked administrative data described in Chapter 4 to investigate whether use of smoking cessation pharmacotherapy during the inter-pregnancy interval was associated with smoking cessation prior to the subsequent pregnancy. Among women smoking at the time of birth, this study found that varenicline use during the inter-pregnancy interval was effective in assisting a woman to quit smoking prior to the next pregnancy. Compared to the women who smoked and did not use any pharmacotherapy during the inter-pregnancy interval, women who used varenicline were more likely to quit prior to their second pregnancy (aOR 1.29, 95% CI 1.03-1.60). This finding calls for considerable effort to encourage greater uptake of pharmacotherapy for smoking cessation after giving birth. This study was unable to determine whether bupropion and NRT patches are effective during the postpartum period as few women used bupropion and NRT patches in the inter- pregnancy interval.

172

Strengths and limitations of the thesis Given the strengths and limitations of each study are discussed in their respective chapters, this section focuses on how they apply to the overall thesis. The thesis findings and their interpretation must be considered within the methodological limitations, and the interpretation made accordingly.

A key strength of the research reported in this thesis is that it employed different methods to address the thesis aims. In particular, this thesis combined primary data collection and secondary analysis of existing data sources. This mixed-method approach allowed for an examination of a wider range of factors that potentially influence the low uptake of pharmacotherapy for smoking cessation during pregnancy and the postpartum period as compared to previous research.

One of the key strengths of using a primary data source, such as through a survey (Chapter 3), is that it allows the data collection tool to be designed to directly solicit the information required to answer the research question. However, not all potentially relevant factors were measured in the survey. Questions relating to training in smoking cessation were not included, and these questions may influence one of the study factors regarding self-efficacy to prescribe smoking cessation pharmacotherapies. Furthermore, the poor response rate from the survey in Chapter 3 and the finding that the survey respondents did not represent the entire population of obstetricians and gynaecologists in Australia means that response bias was introduced in the study. The findings from this survey might not apply to those who did not respond to the survey.

While this risk of selection bias was apparent in the use of primary data sources, this risk was minimised in those studies using routinely-collected data. This thesis used two sources of routinely collected data; records of calls made to a teratology information service (Chapter 2) and linked administrative health services data (Chapters 4 and 5).

Using data from routinely-collected sources minimises the risk of selection bias as the data source covers the entire population that uses the said service. In experimental research, participants whose characteristics meet the strict eligibility criteria are selected [233, 409]. This may mean that, for example in trials examining the role of mental health disorders in influencing the use of these quit smoking medications,

173

women who present with comorbidities such as cardiovascular diseases, are excluded. Furthermore, the inclusion of health care records and outcomes as they occur reflects health and healthcare in real-world settings, as opposed to that measured in trials and community-based studies. Importantly, the voluntary nature of many experimental studies and other primary data collection studies means that women who participated are more likely to possess a high level of motivation to quit, which may not represent the general population of smokers [410, 411].

Although the findings of this thesis are generalisable to the users of services represented in the data sources used, there are ways in which these routinely collected data are limited in their generalisability. First, the population of TIS callers (Chapter 2) may not be representative of all concerned individuals as they may differ in terms of their information-seeking behaviour, access to alternative resources, and personality traits [232]. Second, the findings of the studies reported in Chapters 4 and 5 might not apply to the women whose births were not captured in the perinatal records including those resulting in miscarriage before 20 weeks of gestation or where the foetus weighed less than 400g. Given miscarriage is an important outcome of smoking during pregnancy [7-9] , this may introduce the risk of selection bias in the study population of women who smoke at delivery. Third, the lack of dispensing data for general beneficiaries prior to July 2012 limits the extrapolation of some of the thesis findings beyond concessional beneficiaries, where concessional beneficiaries were demonstrated to differ from all beneficiaries on some important characteristics (see section 4.5.3 Limitations). Third, by excluding women who did not have information on nicotine dependence in Chapters 4 and 5, the thesis findings are limited to women whose information on nicotine dependence was reported.

One strong advantage of using secondary data sources is that it allows a timely and cost-efficient manner to study the temporal relationship between study factors and outcomes (Chapters 4 and 5). As evidenced by the review of literature in the introductory chapter (Chapter 1), few pregnant women and postpartum smokers used smoking cessation pharmacotherapies. To meaningfully explore the factors associated with the use of these pharmacotherapies during the postpartum period, a relatively large number of women using smoking cessation pharmacotherapies is needed, and this requires the study to be based on a large study sample. In the study described in

174

Chapter 2, the 16-year study period allowed for the investigation of the relationship between the calls regarding smoking cessation pharmacotherapies and those of other medications, including infrequently used medications which are an important focus of medication safety research due to their strong risk of causing irreversible harm to the foetus (category D and X medications). Using primary data collection for this purpose is unlikely to be feasible as it would come at a high financial and time cost.

Using routinely-collected data, however, has limitations. As the data sources were not designed for the research reported in this thesis, existing variables were used as proxy measures for the study exposures, outcomes, and potential confounders. Calls made to a TIS were taken as a surrogate indicator of concern (Chapter 2), women who were dispensed with pharmacotherapy were assumed to use these medications (Chapters 4 and 5), and nicotine dependence was measured by the number of cigarettes smoked (Chapters 4 and 5). In the study described in Chapter 5, smoking cessation following varenicline use in the inter-pregnancy interval was defined as not smoking in the subsequent pregnancy, in place of the direct measurement of smoking abstinence following the use of varenicline. This thesis identified maternal morbidities from clinical encounters that resulted in the recording of a diagnostic code and/or dispensing of a related medication (Chapter 4). This definition might result in under-ascertainment of many maternal morbidities especially mental health disorders and substance use disorders because identification of these disorders requires hospitalisation or prescribing of medications that arise from acute and/or severe manifestations of the illnesses. The use of these proxies may have led to some misclassification of study exposures, outcomes and potential confounders.

A related drawback of routinely collected data is the absence of some important variables. Although this thesis was able to adjust for a large number of potential confounders, there remains the likelihood of residual confounding. There was no information available on women’s educational status and their partners’ smoking status (Chapters 2, 4 and 5). Both variables uniquely contribute to the likelihood of continuing smoking during pregnancy and after giving birth [29, 412]. Due to ethical reasons, there was no identifying information made available for the callers (Chapter 2). This thesis, therefore, could not adjust for potential clustering of calls within callers.

175

One key limitation of some studies in this thesis was the lack of statistical power to carry out analyses as set out in the thesis aims due to the small number of participants in each level of the key exposures, outcomes, and potential confounders. In the survey (Chapter 3), the high non-response rate reduced the power of the study to detect any observed relationship between the study factors and the study outcome. This precluded all planned analyses pertaining to varenicline. The desired granularity to investigate obstetricians who intend to prescribe compared to those not intending to prescribe was not achieved. In its place, the study combined the ‘unsure’ and ‘did not intend to’ responses into a single ‘did not intend’ response. In addition, it was not possible to adjust for potential confounders when examining the relationship between obstetricians’ beliefs and intention to prescribe these pharmacotherapies. Qualitative studies, including interviews and focus groups, were not carried out to provide data that would complement those from the survey. These methods could provide additional insight into factors associated with prescribing smoking cessation pharmacotherapies to pregnant women. Insufficient statistical power also arose in Chapter 5 whereby too few women used NRT patches to allow for examination of the effectiveness of NRT patches.

176

Implications for clinical practice and for future research Maternal smoking is the leading modifiable risk factor for maternal, neonatal and infant morbidity and mortality. Contacts with antenatal and postnatal care providers present an opportunity to support smoking cessation by delivering smoking cessation interventions to maternal smokers. As acknowledged in the introductory chapter (Chapter 1), pharmacological smoking cessation interventions, which should be accompanied with behavioural support, may not be indicated in all women who smoke during pregnancy and in the postpartum period. Yet, it is increasingly recognised that NRT may benefit pregnant women who have not been successful with non- pharmacological interventions. Moreover, there may soon be calls for wider use of varenicline and/or bupropion by pregnant smokers, given recent evidence of their safety [127, 129], and the improved birth outcomes for women who use varenicline, in particular [127]. There is an opportunity to increase quit smoking rates and improve birth and infant outcomes by increasing the use of smoking cessation pharmacotherapies among pregnant and postpartum women.

Beyond pregnancy, greater use of pharmacotherapies is likely to be beneficial as evidence now exists showing that varenicline is effective among postpartum smokers (findings from Chapter 5). This will help reduce second-hand smoke exposure to newborns, and help improve health outcomes for mothers, especially among women whose comorbid health conditions are worsened by continued smoking. Although there are concerns regarding the safety of these pharmacotherapies while breastfeeding, this is unlikely to account for much of the non-use of smoking cessation pharmacotherapies during the 12-month postpartum period; Chapter 4 found that only 8.7% of maternal smokers used a smoking cessation pharmacotherapy.

To increase uptake of smoking cessation pharmacotherapy during and after pregnancy, it is necessary to first identify and understand which factors influence the prescribing, and use of smoking cessation pharmacotherapies in these two important life stages of maternal smokers. The findings generated from this thesis open up new questions for future research.

Safety concerns regarding the use of pharmacotherapies for smoking cessation during pregnancy is a recommended focus for future interventions aiming to increase

177

prescribing of these pharmacotherapies to pregnant women. Findings from Chapter 2 suggested that healthcare providers are overly concerned about varenicline and bupropion compared to other medications in the same risk categories, while NRT was accorded the same level of concern as teratogenic medications. The concerns for NRT safety in pregnant women are consistent with other studies that were carried out among antenatal care providers including obstetricians, GPs, and midwives [215, 217, 413]. Yet, recent studies demonstrate that the use of NRT is safer than smoking [88]. Moreover, recent, high-quality population based studies did not show clear evidence of harm from using bupropion and varenicline during pregnancy [127, 129]. Efforts are needed to gather more robust safety data regarding the use of these pharmacotherapies during pregnancy to alleviate safety concerns, in particular the concerns for pregnancy outcomes such as miscarriages, specific congenital malformations, and long-term child health and development.

Recommendations in smoking cessation care guidelines should be consistent with the information conveyed through the pregnancy risk categories to support greater use smoking cessation pharmacotherapies during pregnancy. Guidelines recommend that NRT be used in pregnant women who are unable to quit smoking [57, 61, 108, 130]. However, in Australia’s TGA pregnancy risk categorisation system, NRT is categorised as a medication that is likely to cause irreversible foetal harm, due to the presence of nicotine (pregnancy risk category D [120]). This may lead healthcare providers who are guided by these pregnancy risk categories to overestimate the risk arising from NRT use in pregnancy (findings from study described in Chapter 2). Therefore, the use of TGA pregnancy risk categories should be accompanied by a descriptive narrative that acknowledge the risk of continued smoking against the risk of NRT use, consistent with smoking cessation guidelines. In contrast, the guidelines’ prohibitory stance on the use of bupropion and varenicline are not consistent with the pregnancy risk categories for these pharmacotherapies. Current guidelines recommendations advise against the use of bupropion and varenicline during pregnancy due to the lack of evidence on the benefit and risks, without adequately recognising that evidence demonstrating harm is also lacking [57, 61, 108, 130]. In contrast, the pregnancy risk categories for bupropion and varenicline acknowledge that human and animal data are inadequate to demonstrate risk. Therefore, the guidelines should provide a more balanced representation of the lack of evidence regarding safety as well as harm. This may

178

support greater use of bupropion and varenicline especially in high-risk pregnancies where the established risks of continuing smoking may outweigh the lack of evidence of harm arising from using these pharmacotherapies during pregnancy.

In contrast, the findings from Chapter 3 indicated that safety concerns were not related to the obstetricians’ and gynaecologists’ intention to prescribe NRT patches among those who were familiar with pharmacotherapies. Notably, this finding is inconsistent with other studies carried out among obstetricians which cited safety concerns as a barrier to prescribing NRT [215, 219]. A key point that differentiates between the study reported in Chapter 3, and previous studies showing discrepant findings, is that the analyses in Chapter 3 were restricted to obstetricians who were familiar with NRT patches. This implies that the concerns regarding the safety of smoking cessation pharmacotherapies may not be the prevailing factor among obstetricians who possess sufficient level of knowledge of smoking cessation pharmacotherapies. Research into whether safety concerns for smoking cessation pharmacotherapies during pregnancy is still salient to other categories of obstetric care providers including GPs, midwives, and pharmacists, among those who are familiar with smoking cessationpharmacotherapies, is needed. This may include qualitative research comprising interviews with the abovementioned groups of obstetric care providers.

This raises the possibility that low level of knowledge regarding smoking cessation pharmacotherapies may be one of the factors limiting the use of these pharmacotherapies during pregnancy. As demonstrated in the study described in Chapter 3, most of the surveyed obstetricians had limited familiarity regarding the newer pharmacotherapies (bupropion and varenicline). Indeed, it has been shown that increasing knowledge is an essential element of any intervention that aims to improve clinical practice [414]. While correcting the misperception of risk calls for improving the knowledge base of healthcare providers about pharmacologic therapies for smoking cessation during pregnancy, the findings from Chapter 3 revealed that there remains a suboptimal intention to prescribe despite being familiar with pharmacotherapies. Future studies are therefore necessary to examine whether improving the knowledge base regarding smoking cessation pharmacotherapies among healthcare providers will translate to an improved prescribing of these pharmacotherapies to pregnant women. If it is determined that improving the knowledge base is warranted, the next step will be

179

to establish how the knowledge gaps should be addressed. This may require a content review of smoking cessation curriculum and training in obstetrics programs to ascertain whether it might be improved. Alternatively, information-based training may be delivered using incentives including continuous medical education points.

Research on the safety of smoking cessation pharmacotherapies during the postpartum period might also help increase the uptake of these pharmacotherapies among postpartum maternal smokers. Although this thesis did not set out to measure whether these pharmacotherapies are safe in women after giving birth, concerns for the safety of these pharmacotherapies during breastfeeding might account for the low use of varenicline in the immediate 8 weeks after giving birth (findings from Chapter 4). In light of limited data on the safety of bupropion and varenicline during lactation, concerns regarding their safety while breastfeeding is plausible.

An overall consideration of the evidence generated in this thesis reveals an additional factor that is potentially contributing to the limited use of pharmacotherapy during and after pregnancy. There may be inadequate consideration of the known risks of continued smoking when assessing the risks and benefits of prescribing or using smoking cessation medications during and after pregnancy. This appeared to be the case in the study reported in Chapter 2, where the magnitude of concerns regarding NRT was similar to that for teratogenic medications. It was also apparent in Chapter 4 which revealed that women with pre-existing health conditions and women who experienced poor birth outcomes, were not more likely to use a pharmacotherapy postpartum, despite being most likely to benefit from smoking cessation. It can be argued that the known risk of continuing smoking may be insufficient to overcome the concern arising from the unknown safety of smoking cessation pharmacotherapy after pregnancy. Yet, this seems unwarranted as Chapter 5 indicated that pharmacotherapy including varenicline is effective among postpartum maternal smokers. While this insufficient smoking risk recognition can be improved by means of appropriate education and training of healthcare providers [415], gaps remaining include an understanding of how these providers assess the overall risk-benefit of smoking cessation pharmacotherapies and what information is used to guide the prescribing behaviour, especially among those healthcare providers who treat women with comorbid health conditions. These questions should be the focus of future studies.

180

Although this thesis focuses on cigarette smoking as it is the predominant form of tobacco use, the significance of this thesis findings may also apply to other tobacco products including smokeless tobacco, waterpipe/hookah, and electronic cigarettes. Women who use these tobacco products may benefit from using pharmacotherapy to quit, given the neuropharmacological effects of nicotine and nicotine withdrawal are addressed by pharmacotherapy. However, further research involving the users of these tobacco products, particularly electronic cigarettes, are needed.

181

Conclusion Using different data sources, this body of work identified a range of factors to target in interventions that encourage better uptake of pharmacotherapy for smoking cessation among pregnant women and women who smoke after giving birth. Evidence for misplaced concerns regarding these pharmacotherapies may limit their prescribing among healthcare providers. Among one key group of providers (obstetricians and gynaecologists), there is also evidence of an association between facility-level factors and lack of knowledge regarding NRT patches, and of a relationship between individual-level factors and suboptimal intention to prescribe among those who are familiar with NRT patches. This thesis also indicates that there appears to be a lack of judicious use of pharmacotherapies among women with a favourable risk-benefit profile after giving birth. This thesis also contains the first study that provides data on varenicline effectiveness in addressing postpartum maternal smoking.

Notwithstanding the limitations of this thesis, the indicative findings generated from this thesis suggest there remains much scope to address important questions relating to the factors examined in this thesis. Safety concerns regarding the use of pharmacotherapies for smoking cessation during pregnancy are the recommended focus for future interventions aiming to increase prescribing of these pharmacotherapies to pregnant women. There may also be insufficient consideration of the known risks of continued smoking when assessing the overall risks and benefits of prescribing smoking cessation medications to pregnant women and women who smoke after giving birth. Future research opportunities should focus on gathering robust safety data regarding the use of these pharmacotherapies during pregnancy to determine the safety risks, specifically addressing the concerns for: (a) under-studied pregnancy outcomes such as those pregnancies resulting in miscarriages, termination and specific malformations, and (b) infant outcomes during breastfeeding. Information is required regarding whether concerns regarding the safety of smoking cessation pharmacotherapies during pregnancy are salient to the other obstetric care providers apart from obstetricians and gynaecologists who possess sufficient level of knowledge regarding smoking cessation pharmacotherapies.

182

In closing, the discussion throughout this chapter shows that although much has been accomplished to treat smoking during pregnancy and postpartum, there is room for improvement. This calls for optimising the provision of effective smoking cessation interventions. This thesis demonstrated that there is an opportunity to promote greater use of smoking cessation pharmacotherapy among pregnant and postpartum smokers on a broader scale than the current practice. This means that greater use should be supported among women who most likely to benefit from quitting smoking and among women with newborn infants to reduce risks associated with second-hand smoke exposure.

183

References 1. Kurti AN, Redner R, Lopez AA, Keith DR, Villanti AC, Stanton CA, et al. Tobacco and nicotine delivery product use in a national sample of pregnant women. Preventive Medicine. 2017;104:50-6. 2. Australian Institute of Health and Welfare. National Drug Strategy Household Survey 2016: detailed findings. Drug Statistics series no. 31. Cat. no. PHE 214. Canberra: AIHW; 2017. 3. Gould GS, Havard A, Lim LL, Group TPSiPE, Kumar R. Exposure to Tobacco, Environmental Tobacco Smoke and Nicotine in Pregnancy: A Pragmatic Overview of Reviews of Maternal and Child Outcomes, Effectiveness of Interventions and Barriers and Facilitators to Quitting. International journal of environmental research and public health. 2020;17(6):2034. 4. Roelands J, Jamison MG, Lyerly AD, James AH. Consequences of Smoking during Pregnancy on Maternal Health. Journal of Women's Health. 2009;18(6):867-72. 5. Chen HL, Cai JY, Zha ML, Shen WQ. Prenatal smoking and postpartum depression: a meta-analysis. Journal of Psychosomatic Obstetrics & Gynecology. 2019;40(2):97-105. 6. Wallin HP, Gissler M, Korhonen PE, Ekblad M. Maternal Smoking and Hospital Treatment During Pregnancy. Nicotine & Tobacco Research [Internet]. 2019. [cited 2019 20 August]. Available from: https://doi.org/10.1093/ntr/ntz137. doi:10.1093/ntr/ntz137. 7. Castles A, Adams EK, Melvin CL. Effects of smoking during pregnancy – five meta analyses. American journal of preventive medicine. 1999;16. 8. Pineles BL, Park E, Samet JM. Systematic Review and Meta-Analysis of Miscarriage and Maternal Exposure to Tobacco Smoke During Pregnancy. American Journal of Epidemiology. 2014;179(7):807-23. 9. Salihu HM, Wilson RE. Epidemiology of prenatal smoking and perinatal outcomes. Early Human Development. 2007;83(11):713-20. 10. Ankum WM, Mol BWJ, Van der Veen F, Bossuyt PMM. Risk factors for ectopic pregnancy: a meta-analysis. Fertility and Sterility. 1996;65(6):1093-9. 11. Rogers JM. Tobacco and pregnancy. Reproductive Toxicology. 2009;28(2):152-60. 12. Little J, Cardy A, Munger RG. Tobacco smoking and oral clefts: a meta-analysis. Bulletin of the World Health Organization. 2004;82(3):213-8. 13. Hackshaw A, Rodeck C, Boniface S. Maternal smoking in pregnancy and birth defects: a systematic review based on 173 687 malformed cases and 11.7 million controls. Human Reproduction Update. 2011;17(5):589-604. 14. Shah NR, Bracken MB. A systematic review and meta-analysis of prospective studies on the association between maternal cigarette smoking and preterm delivery. American Journal of Obstetrics and Gynecology. 2000;182(2):465-72. 15. Pereira PPdS, Da Mata FAF, Figueiredo ACG, de Andrade KRC, Pereira MG. Maternal Active Smoking During Pregnancy and Low Birth Weight in the Americas: A Systematic Review and Meta-analysis. Nicotine & Tobacco Research. 2017;19(5):497-505. 16. Marufu TC, Ahankari A, Coleman T, Lewis S. Maternal smoking and the risk of still birth: systematic review and meta-analysis. BMC public health. 2015;15(1):239. 17. DiFranza JR, Aligne CA, Weitzman M. Prenatal and Postnatal Environmental Tobacco Smoke Exposure and Children’s Health. Pediatrics. 2004;113(Supplement 3):1007-15. 18. Malanchini M, Smith-Woolley E, Ayorech Z, Rimfeld K, Krapohl E, Vuoksimaa E, et al. Aggressive behaviour in childhood and adolescence: the role of smoking during pregnancy, evidence from four twin cohorts in the EU-ACTION consortium. Psychological medicine. 2018;49(4):646-54. 19. Clifford A, Lang L, Chen R. Effects of maternal cigarette smoking during pregnancy on cognitive parameters of children and young adults: A literature review. Neurotoxicology and Teratology. 2012;34(6):560-70. 20. Hodyl NA, Stark MJ, Scheil W, Grzeskowiak LE, Clifton VL. Perinatal outcomes following maternal asthma and cigarette smoking during pregnancy. European Respiratory Journal. 2014;43(3):704-16. 21. Oken E, Levitan EB, Gillman MW. Maternal smoking during pregnancy and child overweight: systematic review and meta-analysis. International Journal of Obesity. 2008;32(2):201-10.

184

22. Li R, Lodge J, Flatley C, Kumar S. The burden of adverse obstetric and perinatal outcomes from maternal smoking in an Australian cohort. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2019;59(3):356-61. 23. Adams EK, Melvin CL, Raskind-Hood C, Joski PJ, Galactionova E. Infant Delivery Costs Related to Maternal Smoking: An Update. Nicotine & Tobacco Research. 2011;13(8):627- 37. 24. Godfrey C, Pickett K, Parrott S, Mdege N, Eapen P. Estimating the Costs to the NHS of Smoking in Pregnancy for Pregnant Women and Infants. Final Report. York: Public Health Research Consortium, University of York; 2010. 25. Lange S, Probst C, Rehm J, Popova S. National, regional, and global prevalence of smoking during pregnancy in the general population: a systematic review and meta-analysis. The Lancet Global Health. 2018;6(7):e769-e76. 26. Australian Institute of Health and Welfare. Australia’s mothers and babies 2017 — in brief. Perinatal statistics series no. 35. Cat no. PER 100. Canberra: AIHW; 2019. 27. Centre for Epidemiology and Evidence. New South Wales Mothers and Babies 2017 [online]. Sydney: NSW Ministry of Health; 2018. 28. Moolchan ET, Fagan P, Fernander AF, Velicer WF, Hayward MD, King G, et al. Addressing tobacco-related health disparities. Addiction. 2007;102:30-42. 29. Riaz M, Lewis S, Naughton F, Ussher M. Predictors of smoking cessation during pregnancy: a systematic review and meta-analysis. Addiction. 2018;113(4):610-22. 30. Schneider S, Schütz J. Who smokes during pregnancy? A systematic literature review of population-based surveys conducted in developed countries between 1997 and 2006. The European Journal of Contraception & Reproductive Health Care. 2008;13(2):138-47. 31. Gould GS, Patten C, Glover M, Kira A, Jayasinghe H. Smoking in Pregnancy Among Indigenous Women in High-Income Countries: A Narrative Review. Nicotine & Tobacco Research. 2017;19(5):506-17. 32. Drake P, Driscoll AK, Mathews TJ. Cigarette smoking during pregnancy: United States, 2016. NCHS Data Brief, no 305. Hyattsville, MD: National Center for Health Statistics.; 2018. 33. Tong VT, Farr SL, Bombard J, D'Angelo D, Ko JY, England LJ. Smoking Before and During Pregnancy Among Women Reporting Depression or Anxiety. Obstetrics & Gynecology. 2016;128(3):562-70. 34. Lamy S, Houivet E, Marret S, Hennart B, Delavenne H, Benichou J, et al. Risk factors associated to tobacco and alcohol use in a large French cohort of pregnant women. Archives of Women's Mental Health. 2019;22(2):267-77. 35. Seybold DJ, Broce M, Siegel E, Findley J, Calhoun BC. Smoking in Pregnancy in West Virginia: Does Cessation/Reduction Improve Perinatal Outcomes? Maternal and Child Health Journal. 2012;16(1):133-8. 36. Bickerstaff M, Beckmann M, Gibbons K, Flenady V. Recent cessation of smoking and its effect on pregnancy outcomes. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2012;52(1):54-8. 37. McCowan LME, Dekker GA, Chan E, Stewart A, Chappell LC, Hunter M, et al. Spontaneous preterm birth and small for gestational age infants in women who stop smoking early in pregnancy: prospective cohort study. BMJ. 2009;338. 38. Soneji S, Beltrán-Sánchez H. Association of Maternal Cigarette Smoking and Smoking Cessation With Preterm Birth. JAMA Network Open. 2019;2(4):e192514-e. 39. Taylor G, McNeill A, Girling A, Farley A, Lindson-Hawley N, Aveyard P. Change in mental health after smoking cessation: systematic review and meta-analysis. British Medical Journal. 2014;348:g1151. 40. Wehby GL, Prater K, McCarthy AM, Castilla EE, Murray JC. The Impact of Maternal Smoking during Pregnancy on Early Child Neurodevelopment. Journal of Human Capital. 2011;5(2):207-54. 41. Kristjansson AL, Thomas S, Lilly CL, Thorisdottir IE, Allegrante JP, Sigfusdottir ID. Maternal smoking during pregnancy and academic achievement of offspring over time: A registry data-based cohort study. Preventive Medicine. 2018;113:74-9.

185

42. Cooper S, Orton S, Leonardi-Bee J, Brotherton E, Vanderbloemen L, Bowker K, et al. Smoking and quit attempts during pregnancy and postpartum: a longitudinal UK cohort. BMJ Open. 2017;7(11):e018746. 43. Emery JL, Sutton S, Naughton F. Cognitive and Behavioral Predictors of Quit Attempts and Biochemically-Validated Abstinence During Pregnancy. Nicotine & Tobacco Research. 2017;19(5):547-54. 44. Hyland A, Borland R, Li Q, Yong H-H, McNeill A, Fong GT, et al. Individual-level predictors of cessation behaviours among participants in the International Tobacco Control (ITC) Four Country Survey. Tobacco Control. 2006;15(suppl 3):iii83-iii94. 45. Kale D, Gilbert HM, Sutton S. Are predictors of making a quit attempt the same as predictors of 3-month abstinence from smoking? Findings from a sample of smokers recruited for a study of computer-tailored smoking cessation advice in primary care. Addiction. 2015;110(10):1653-64. 46. Kurti AN, Redner R, Bunn JY, Tang K, Nighbor T, Lopez AA, et al. Examining the relationship between pregnancy and quitting use of tobacco products in a U.S. national sample of women of reproductive age. Preventive Medicine. 2018;117:52-60. 47. Passmore E, McGuire R, Correll P, Bentley J. Demographic factors associated with smoking cessation during pregnancy in New South Wales, Australia, 2000–2011. BMC public health. 2015;15(1):398. 48. Schneider S, Huy C, Schütz J, Diehl K. Smoking cessation during pregnancy: a systematic literature review. Drug Alcohol Review. 2010;29(1):81-90. 49. Giglia RC, Binns CW, Alfonso HS. Which women stop smoking during pregnancy and the effect on breastfeeding duration. BMC public health. 2006;6(1):195. 50. Flower A, Shawe J, Stephenson J, Doyle P. Pregnancy planning, smoking behaviour during pregnancy, and neonatal outcome: UK millennium cohort study. BMC Pregnancy and Childbirth. 2013;13(1):238. 51. Barnett MJ, Fealy S, Wilson A. Barriers and enablers for smoking cessation amongst pregnant women: An Umbrella Review. Women and Birth. 2019;32(4):310-7. 52. Arizmendi TG, Affonso DD. Stressful events related to pregnancy and postpartum. Journal of Psychosomatic Research. 1987;31(6):743-56. 53. Dempsey D, Jacob P, Benowitz NL. Accelerated Metabolism of Nicotine and Cotinine in Pregnant Smokers. Journal of Pharmacology and Experimental Therapeutics. 2002;301(2):594- 8. 54. Heil SH, Higgins ST, Mongeon JA, Badger GJ, Bernstein IM. Characterizing nicotine withdrawal in pregnant cigarette smokers. Experimental and Clinical Psychopharmacology. 2006;14(2):165-70. 55. Naughton F, Vaz LR, Coleman T, Orton S, Bowker K, Leonardi-Bee J, et al. Interest in and use of smoking cessation support across pregnancy and postpartum. Nicotine & Tobacco Research [Internet]. 2019 [cited 2019 10 September]:[1-9 pp.]. Available from: https://doi.org/10.1093/ntr/ntz151. doi:10.1093/ntr/ntz151. 56. Ussher M, West R, Hibbs N. A survey of pregnant smokers’ interest in different types of smoking cessation support. Patient Education and Counseling. 2004;54(1):67-72. 57. World Health Organization. WHO Recommendations for the Prevention and Management of Tobacco Use and Second-Hand Smoke Exposure in Pregnancy. Geneva: World Health Organization; 2013 [cited 2016 24 July]. Available from: https://apps.who.int/iris/bitstream/handle/10665/94555/9789241506076_eng.pdf. 58. US Public Health Service. Treating Tobacco Use and Dependence: 2008 Update. Clinical Practice Guideline. Rockville, MD: US Department of Health and Human Services; 2008 [cited 2016 24 July]. Available from: https://www.ahrq.gov/sites/default/files/wysiwyg/professionals/clinicians-providers/guidelines- recommendations/tobacco/clinicians/update/treating_tobacco_use08.pdf. 59. Siu AL, for US Preventive Services Task Force. Behavioral and Pharmacotherapy Interventions for Tobacco Smoking Cessation in Adults, Including Pregnant Women: U.S. Preventive Services Task Force Recommendation Statement. Annals of Internal Medicine. 2015;163(8):622-34.

186

60. West R, McNeill A, Raw M. Smoking cessation guidelines for health professionals: an update. Thorax. 2000;55(12):987-99. 61. The Royal Australian College of General Practitioners. Supporting smoking cessation: A guide for health professionals. 2nd edn. East Melbourne, Vic: RACGP; 2019. 62. Gould GS, Twyman L, Stevenson L, Gribbin GR, Bonevski B, Palazzi K, et al. What components of smoking cessation care during pregnancy are implemented by health providers? A systematic review and meta-analysis. BMJ Open. 2019;9(8):e026037. 63. Chamberlain C, O'Mara‐Eves A, Porter J, Coleman T, Perlen SM, Thomas J, et al. Psychosocial interventions for supporting women to stop smoking in pregnancy. Cochrane Database of Systematic Reviews. 2017;Issue 2:Art. No.: CD001055. doi: 10.1002/14651858.CD001055.pub5. 64. Victoria Government. Quit Coach 2019 [cited 2019 31 October ]. Available from: www.quitcoach.org.au. 65. New South Wales Government. iCanQuit. 2019 [cited 2019 1 November]. Available from: https://www.icanquit.com.au/. 66. Australian Government Department of Health. Quit for You - Quit for Two app. [cited 2019 31 October]. Available from: https://www.health.gov.au/resources/apps-and-tools/quit-for- you-quit-for-two-app. 67. Australian Government Department of Health. Quit Because You can 2019 [Available from: https://makesmokinghistory.org.au/im-ready-to-quit/getting-help-to-quit/. 68. Daniella G, Tessa L, Thomas F. Evaluation of the Victorian Fresh Start Short Course: January 2000 to February 2003. CBRC Research Paper Series No. 7. Melbourne, Australia: Centre for Behavioural Research in Cancer,The Cancer Council Victoria; June 2004. 69. New South Wales Ministry of Health. Quit for new life. 2018 [cited 2020 16 February]. Available from: https://www.health.nsw.gov.au/tobacco/Pages/quit-for-new-life.aspx. 70. Jones M, Lewis S, Parrott S, Wormall S, Coleman T. Re-starting smoking in the postpartum period after receiving a smoking cessation intervention: a systematic review. Addiction. 2016;111(6):981-90. 71. Coleman T, Chamberlain C, Davey MA, Cooper SE, Leonardi‐Bee J. Pharmacological interventions for promoting smoking cessation during pregnancy. Cochrane Database of Systematic Reviews. 2015;Issue 12. Art. No.: CD010078:doi:10.1002/14651858.CD010078.pub2. 72. Australian Government Department of Health. Pharmaceutical Benefits Scheme (PBS). Patient charges [cited 2019 3 March]. Available from: http://www.pbs.gov.au/info/healthpro/explanatory- notes/section1/Section_1_4_Explanatory_Notes. 73. West RJ, Jarvis MJ, Russell MA, Carruthers ME, Feyerabend C. Effect of nicotine replacement on the cigarette withdrawal syndrome. British Journal of Addiction. 1984;79(2):215- 19. 74. Slemmer JE, Martin BR, Damaj MI. Bupropion is a nicotinic antagonist. Journal of Pharmacology and Experimental Therapeutics. 2000;295(1):321-27. 75. Coe JW, Brooks PR, Vetelino MG, Wirtz MC, Arnold EP, Huang J, et al. Varenicline: An α4β2 Nicotinic Receptor Partial Agonist for Smoking Cessation. Journal of Medicinal Chemistry. 2005;48(10):3474-7. 76. Howes S, Hartmann‐Boyce J, Livingstone‐Banks J, Hong B, Lindson N. Antidepressants for smoking cessation. Cochrane Database of Systematic Reviews. 2020(4). 77. Cahill K, Lindson‐Hawley N, Thomas KH, Fanshawe TR, Lancaster T. Nicotine receptor partial agonists for smoking cessation. Cochrane Database of Systematic Reviews. 2016(5). 78. Hartmann‐Boyce J, Chepkin SC, Ye W, Bullen C, Lancaster T. Nicotine replacement therapy versus control for smoking cessation. Cochrane Database of Systematic Reviews. 2018;Issue 5. Art. No.: CD000146. doi: 10.1002/14651858.CD000146.pub5. 79. Loebstein R, Lalkin A, Koren G. Pharmacokinetic Changes During Pregnancy and Their Clinical Relevance. Clinical Pharmacokinetics. 1997;33(5):328-43. 80. Costantine MM. Physiologic and pharmacokinetic changes in pregnancy. Frontiers in Pharmacology. 2014;5(65):1-5.

187

81. Holford NHG, Sheiner LB. Kinetics of pharmacologic response. Pharmacology & Therapeutics. 1982;16(2):143-66. 82. Pariente G, Leibson T, Carls A, Adams-Webber T, Ito S, Koren G. Pregnancy- Associated Changes in Pharmacokinetics: A Systematic Review. PLOS Medicine. 2016;13(11):e1002160. 83. Nordeng H, Ystrøm E, Einarson A. Perception of risk regarding the use of medications and other exposures during pregnancy. European journal of clinical pharmacology. 2010;66(2):207-14. 84. Ceulemans M, Lupattelli A, Nordeng H, Odalovic M, Twigg M, Foulon V. Beliefs About Medicines and Adherence to Pharmacotherapy in Pregnancy: Opportunities for Community Pharmacists. Current Pharmaceutical Design. 2019;25(5):469-82. 85. Lupattelli A, Spigset O, Nordeng H. Adherence to medication for chronic disorders during pregnancy: results from a multinational study. International Journal of Clinical Pharmacy. 2014;36(1):145-53. 86. Lupattelli A, Spigset O, Björnsdóttir I, Hämeen-Anttila K, Mårdby A-C, Panchaud A, et al. Patterns and factors associated with low adherence to psychotropic medications during pregnancy-a cross-sectional, multinational web-based study. Depression and Anxiety. 2015;32(6):426-36. 87. Amundsen S, Øvrebø TG, Amble NMS, Poole AC, Nordeng H. Risk perception, beliefs about medicines and medical adherence among pregnant and breastfeeding women with migraine: findings from a cross-sectional study in Norway. BMJ Open. 2019;9(2):e026690. 88. Hickson C, Lewis S, Campbell KA, Cooper S, Berlin I, Claire R, et al. Comparison of nicotine exposure during pregnancy when smoking and abstinent with nicotine replacement therapy: systematic review and meta-analysis. Addiction. 2019;114(3):406-24. 89. Patnode CD, Henderson JT, Thompson JH, Senger CA, Fortmann SP, Whitlock EP. Behavioral Counseling and Pharmacotherapy Interventions for Tobacco Cessation in Adults, Including Pregnant Women: A Review of Reviews for the U.S. Preventive Services Task Force. Evidence Synthesis No. 134. AHRQ Publication No. 14-05200-EF-1. Rockville, MD: Agency for Healthcare Research and Quality; 2015. 90. Oncken C, Dornelas EA, Kuo C-L, Sankey HZ, Kranzler HR, Mead EL, et al. Randomized trial of nicotine inhaler for pregnant smokers. American Journal of Obstetrics & Gynecology. 2019;1(1):10-8. 91. El-Mohandes AAE, Windsor R, Tan S, Perry DC, Gantz MG, Kiely M. A Randomized Clinical Trial of Trans-Dermal Nicotine Replacement in Pregnant African-American Smokers. Maternal and Child Health Journal. 2013;17(5):897-906. 92. Berlin I, Grangé G, Jacob N, Tanguy M-L. Nicotine patches in pregnant smokers: randomised, placebo controlled, multicentre trial of efficacy. British Medical Journal. 2014;348:g1622. 93. Coleman T, Cooper S, Thornton JG, Grainge MJ, Watts K, Britton J, et al. A Randomized Trial of Nicotine-Replacement Therapy Patches in Pregnancy. New England Journal of Medicine. 2012;366(9):808-18. 94. Hotham ED, Gilbert AL, Atkinson ER. A randomised-controlled pilot study using nicotine patches with pregnant women. Addictive Behaviors. 2006;31(4):641-8. 95. Pollak KI, Oncken CA, Lipkus IM, Lyna P, Swamy GK, Pletsch PK, et al. Nicotine Replacement and Behavioral Therapy for Smoking Cessation in Pregnancy. American journal of preventive medicine. 2007;33(4):297-305. 96. Kapur B, Hackman R, Selby P, Klein J, Koren G. Randomized, double-blind, placebo- controlled trial of nicotine replacement therapy in pregnancy. Current Therapeutic Research. 2001;62(4):274-8. 97. Oncken C, Dornelas E, Greene J, Sankey H, Glasmann A, Feinn R, et al. Nicotine Gum for Pregnant Smokers: A Randomized Controlled Trial. Obstetrics & Gynecology. 2008;112(4):859-67. 98. Wisborg K, Henriksen TB, Jespersen LB, Secher NJ. Nicotine patches for pregnant smokers: a randomized controlled study. Obstetrics & Gynecology. 2000;96(6):967-71.

188

99. Stotts AL, Northrup TF, Cinciripini PM, Minnix JA, Blalock JA, Mullen PD, et al. Randomized, controlled pilot trial of bupropion for pregnant smokers: challenges and future directions. American Journal of Perinatology. 2015;32(4):351-56. 100. Nanovskaya TN, Oncken C, Fokina VM, Feinn RS, Clark SM, West H, et al. Bupropion sustained release for pregnant smokers: a randomized, placebo-controlled trial. American Journal of Obstetrics and Gynecology. 2017;216(4):420.e1-.e9. 101. Miller H. Bupropion for Smoking Cessation in Pregnancy. NLM identifier: NCT01875172 [updated June 11, 2013; cited 2019 June 18]. Available from: https://clinicaltrials.gov/ct2/show/NCT01875172. 102. Pfizer. Varenicline Pregnancy Cohort Study. NLM identifier: NCT0 1875172. [updated 22 January, 2019; cited 2019 June 18]. Available from: https://clinicaltrials.gov/ct2/show/NCT01290445. 103. Choi SK, Tran DT, Preen DB, Randall D, Kemp-Casey A, Einarsdottir K, et al. Varenicline is More Effective than Nicotine Replacement Therapy During Pregnancy: Findings from the Smoking MUMS (Maternal Use of Medications and Safety) Study. International Journal of Population Data Science. 2018;3(4). 104. Jarlenski MP, Chisolm MS, Kachur S, Neale DM, Bennett WL. Use of Pharmacotherapies for Smoking Cessation: Analysis of Pregnant and Postpartum Medicaid Enrollees. American journal of preventive medicine. 2015;48(5):528-34. 105. Scheuermann TS, Richter KP, Jacobson LT, Shireman TI. Medicaid Coverage of Smoking Cessation Counseling and Medication Is Underutilized for Pregnant Women. Nicotine & Tobacco Research. 2017;19(5):656-9. 106. Brose LS, McEwen A, West R. Association between nicotine replacement therapy use in pregnancy and smoking cessation. Drug and Alcohol Dependence. 2013;132(3):660-4. 107. Dhalwani NN, Szatkowski L, Coleman T, Fiaschi L, Tata LJ. Prescribing of nicotine replacement therapy in and around pregnancy: a population-based study using primary care data. British Journal of General Practice. 2014;64(626):e554-e60. 108. National Institute for Health and Care Excellence. Smoking: stopping in pregnancy and after childbirth. 2010 ed. London, UK: NICE; 2013. 109. Bérard A, Zhao J-P, Sheehy O. Success of smoking cessation interventions during pregnancy. American Journal of Obstetrics and Gynecology. 2016;215(5):611.e1-.e8. 110. Roper L, Tran DT, Einarsdóttir K, Preen DB, Havard A. Algorithm for resolving discrepancies between claims for smoking cessation pharmacotherapies during pregnancy and smoking status in delivery records: The impact on estimates of utilisation. PLOS ONE. 2018;13(8):e0202999. 111. Lassen TH, Madsen M, Skovgaard LT, Strandberg-Larsen K, Olsen J, Andersen A-MN. Maternal use of nicotine replacement therapy during pregnancy and offspring birthweight: a study within the Danish National Birth Cohort. Paediatric and perinatal epidemiology. 2010;24(3):272-81. 112. Strandberg-Larsen K, Tinggaard M, Nybo Andersen A-M, Olsen J, Grønbæk M. Use of nicotine replacement therapy during pregnancy and stillbirth: a cohort study. BJOG: An International Journal of Obstetrics & Gynaecology. 2008;115(11):1405-10. 113. Irvine L, Flynn RWV, Libby G, Crombie IK, Evans JMM. Drugs Dispensed in Primary Care During Pregnancy. Drug Safety. 2010;33(7):593-604. 114. Tappin DM, MacAskill S, Bauld L, Eadie D, Shipton D, Galbraith L. Smoking prevalence and smoking cessation services for pregnant women in Scotland. Substance Abuse Treatment, Prevention, and Policy. 2010;5(1):1. 115. Harris BM, Harris ML, Rae K, Chojenta C. Barriers and facilitators to smoking cessation within pregnant Aboriginal and/or Torres Strait Islander women: An integrative review. Midwifery. 2019;73:49-61. 116. Bedford K, Wallace C, Carroll T, Rissel C. Pregnant smokers are receptive to smoking cessation advice and use of nicotine replacement therapy. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2008;48(4):424-6. 117. Hoekzema L, Werumeus Buning A, Bonevski B, Wolke L, Wong S, Drinkwater P, et al. Smoking rates and smoking cessation preferences of pregnant women attending antenatal

189

clinics of two large Australian maternity hospitals. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2014;54(1):53-8. 118. Hotham ED, Atkinson ER, Gilbert AL. Focus groups with pregnant smokers: barriers to cessation, attitudes to nicotine patch use and perceptions of cessation counselling by care providers. Drug and alcohol review. 2002;21(2):163-8. 119. World Health Organisation. Guide to Good Prescribing - A Practical Manual [online]. Geneva, Switzerland: WHO Department of Essential Drugs and Medicines Policy; 1994 [cited 2019 27 August ]. Available from: http://apps.who.int/medicinedocs/pdf/whozip23e/whozip23e.pdf. 120. Australian Therapeutic Goods Administration. Prescribing medicines in pregnancy database [cited 2016 25 November]. Available from: https://www.tga.gov.au/prescribing- medicines-pregnancy-database. 121. Cooper S, Taggar J, Lewis S, Marlow N, Dickinson A, Whitemore R, et al. Effect of nicotine patches in pregnancy on infant and maternal outcomes at 2 years: follow-up from the randomised, double-blind, placebo-controlled SNAP trial. The Lancet Respiratory Medicine. 2014;2(9):728-37. 122. Iyen B, Vaz LR, Taggar J, Cooper S, Lewis S, Coleman T. Is the apparently protective effect of maternal nicotine replacement therapy (NRT) used in pregnancy on infant development explained by smoking cessation?: secondary analyses of a randomised controlled trial. BMJ Open. 2019;9(7):e024923. 123. Dhalwani NN, Szatkowski L, Coleman T, Fiaschi L, Tata LJ. Stillbirth Among Women Prescribed Nicotine Replacement Therapy in Pregnancy: Analysis of a Large UK Pregnancy Cohort. Nicotine & Tobacco Research. 2018;21(4):409-15. 124. Dhalwani NN, Szatkowski L, Coleman T, Fiaschi L, Tata LJ. Nicotine Replacement Therapy in Pregnancy and Major Congenital Anomalies in Offspring. Pediatrics. 2015;135(5):859-67. 125. Morales-Suarez-Varela MM, Bille C, Christensen K, Olsen J. Smoking habits, nicotine use, and congenital malformations. Obstetrics & Gynecology. 2006;107(1):51-7. 126. Gaither KH, Huber LRB, Thompson ME, Huet-Hudson YM. Does the Use of Nicotine Replacement Therapy During Pregnancy Affect Pregnancy Outcomes? Maternal and Child Health Journal. 2008;13(4):497-504. 127. Tran DT, Preen DB, Einarsdottir K, Kemp-Casey A, Randall D, Jorm LR, et al. Use of smoking cessation pharmacotherapies during pregnancy is not associated with increased risk of adverse pregnancy outcomes: a population-based cohort study. BMC Medicine. 2020;18(1):15. 128. Turner E, Jones M, Vaz LR, Coleman T. Systematic Review and Meta-Analysis to Assess the Safety of Bupropion and Varenicline in Pregnancy. Nicotine & Tobacco Research. 2018;21(8):1001-10. 129. Pedersen L, Petronis KR, Nørgaard M, Mo J, Frøslev T, Stephansson O, et al. Risk of adverse birth outcomes after maternal varenicline use: A population-based observational study in Denmark and Sweden. Pharmacoepidemiology and Drug Safety.n/a(n/a). 130. American College of Obstetricians and Gynecologists. Committee Opinion No. 721: Smoking Cessation During Pregnancy. Obstetrics & Gynecology. 2017;130(4):e200-e4. 131. Campbell KA, Fergie L, Coleman-Haynes T, Cooper S, Lorencatto F, Ussher M, et al. Improving Behavioral Support for Smoking Cessation in Pregnancy: What Are the Barriers to Stopping and Which Behavior Change Techniques Can Influence Them? Application of Theoretical Domains Framework. International journal of environmental research and public health. 2018;15(2):359. 132. Bar-Zeev Y, Skelton E, Bonevski B, Gruppetta M, Gould GS. Overcoming Challenges to Treating Tobacco use During Pregnancy - A Qualitative study of Australian General Practitioners Barriers. BMC Pregnancy and Childbirth. 2019;19(1):61. 133. Longman JM, Adams CM, Johnston JJ, Passey ME. Improving implementation of the smoking cessation guidelines with pregnant women: How to support clinicians? Midwifery. 2018;58:137-44. 134. Bell R, Glinianaia SV, Waal Zvd, Close A, Moloney E, Jones S, et al. Evaluation of a complex healthcare intervention to increase smoking cessation in pregnant women: interrupted time series analysis with economic evaluation. Tobacco Control. 2018;27(1):90-8.

190

135. Flitter AS, Lubitz SF, Ziedonis D, Stevens N, Leone FT, Mandell D, et al. A Cluster- Randomized Clinical Trial Testing the Effectiveness of the Addressing Tobacco Through Organizational Change Model for Improving the Treatment of Tobacco Use in Community Mental Health Care: Preliminary Study Feasibility and Baseline Findings. Nicotine & Tobacco Research. 2018;21(5):559-67. 136. Thomas D, Abramson MJ, Bonevski B, George J. System change interventions for smoking cessation. Cochrane Database of Systematic Reviews. 2017;Issue 2. Art. No.: CD010742. DOI: 10.1002/14651858.CD010742.pub2. 137. Fiore M, Adsit R, Zehner M, McCarthy D, Lundsten S, Hartlaub P, et al. An electronic health record–based interoperable eReferral system to enhance smoking Quitline treatment in primary care. Journal of the American Medical Informatics Association. 2019;26(8-9):778–86. 138. Paredes P, de la Peña M, Flores-Guerra E, Diaz J, Trostle J. Factors influencing physicians' prescribing behaviour in the treatment of childhood diarrhoea: Knowledge may not be the clue. Social Science & Medicine. 1996;42(8):1141-53. 139. Bar-Zeev Y, Bovill M, Bonevski B, Gruppetta M, Oldmeadow C, Palazzi K, et al. Improving smoking cessation care in pregnancy at Aboriginal Medical Services: ‘ICAN QUIT in Pregnancy’ step-wedge cluster randomised study. BMJ Open. 2019;9(6):e025293. 140. De Wilde K, Tency I, Steckel S, Temmerman M, Boudrez H, Maes L. Which role do midwives and gynecologists have in smoking cessation in pregnant women? – A study in Flanders, Belgium. Sexual & Reproductive Healthcare. 2015;6(2):66-73. 141. Ajzen I. From Intentions to Actions: A Theory of Planned Behavior. In: Kuhl J, Beckmann J, editors. Action Control: From Cognition to Behavior. Berlin, Heidelberg: Springer Berlin Heidelberg; 1985. p. 11-39. 142. Godin G, Bélanger-Gravel A, Eccles M, Grimshaw J. Healthcare professionals' intentions and behaviours: A systematic review of studies based on social cognitive theories. Implementation Science. 2008;3(1):36. 143. Farber HJ, Groner J, Walley S, Nelson K. Protecting Children From Tobacco, Nicotine, and Tobacco Smoke. Pediatrics. 2015;136(5):e1439-e67. 144. U.S. Department of Health and Human Services. The Health Consequences of Smoking: 50 Years of Progress. A Report of the Surgeon General. Atlanta, GA: U.S. : 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. 145. Öberg M, Woodward A, Jaakkola MS, Peruga A, Prüss-Ustün A. Global estimate of the burden of disease from second-hand smoke. Geneva: World Health Organisation; 2011 [cited 2019 16 November]. Available from: https://apps.who.int/iris/bitstream/handle/10665/44426/9789241564076_eng.pdf. 146. Stocks J. The functional growth and development of the lung during the first year of life. Early Human Development. 1977;1(3):285-309. 147. Bartelink IH, Rademaker CMA, Schobben AFAM, van den Anker JN. Guidelines on Paediatric Dosing on the Basis of Developmental Physiology and Pharmacokinetic Considerations. Clinical Pharmacokinetics. 2006;45(11):1077-97. 148. Anderson TM, Lavista Ferres JM, Ren SY, Moon RY, Goldstein RD, Ramirez J-M, et al. Maternal Smoking Before and During Pregnancy and the Risk of Sudden Unexpected Infant Death. Pediatrics. 2019;143(4):e20183325. 149. Burke H, Leonardi-Bee J, Hashim A, Pine-Abata H, Chen Y, Cook DG, et al. Prenatal and Passive Smoke Exposure and Incidence of Asthma and Wheeze: Systematic Review and Meta-analysis. Pediatrics. 2012;129(4):735-44. 150. Leung GM, Ho L-M, Lam T-H. Secondhand Smoke Exposure, Smoking Hygiene, and Hospitalization in the First 18 Months of Life. JAMA Pediatrics. 2004;158(7):687-93. 151. Groh CA, Vittinghoff E, Benjamin EJ, Dupuis J, Marcus GM. Childhood Tobacco Smoke Exposure and Risk of Atrial Fibrillation in Adulthood. Journal of the American College of Cardiology. 2019;74(13):1658-64. 152. Öberg M, Jaakkola MS, Woodward A, Peruga A, Prüss-Ustün A. Worldwide burden of disease from exposure to second-hand smoke: a retrospective analysis of data from 192 countries. The Lancet. 2011;377(9760):139-46.

191

153. Gutvirtz G, Wainstock T, Landau D, Sheiner E. Maternal smoking during pregnancy and long-term neurological morbidity of the offspring. Addictive Behaviors. 2019;88:86-91. 154. Leonardi-Bee J, Jere ML, Britton J. Exposure to parental and sibling smoking and the risk of smoking uptake in childhood and adolescence: a systematic review and meta-analysis. Thorax. 2011;66(10):847-55. 155. Saygın Avşar T, McLeod H, Jackson L. Health outcomes of maternal smoking during pregnancy and postpartum period for the mother and infant: protocol for an umbrella review. Systematic Reviews. 2018;7(1):235. 156. Hall ES, Venkatesh M, Greenberg JM. A population study of first and subsequent pregnancy smoking behaviors in Ohio. Journal of Perinatology. 2016;36(11):948-53. 157. Makate M, Whetton S, Tait RJ, Dey T, Scollo M, Banks E, et al. Tobacco Cost of Illness Studies: A Systematic Review. Nicotine & Tobacco Research. 2019;22(4):458-65. 158. Shisler S, Homish GG, Molnar DS, Schuetze P, Colder CR, Eiden RD. Predictors of Changes in Smoking From Third Trimester to 9 Months Postpartum. Nicotine & Tobacco Research. 2016;18(1):84-7. 159. Kondracki AJ. Prevalence and patterns of cigarette smoking before and during early and late pregnancy according to maternal characteristics: the first national data based on the 2003 birth certificate revision, United States, 2016. Reproductive Health. 2019;16(1):142. 160. NHS Digital. Statistics on Women's Smoking Status at Time of Delivery, England ‐ Quarter 4, 2018‐19 2019 [cited 2019 30 August]. Available from: https://digital.nhs.uk/data-and- information/publications/statistical/statistics-on-women-s-smoking-status-at-time-of-delivery- england/statistics-on-womens-smoking-status-at-time-of-delivery-england-quarter-4-january- 2019-to-march-2019. 161. Levine MD, Cheng Y, Marcus MD, Kalarchian MA, Emery RL. Preventing Postpartum Smoking Relapse: A Randomized Clinical Trial. JAMA Internal Medicine. 2016;176(4):443-52. 162. Coleman-Cowger VH, Mark KS, Rosenberry ZR, Koszowski B, Terplan M. A Pilot Randomized Controlled Trial of a Phone-based Intervention for Smoking Cessation and Relapse Prevention in the Postpartum Period. Journal of Addiction Medicine. 2018;12(3):193- 200. 163. Pirie K, Peto R, Reeves GK, Green J, Beral V. The 21st century hazards of smoking and benefits of stopping: a prospective study of one million women in the UK. The Lancet. 2013;381(9861):133-41. 164. Carter CS, Altemus M, Pchrousos G. Neuroendocrine and emotional changes in the post-partum period. Progress in Brain Research. 2001;133:241-9. 165. Tully KP, Stuebe AM, Verbiest SB. The fourth trimester: a critical transition period with unmet maternal health needs. American Journal of Obstetrics and Gynecology. 2017;217(1):37- 41. 166. Levine MD, Marcus MD, Kalarchian MA, Houck PR, Cheng Y. Weight Concerns, Mood, and Postpartum Smoking Relapse. American journal of preventive medicine. 2010;39(4):345- 51. 167. Negron R, Martin A, Almog M, Balbierz A, Howell EA. Social Support During the Postpartum Period: Mothers’ Views on Needs, Expectations, and Mobilization of Support. Maternal and Child Health Journal. 2013;17(4):616-23. 168. Flemming K, Graham H, Heirs M, Fox D, Sowden A. Smoking in pregnancy: a systematic review of qualitative research of women who commence pregnancy as smokers. Journal of Advanced Nursing. 2013;69(5):1023-36. 169. Orton S, Coleman T, Lewis S, Cooper S, Jones LL. “I Was a Full Time Proper Smoker”: A Qualitative Exploration of Smoking in the Home after Childbirth among Women Who Relapse Postpartum. PLOS ONE. 2016;11(6):e0157525. 170. Firoz T, McCaw-Binns A, Filippi V, Magee LA, Costa ML, Cecatti JG, et al. A framework for healthcare interventions to address maternal morbidity. International Journal of Gynecology & Obstetrics. 2018;141(S1):61-8. 171. The Royal Australian College of General Practitioners. Guidelines for preventive activities in general practice. 9th ed. East Melbourne, VIC: RACGP; 2018.

192

172. Levitt C, Shaw E, Wong S, Kaczorowski J, Group TMUPR. Systematic Review of the Literature on Postpartum Care: Effectiveness of Interventions for Smoking Relapse Prevention, Cessation, and Reduction in Postpartum Women. Birth. 2007;34(4):341-7. 173. Jiménez-Muro A, Nerín I, Samper P, Marqueta A, Beamonte A, Gargallo P, et al. A proactive smoking cessation intervention in postpartum women. Midwifery. 2013;29(3):240-5. 174. Cummins SE, Tedeschi GJ, Anderson CM, Zhu S-H. Telephone Intervention for Pregnant Smokers: A Randomized Controlled Trial. American journal of preventive medicine. 2016;51(3):318-26. 175. Hennrikus D, Pirie P, Hellerstedt W, Lando HA, Steele J, Dunn C. Increasing support for smoking cessation during pregnancy and postpartum: Results of a randomized controlled pilot study. Preventive Medicine. 2010;50(3):134-7. 176. Hayes CB, Collins C, O’Carroll H, Wyse E, Gunning M, Geary M, et al. Effectiveness of Motivational Interviewing in Influencing Smoking Cessation in Pregnant and Postpartum Disadvantaged Women. Nicotine & Tobacco Research. 2013;15(5):969-77. 177. Severson HH, Andrews JA, Lichtenstein E, Wall M, Akers L. Reducing Maternal Smoking and Relapse: Long-Term Evaluation of a Pediatric Intervention. Preventive Medicine. 1997;26(1):120-30. 178. Orton S, Coleman T, Coleman-Haynes T, Ussher M. Predictors of Postpartum Return to Smoking: A Systematic Review. Nicotine & Tobacco Research. 2017;20(6):665-73. 179. Fossum B, Arborelius E, Bremberg S. Evaluation of a counseling method for the prevention of child exposure to tobacco smoke: an example of client-centered communication. Preventive Medicine. 2004;38(3):295-301. 180. Sachs HC. The Transfer of Drugs and Therapeutics Into Human Breast Milk: An Update on Selected Topics. Pediatrics. 2013;132(3):e796-e809. 181. Ilett KF, Hale TW, Page-Sharp M, Kristensen JH, Kohan R, Hackett LP. Use of nicotine patches in breast-feeding mothers: Transfer of nicotine and cotinine into human milk. Clinical Pharmacology & Therapeutics. 2003;74(6):516-24. 182. Svensson CK. Clinical Pharmacokinetics of Nicotine. Clinical Pharmacokinetics. 1987;12(1):30-40. 183. Vio F, Salazar G, Infante C. Smoking during pregnancy and lactation and its effects on breast-milk volume. The American Journal of Clinical Nutrition. 1991;54(6):1011-6. 184. Fríguls B, Joya X, García-Algar O, Pallás CR, Vall O, Pichini S. A comprehensive review of assay methods to determine drugs in breast milk and the safety of breastfeeding when taking drugs. Analytical and Bioanalytical Chemistry. 2010;397(3):1157-79. 185. Weiser TM, Lin M, Garikapaty V, Feyerharm RW, Bensyl DM, Zhu B-P. Association of Maternal Smoking Status With Breastfeeding Practices: Missouri, 2005. Pediatrics. 2009;124(6):1603-10. 186. Giglia R, Binns CW, Alfonso H. Maternal cigarette smoking and breastfeeding duration. Acta Paediatrica. 2006;95(11):1370-4. 187. Drehmer JE, Ossip DJ, Nabi-Burza E, Hipple Walters B, Gorzkowski JA, Winickoff JP. Pediatric Office Delivery of Smoking Cessation Assistance for Breast-Feeding Mothers. Nicotine & Tobacco Research. 2018;22(3):346-53. 188. World Health Organization. Global strategy for infant and young child feeding Geneva, Switzerland: WHO; 2003 [cited 2019 10 January]. Available from: https://apps.who.int/iris/bitstream/handle/10665/42590/9241562218.pdf?sequence=1. 189. Romano M, Cacciatore A, Giordano R, La Rosa B. Postpartum period: three distinct but continuous phases. J Prenat Med. 2010;4(2):22-5. 190. Allen A, Tosun N, Carlson S, Allen S. Postpartum Changes in Mood and Smoking- Related Symptomatology: An Ecological Momentary Assessment Investigation. Nicotine & Tobacco Research. 2017;20(6):681-9. 191. Hung C-H, Lin C-J, Stocker J, Yu C-Y. Predictors of postpartum stress. Journal of Clinical Nursing. 2011;20(5‐6):666-74. 192. Maxwell S. Rational prescribing: the principles of drug selection. Clinical Medicine. 2009;9(5):481-5. 193. Harrison-Woolrych M, Ashton J. Psychiatric Adverse Events Associated with Varenicline. Drug Safety. 2011;34(9):763-72.

193

194. Pasternak B, Svanström H, Hviid A. Use of varenicline versus bupropion and risk of psychiatric adverse events. Addiction. 2013;108(7):1336-43. 195. US Food and Drug Administration. Drug Safety Communications. FDA revises description of mental health side effects of the stop-smoking medicines Chantix (varenicline) and Zyban (bupropion) to reflect clinical trial findings, 2016[cited 2016 16 December]. Available from: https://www.fda.gov/Drugs/DrugSafety/ucm532221.htm. 196. Ascher JA, Cole JO, Colin JN, Feighner JP, Ferris RM, Fibiger HC, et al. Bupropion: a review of its mechanism of antidepressant activity. The Journal of clinical psychiatry. 1995;56(9):395-401. 197. Gadde KM, Xiong GL. Bupropion for weight reduction. Expert Review of Neurotherapeutics. 2007;7(1):17-24. 198. Lindson N, Chepkin SC, Ye W, Fanshawe TR, Bullen C, Hartmann‐Boyce J. Different doses, durations and modes of delivery of nicotine replacement therapy for smoking cessation. Cochrane Database of Systematic Reviews. 2019;Issue 4. Art. No.: CD013308. DOI: 10.1002/14651858.CD013308. 199. Foley KF, DeSanty KP, Kast RE. Bupropion: pharmacology and therapeutic applications. Expert Review of Neurotherapeutics. 2006;6(9):1249-65. 200. Australian Government Services Australia. Streamlined authority process,updated 7 May 2020 [cited 2020 20 June]. Available from: https://www.servicesaustralia.gov.au/organisations/health-professionals/topics/streamlined- authority-process/31576. 201. Cahill K, Stevens S, Perera R, Lancaster T. Pharmacological interventions for smoking cessation: an overview and network meta‐analysis. Cochrane Database of Systematic Reviews. 2013:Issue 5. Art. No.: CD009329. doi: 10.1002/14651858.CD009329.pub2. 202. Mills EJ, Wu P, Lockhart I, Wilson K, Ebbert JO. Adverse events associated with nicotine replacement therapy (NRT) for smoking cessation. A systematic review and meta- analysis of one hundred and twenty studies involving 177,390 individuals. Tobacco Induced Diseases. 2010;8(1):8. 203. Anthenelli RM, Benowitz NL, West R, St Aubin L, McRae T, Lawrence D, 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. The Lancet. 2016;387(10037):2507-20. 204. US Food and Drug Administration. FDA Drug Safety Communication: Chantix (varenicline) may increase the risk of certain cardiovascular adverse events in patients with cardiovascular disease, 2011[cited 2020 3 January]. Available from: https://www.fda.gov/drugs/drug-safety-and-availability/fda-drug-safety-communication-chantix- varenicline-may-increase-risk-certain-cardiovascular-adverse. 205. Pfizer. Australian Product Information - Champix (Varenilcine Tartrate), revised 15 March 2019 [cited 2019 5 August ]. Available from: https://www.ebs.tga.gov.au/ebs/picmi/picmirepository.nsf/pdf?OpenAgent&id=CP-2010-PI- 06102-3&d=202003231016933. 206. Prochaska JJ, Hilton JF. Risk of cardiovascular serious adverse events associated with varenicline use for tobacco cessation: systematic review and meta-analysis. British Medical Journal. 2012;344:e2856. 207. Kotz D, Viechtbauer W, Simpson C, van Schayck OCP, West R, Sheikh A. Cardiovascular and neuropsychiatric risks of varenicline: a retrospective cohort study. The Lancet Respiratory Medicine. 2015;3(10):761-8. 208. Aspen Pharmacare Australia Pty Ltd. Zyban SR (bupropion hydrochloride) sustained release tablets, revised 10 March 2020 [cited 2020 11 March]. Available from: https://www.ebs.tga.gov.au/ebs/picmi/picmirepository.nsf/pdf?OpenAgent&id=CP-2013-PI- 01138-1. 209. Lee ML, Tran DT, Welsh A, Kennedy D, Havard A. Health-care providers' concern regarding smoking cessation pharmacotherapies during pregnancy: Calls to a teratology information service. Drug and alcohol review. 2020;39(3):223-31.

194

210. Dietz PM, England LJ, Shapiro-Mendoza CK, Tong VT, Farr SL, Callaghan WM. Infant morbidity and mortality attributable to prenatal smoking in the US. American journal of preventive medicine. 2010;39(1):45-52. 211. Lumley J, Chamberlain C, Dowswell T, Oliver S, Oakley L, Watson L. Interventions for promoting smoking cessation during pregnancy. Cochrane Database of Systematic Reviews. 2009;Issue 3. Art. No.: CD001055. DOI: 10.1002/14651858.CD001055.pub3. 212. Fahy SJ, Cooper S, Coleman T, Naughton F, Bauld L. Provision of smoking cessation support for pregnant women in England: results from an online survey of NHS stop smoking services for pregnant women. BMC Health Services Research. 2014;14(1):107. 213. Eiden RD, Homish GG, Colder CR, Schuetze P, Gray TR, Huestis MA. Changes in Smoking Patterns During Pregnancy. Substance Use & Misuse. 2013;48(7):513-22. 214. Mohsin M, Bauman AE. Socio-demographic factors associated with smoking and smoking cessation among 426,344 pregnant women in New South Wales, Australia. BMC public health. 2005;5(1):138. 215. Price JH, Jordan TR, Dake JA. Obstetricians and gynecologists’ perceptions and use of nicotine replacement therapy. Journal of Community Health. 2006;31(3):160. 216. Hartmann KE, Wechter ME, Payne P, Salisbury K, Jackson RD, Melvin CL. Best Practice Smoking Cessation Intervention and Resource Needs of Prenatal Care Providers. Obstetrics & Gynecology. 2007;110(4):765-70. 217. Herbert R, Coleman T, Britton J. U.K. General Practitioners' Beliefs, Attitudes, and Reported Prescribing of Nicotine Replacement Therapy in Pregnancy. Nicotine & Tobacco Research. 2005;7(4):541-6. 218. Jordan TR, Dake JA, Price JH. Best practices for smoking cessation in pregnancy: Do obstetrician/gynecologists use them in practice? Journal of Women's Health. 2006;15(4):400- 41. 219. Bar-Zeev Y, Bonevski B, Gruppetta M, Twyman L, Atkins L, Palazzi K, et al. Clinician factors associated with prescribing nicotine replacement therapy in pregnancy: A cross- sectional survey of Australian obstetricians and general practitioners. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2018;58(3):366-70. 220. Oncken CA, Pbert L, Ockene JK, Zapka J, Stoddard A. Nicotine replacement prescription practices of obstetric and pediatric clinicians. Obstetrics & Gynecology. 2000;96(2):261-5. 221. Pijpers EL, Kreijkamp-Kaspers S, McGuire TM, Deckx L, Brodribb W, van Driel ML. Women's questions about medicines in pregnancy – An analysis of calls to an Australian national medicines call centre. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2017;57(3):334-41. 222. Gendron M-P, Martin B, Oraichi D, Bérard A. Health care providers’ requests to Teratogen Information Services on medication use during pregnancy and lactation. European journal of clinical pharmacology. 2009;65(5):523-31. 223. Richardson JL, Stephens S, Yates LM, Diav-Citrin O, Arnon J, Beghin D, et al. Pregnancy outcomes after maternal varenicline use: analysis of surveillance data collected by the European Network of Teratology Information Services. Reproductive Toxicology. 2017;67:26-34. 224. Lim JM, Sullivan E, Kennedy D. MotherSafe: Review of three years of counselling by an Australian teratology Information Service. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2009;49(2):168-72. 225. Corral JE, Yarur AJ, Diaz L, Simmons OL, Sussman DA. Cross-sectional analysis of patient phone calls to an inflammatory bowel disease clinic. Annals of gastroenterology. 2015;28(3):357-65. 226. Lau SM, McGuire TM, van Driel ML. Consumer concerns about paracetamol: a retrospective analysis of a medicines call centre. BMJ Open. 2016;6(6):e010860. 227. Office of the Australian Information Commissioner. Health: What are health service providers? [cited 2019 22 January ]. Available from: https://www.oaic.gov.au/individuals/faqs- for-individuals/health/what-are-health-service-providers 228. Australian Bureau of Statistics. 2033.0.55.001. Census of Population and Housing: Socio-Economic Indexes for Areas (SEIFA), Australia 2006 2006 [updated 2008]; cited 2017 12

195

January]. Available from: http://www.abs.gov.au/AUSSTATS/[email protected]/Lookup/2033.0.55.001Main+Features12006?Ope nDocument. 229. Australian Bureau of Statistics. Australian Standard Geographic Classification Remoteness Structure: Commonwealth of Australia; 2010 [cited 2017 28 January]. Available from: http://www.abs.gov.au/websitedbs/D3310114.nsf/home/remoteness+structure. 230. Tran DT HA, Preen DB, Einarsdottir K, Kemp-Casey A, Randall D, Jorm LR. Birth outcomes associated with the use of smoking cessation pharmacotherapies in pregnancy: Findings from the Smoking MUMS (Maternal Use of Medications and Safety) Study. . Pharmacoepidemiology and Drug Safety 2017;26(S2):435. 231. Smedberg J, Lupattelli A, Mårdby A-C, Nordeng H. Characteristics of women who continue smoking during pregnancy: a cross-sectional study of pregnant women and new mothers in 15 European countries. BMC Pregnancy and Childbirth. 2014;14(1):213. 232. Beck E, Lechner A, Schaefer C. Who seeks Teratology Information Service’s advice? Assessing the risk of selection bias in observational cohort studies on drug risks in pregnancy. Reproductive Toxicology. 2017;67:79-84. 233. Rothman KJ, Greenland S, Lash T. Modern Epidemiology. 3rd ed. Philadelphia, PA: Lippincott Williams & Wilkins 2008. 234. Hobel CJ, Hyvarinen MA, Okada DM, Oh W. Prenatal and intrapartum high-risk screening: I. Prediction of the high-risk neonate. American Journal of Obstetrics and Gynecology. 1973;117(1):1-9. 235. Aubry RH, Pennington JC. Identification and Evaluation of High-Risk Pregnancy: The Perinatal Concept. Clinical Obstetrics and Gynecology. 1973;16(1):3-27. 236. Australian Government Department of Health. Clinical Practice Guidelines: Pregnancy Care 2018 [cited 2020 7 February]. Available from: https://ranzcog.edu.au/RANZCOG_SITE/media/RANZCOG- MEDIA/Women's%20Health/Patient%20information/Pregnancy-Care-Guidelines.pdf. 237. Zeev YB, Bonevski B, Twyman L, Watt K, Atkins L, Palazzi K, et al. Opportunities Missed: A Cross-Sectional Survey of the Provision of Smoking Cessation Care to Pregnant Women by Australian General Practitioners and Obstetricians. Nicotine & Tobacco Research. 2017;19(5):636-41. 238. Bonevski B, Guillaumier A, Shakeshaft A, Farrell M, Tzelepis F, Walsberger S, et al. An organisational change intervention for increasing the delivery of smoking cessation support in addiction treatment centres: study protocol for a randomized controlled trial. Trials. 2016;17(1):290. 239. Colomar M, Tong VT, Morello P, Farr SL, Lawsin C, Dietz PM, et al. Barriers and Promoters of an Evidenced-Based Smoking Cessation Counseling During Prenatal Care in Argentina and Uruguay. Maternal and Child Health Journal. 2015;19(7):1481-9. 240. Beenstock J, Sniehotta FF, White M, Bell R, Milne EMG, Araujo-Soares V. What helps and hinders midwives in engaging with pregnant women about stopping smoking? A cross- sectional survey of perceived implementation difficulties among midwives in the North East of England. Implementation Science. 2012;7(1):36. 241. Naughton F, Hopewell S, Sinclair L, McCaughan D, McKell J, Bauld L. Barriers and facilitators to smoking cessation in pregnancy and in the post-partum period: The health care professionals’ perspective. British Journal of Health Psychology. 2018;23(3):741-57. 242. Webb TL, Sheeran P. Does changing behavioral intentions engender behavior change? A meta-analysis of the experimental evidence. Psychological Bulletin. 2006;132(2):249-68. 243. Eccles MP, Hrisos S, Francis J, Kaner EF, Dickinson HO, Beyer F, et al. Do self- reported intentions predict clinicians' behaviour: a systematic review. Implementation Science. 2006;1(1):28. 244. Raisch DW. A Model of Methods for Influencing Prescribing: Part II. A Review of Educational Methods, Theories of Human Inference, and Delineation of the Model. DICP. 1990;24(5):537-42. 245. Bandura A. Social cognitive theory of self-regulation. Organizational Behavior and Human Decision Processes. 1991;50(2):248-87.

196

246. Warreman EB, Lambregts MMC, Wouters RHP, Visser LG, Staats H, van Dijk E, et al. Determinants of in-hospital antibiotic prescription behaviour: a systematic review and formation of a comprehensive framework. Clinical Microbiology and Infection. 2019;25(5):538-45. 247. Liu C, Liu C, Wang D, Deng Z, Tang Y, Zhang X. Determinants of antibiotic prescribing behaviors of primary care physicians in Hubei of China: a structural equation model based on the theory of planned behavior. Antimicrobial Resistance & Infection Control. 2019;8(1):23. 248. Rashidian A, Russell I. Intentions and statins prescribing: can the Theory of Planned Behaviour explain physician behaviour in following guideline recommendations? Journal of evaluation in clinical practice. 2011;17(4):749-57. 249. Guthrie B, Kavanagh K, Robertson C, Barnett K, Treweek S, Petrie D, et al. Data feedback and behavioural change intervention to improve primary care prescribing safety (EFIPPS): multicentre, three arm, cluster randomised controlled trial. British Medical Journal. 2016;354:i4079. 250. Ponnet K, Wouters E, Van Hal G, Heirman W, Walrave M. Determinants of physicians’ prescribing behaviour of methylphenidate for cognitive enhancement. Psychology, Health & Medicine. 2014;19(3):286-95. 251. Ali S, Smith TL, Mican L, Brown C. Psychiatric Providers’ Willingness to Participate in Shared Decision Making When Prescribing Psychotropics. Journal of Pharmacy Technology. 2015;31(5):212-8. 252. Mc Grae McDermott M, Hahn EA, Greenland P, Cella D, Ockene JK, Brogan D, et al. Atherosclerotic Risk Factor Reduction in Peripheral Arterial Disease. Journal of General Internal Medicine. 2002;17(12):895-904. 253. Millstein SG. Utility of the theories of reasoned action and planned behavior for predicting physician behavior: A prospective analysis. Health Psychology. 1996;15(5):398-402. 254. Sinclair PM, Kable A, Levett-Jones T, Holder C, Oldmeadow CJ. The CKD-DETECT study: An RCT aimed at improving intention to initiate a kidney health check in Australian practice nurses. Journal of Clinical Nursing. 2019;28(15-16):2745-59. 255. Tsiantou V, Moschandreas J, Bertsias A, Papadakaki M, Saridaki A, Agius D, et al. General Practitioners’ intention to prescribe and prescribing patterns in selected European settings: The OTCSOCIOMED project. Health Policy. 2015;119(9):1265-74. 256. Amemori M, Michie S, Korhonen T, Murtomaa H, Kinnunen TH. Assessing implementation difficulties in tobacco use prevention and cessation counselling among dental providers. Implementation Science. 2011;6(1):50. 257. Freeman T, Roche AM, Williamson P, Pidd K. What Factors Need to be Addressed to Support Dental Hygienists to Assist Their Patients to Quit Smoking? Nicotine & Tobacco Research. 2012;14(9):1040-7. 258. Blankers M, Buisman R, Hopman P, van Gool R, van Laar M. Modelling intentions to provide smoking cessation support among mental health professionals in the Netherlands. Tobacco Induced Diseases. 2016;14(1):32. 259. Bodner ME, Rhodes RE, Miller WC, Dean E. Predictors of physical therapists’ intentions to counsel for smoking cessation: Implications for practice and professional education. Physiotherapy Theory and Practice. 2018:1-10. 260. Chiu Y, Chou Y, Chang Y, Chu C, Lin F, Lai C, et al. Using an extended theory of planned behaviour to predict smoking cessation counsellors’ intentions to offer smoking cessation support in the Taiwanese military: a cross-sectional study. BMJ Open. 2019;9(5):e026203. 261. Mas S, Bernard P, Gourlan M. Determinants of physical activity promotion by smoking cessation advisors. Patient Education and Counseling. 2018;101(11):1942-6. 262. Ajzen I. The theory of planned behavior. Organizational Behavior and Human Decision Processes. 1991;50:179-211. 263. Ajzen I. Perceived Behavioral Control, Self-Efficacy, Locus of Control, and the Theory of Planned Behavior. Journal of Applied Social Psychology. 2002;32(4):665-83. 264. Armitage CJ, Conner M. Efficacy of the Theory of Planned Behaviour: A meta-analytic review. British Journal of Social Psychology. 2001;40(4):471-99.

197

265. Rashidian A, Russell I. General practitioners' intentions and prescribing for asthma: using the theory of planned behavior to explain guideline implementation. Int J Prev Med. 2012;3(1):17-28. 266. Walker AE, Grimshaw JM, Armstrong EM. Salient beliefs and intentions to prescribe antibiotics for patients with a sore throat. British Journal of Health Psychology. 2001;6(4):347- 60. 267. Tseng Y, Wang K, Lin C, Lin Y, Pan H, Chang C. Predictors of smoking cessation in Taiwan: using the theory of planned behavior. Psychology, Health & Medicine. 2018;23(3):270- 6. 268. Ajzen I., M. F. The influence of attitudes on behavior. In: Albarracin D. JBT, Zanna M.P., editor. The Handbook of Attitudes. 1st eds ed. Mahwah, NJ: Erlbaum; 2005. p. 173–221. 269. Coleman-Cowger VH, Anderson BL, Mahoney J, Schulkin J. Smoking Cessation During Pregnancy and Postpartum: Practice Patterns Among Obstetrician-Gynecologists. Journal of addiction medicine. 2014;8(1):14-24. 270. Lutsenko H, Doran CM, Hall WD. Australian smokers' use of bupropion and nicotine replacement therapies and their relation to reimbursement, Australia 2001–05. Drug and alcohol review. 2008;27(2):160-4. 271. Mejia R, Martinez VG, Gregorich SE, Perez-Stable EJ. Physician counseling of pregnant women about active and secondhand smoking in Argentina. Acta Obstetricia et Gynecologica Scandinavica. 2010;89(4):490-5. 272. Nims L, Jordan T, Price J, Dake J, Khubchandani J. Smoking cessation education and training in obstetrics and gynecology residency programs in the United States. Journal of Family Medicine and Primary Care. 2019;8(3):1151-8. 273. Donnolley NR, Chambers GM, Butler-Henderson KA, Chapman MG, Sullivan EA. More than a name: Heterogeneity in characteristics of models of maternity care reported from the Australian Maternity Care Classification System validation study. Women and Birth. 2017;30(4):332-41. 274. Downe S, Finlayson K, Tunçalp Ö, Gülmezoglu AM. Provision and uptake of routine antenatal services: a qualitative evidence synthesis. Cochrane Database of Systematic Reviews. 2019;Issue 6. Art. No.: CD012392. DOI: 10.1002/14651858.CD012392.pub2. 275. Sandall J, Soltani H, Gates S, Shennan A, Devane D. Midwife‐led continuity models versus other models of care for childbearing women. Cochrane Database of Systematic Reviews. 2016;Issue 4. Art. No.: CD004667. DOI: 10.1002/14651858.CD004667.pub5. 276. Sardana M, Tang Y, Magnani Jared W, Ockene Ira S, Allison Jeroan J, Arnold Suzanne V, et al. Provider‐Level Variation in Smoking Cessation Assistance Provided in the Cardiology Clinics: Insights From the NCDR PINNACLE Registry. Journal of the American Heart Association. 2019;8(13):e011307. 277. Australian Bureau of Statistics. 1270.0.55.003 - Australian Statistical Geography Standard (ASGS): Volume 3 - Non ABS Structures, July 2016 [updated 28 July 2017]; cited 2019 28 September]. Available from: https://www.abs.gov.au/ausstats/[email protected]/Lookup/by%20Subject/1270.0.55.003~July%202016 ~Main%20Features~Postal%20Areas%20(POA)~8. 278. The Royal Australian and New Zealand College of Obstetricians and Gynaecologists (RANZCOG). 2017 Activities Report. East Melbourne, Victoria, Australia. 279. Eysenbach G. Improving the quality of Web surveys: the Checklist for Reporting Results of Internet E-Surveys (CHERRIES). Journal of medical Internet research. 2004;6(3):e34-e. 280. Cooke M, Mattick RP, Walsh RA. Differential uptake of a smoking cessation programme disseminated to doctors and midwives in antenatal clinics. Addiction. 2001;96(3):495-505. 281. Australian Health Ministers Advisory Council. National Survey of Access to Public Antenatal Care Services 2012. Canberra: Australian Government; 2013. 282. Rolfe MI, Donoghue DA, Longman JM, Pilcher J, Kildea S, Kruske S, et al. The distribution of maternity services across rural and remote Australia: does it reflect population need? BMC Health Services Research. 2017;17(1):163. 283. Commonwealth Government of Australia. National Strategic Framework for Rural and Remote Health.2010 [cited 2020 14 February ]. Available from: https://www1.health.gov.au/internet/main/publishing.nsf/Content/A76BD33A5D7A6897CA257F9

198

B00095DA3/$File/National%20Strategic%20Framework%20for%20Rural%20and%20Remote% 20Health.pdf. 284. Greenhalgh E, Stillman, S., & Ford, C. 7.10 Role of health professionals and social services. 2016. In: Tobacco in Australia: Facts and issues [Internet]. Melbourne: Cancer Council Victoria. Available from: http://www.tobaccoinaustralia.org.au/chapter-7-cessation/7-10-role-of- general-practice-and-other-health-pro. 285. Yeoh SGJ, Rolnik DL, Regan JA, Lee PYA. Experience and confidence in vaginal breech and twin deliveries among obstetric trainees and new specialists in Australia and New Zealand. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2019;59(4):545- 9. 286. Janda M, Armfield NR, Kerr G, Kurz S, Jackson G, Currie J, et al. Surgical approach to hysterectomy and barriers to using minimally invasive methods. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2018;58(6):690-5. 287. Cook C, Heath F, Thompson RL. A Meta-Analysis of Response Rates in Web- or Internet-Based Surveys. Educational and Psychological Measurement. 2000;60(6):821-36. 288. Cunningham CT, Quan H, Hemmelgarn B, Noseworthy T, Beck CA, Dixon E, et al. Exploring physician specialist response rates to web-based surveys. BMC Medical Research Methodology. 2015;15(1):32. 289. Flenady V, Macphail J, New K, Devenish-Meares P, Smith J. Implementation of a clinical practice guideline for smoking cessation in a public antenatal care setting. The Australian & New Zealand journal of obstetrics & gynaecology. 2008;48(6):552-8. 290. Dodd J, A. Crowther C. Vaginal birth after Caesarean section: A survey of practice in Australia and New Zealand. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2003;43(3):226-31. 291. Cheng HC, Black K, Woods C, de Costa C. Views and practices of induced abortion among Australian Fellows and trainees of The Royal Australian and New Zealand College of Obstetricians and Gynaecologists: A second study. Australian and New Zealand Journal of Obstetrics and Gynaecology.0(0). 292. Fisher J, Chatham E, Haseler S, McGaw B, Thompson J. Uneven implementation of the National Perinatal Depression Initiative: findings from a survey of Australian women's hospitals. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2012;52(6):559-64. 293. Gibbons K, Beckmann M, Flenady V, Rossouw D, Gardener G, Mahomed K, et al. A survey of policies for the monitoring of fetal growth in Australian and New Zealand hospitals. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2011;51(6):493-8. 294. Appleton B, Target C, Rasmussen M, Readman E, Sale F, Permezel M, et al. Knowledge and attitudes about vaginal birth after Caesarean section in Australian hospitals. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2000;40(2):195-9. 295. McColl E, Jacoby A, Thomas L, Soutter J, Bamford C, Steen N, et al. Design and use of questionnaires: a review of best practice applicable to surveys of health service staff and patients. Health technology assessment (Winchester, England). 2001;5(31):1-256. 296. Kapurubandara S, Qin V, Gurram D, Anpalagan A, Merkur H, Hogg R, et al. Opportunistic bilateral salpingectomy during gynaecological surgery for benign disease: A survey of current Australian practice. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2015;55(6):606-11. 297. Robertson P, Pontré J, Tucker P, Hart R. Are obstetric and gynaecological trainees in Australia and New Zealand adequately trained in the management of the infertile couple? Australian and New Zealand Journal of Obstetrics and Gynaecology. 2019;59(4):580-4. 298. Tuohy JF, Harding JE, Crowther CA, Bloomfield FH. Reported adherence to current antenatal corticosteroid guidelines in Australia and New Zealand. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2019;59(3):416-21. 299. Ellis JE, Tung A, Lee H, Lee H, Kasza K. Anesthesiologists’ preferences for preoperative cardiac evaluation before vascular surgery: results of a mail survey. Journal of Clinical Anesthesia. 2010;22(6):402-9. 300. Dillman DA, Smyth JD, Christian LM. Internet, phone, mail, and mixed mode surveys: The tailored design method, 4th ed. Hoboken, NJ, US: John Wiley & Sons Inc; 2014. xvii, 509- xvii, p.

199

301. Carson KV, Verbiest MEA, Crone MR, Brinn MP, Esterman AJ, Assendelft WJJ, et al. Training health professionals in smoking cessation. Cochrane Database of Systematic Reviews. 2012;Issue 5. Art. No.: CD0002:DOI:10.1002/14651858.CD000214.pub2. 302. The Health Consequences of Smoking-50 Years of Progress: A Report of the Surgeon General. Atlanta GA2014. 303. Tran D, Roberts C, Jorm L, Seeho S, Havard A. Change in smoking status during two consecutive pregnancies: a population-based cohort study. BJOG: An International Journal of Obstetrics & Gynaecology. 2014;121(13):1611-20. 304. Dietz PM, Adams MM, Rochat RW, Mathis MP. Prenatal Smoking in Two Consecutive Pregnancies: Georgia, 1989–1992. Maternal and Child Health Journal. 1997;1(1):43-51. 305. Thompson RA, Nelson CA. Developmental science and the media: Early brain development. American Psychologist. 2001;56(1):5-15. 306. Haran C, van Driel M, Mitchell BL, Brodribb WE. Clinical guidelines for postpartum women and infants in primary care–a systematic review. BMC Pregnancy and Childbirth. 2014;14(1):51. 307. Australian Institute of Health and Welfare 2020. Australia’s children. Canberra: AIHW. 308. Hatsukami D, Hughes JR, Pickens R. Characterization of tobacco withdrawal: physiological and subjective effects. In Grabowski J, Hall SM eds. Pharmacological Adjuncts in Smoking Cessation. Bethesda, Md: National Institute on Drug Abuse; 1985. p. 56-67. Monograph 53. 309. Cnattingius S, Akre O, Lambe M, Ockene J, Granath F. Will an adverse pregnancy outcome influence the risk of continued smoking in the next pregnancy? American Journal of Obstetrics and Gynecology. 2006;195(6):1680-6. 310. Varner SB, Ihongbe T, Masho SW. The Impact of Prior Poor Birth Outcomes on Smoking Behavior on Subsequent Pregnancies: Analysis of the National PRAMS Data. Maternal and Child Health Journal. 2016;20(3):583-92. 311. Havard A, Jorm LR, Preen D, Daube M, Kemp A, Einarsdóttir K, et al. The Smoking MUMS (Maternal Use of Medications and Safety) Study: protocol for a population-based cohort study using linked administrative data. BMJ Open. 2013;3(9):e003692. 312. Tran DT, Havard A, Jorm LR. Data cleaning and management protocols for linked perinatal research data: a good practice example from the Smoking MUMS (Maternal Use of Medications and Safety) Study. BMC Medical Research Methodology. 2017;17(1):97. 313. National Centre for Classification in Health. The International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD-10-AM). 5th ed. Sydney: National Centre for Classification in Health; 2006. 314. National Centre for Classification in Health. The Australian Classification of Health Interventions (ACHI) – Seventh Edition - Tabular list of interventions and Alphabetic index of interventions. Sydney: National Centre for Classification in Health, Faculty of Health Sciences, The University of Sydney; 2010. 315. Australian Institute of Health and Welfare. Australia’s health 2014. Australia’s health series no. 14. Cat. no. AUS 178. Canberra: AIHW. 316. Australian Government Department of Social Services. DSS Payment Demographic Data 2019 [cited 2019 3 March]. Available from: https://data.gov.au/dataset/ds-dga-cff2ae8a- 55e4-47db-a66d-e177fe0ac6a0/details?q=payment%20demographic. 317. Department of Health. History of PBS Copayments and Safety Net Thresholds Retrieved 25/05/2018 [Available from: http://www.pbs.gov.au/info/healthpro/explanatory- notes/front/fee. 318. Kelman CW, Bass AJ, Holman CDJ. Research use of linked health data — a best practice protocol. Australian and New Zealand Journal of Public Health. 2002;26(3):251-5. 319. Centre for Health Record Linkage (CHeReL). Master Linkage Key Quality Assurance. 2012 [cited 2019 26 June]. Available from: https://www.cherel.org.au/media/24160/qa_report_2012-a.pdf. 320. Holman CDAJ, Bass AJ, Rouse IL, Hobbs MST. Population-based linkage of health records in Western Australia: development of a health services research linked database. Australian and New Zealand Journal of Public Health. 1999;23(5):453-9.

200

321. Colvin L, Slack-Smith L, Stanley FJ, Bower C. Pharmacovigilance in pregnancy using population-based linked datasets. Pharmacoepidemiology and Drug Safety. 2009;18(3):211-25. 322. World Health Organization. WHO Collaborating Centre for Drug Statistics Methodology: ATC classification index with DDDs and Guidelines for ATC classification and DDD assignment. Oslo,Norway: Norwegian Institute of Public Health; 2006. 323. Australian Institute of Health and Welfare. Mental health services in Australia: Data sources. Cat. no. HSE 228. Canberra: AIHW; 2019. 324. World Health Organisation. WHO Collaborating Centre for Drug Statistics Methodology. ATC alterations from 2005-2019 [cited 2019 7 July]. Available from: https://www.whocc.no/atc_ddd_alterations__cumulative/atc_alterations/?order_by=1. 325. Chou D, Tunçalp Ö, Firoz T, Barreix M, Filippi V, von Dadelszen P, et al. Constructing maternal morbidity – towards a standard tool to measure and monitor maternal health beyond mortality. BMC Pregnancy and Childbirth. 2016;16(1):45. 326. Passey ME, Sanson-Fisher RW, D’Este CA, Stirling JM. Tobacco, alcohol and cannabis use during pregnancy: Clustering of risks. Drug and Alcohol Dependence. 2014;134:44-50. 327. Nordenberg D, Yip R, Binkin NJ. The Effect of Cigarette Smoking on Hemoglobin Levels and Anemia Screening. Journal of American Medical Association. 1990;264(12):1556-9. 328. Orton S, Coleman T, Coleman-Haynes T, Ussher M. Predictors of Postpartum Return to Smoking: A Systematic Review. Nicotine & Tobacco Research. 2018;20(6):665-73. 329. Stapleton JA, Watson L, Spirling LI, Smith R, Milbrandt A, Ratcliffe M, et al. Varenicline in the routine treatment of tobacco dependence: a pre–post comparison with nicotine replacement therapy and an evaluation in those with mental illness. Addiction. 2008;103(1):146- 54. 330. Hughes JR. Smoking and suicide: A brief overview. Drug and Alcohol Dependence. 2008;98(3):169-78. 331. Park ER, Chang Y, Quinn V, Regan S, Cohen L, Viguera A, et al. The association of depressive, anxiety, and stress symptoms and postpartum relapse to smoking: A longitudinal study. Nicotine & Tobacco Research. 2009;11(6):707-14. 332. Glassman AH, Covey LS, Stetner F, Rivelli S. Smoking cessation and the course of major depression: a follow-up study. The Lancet. 2001;357(9272):1929-32. 333. Allen J, Gamble J, Stapleton H, Kildea S. Does the way maternity care is provided affect maternal and neonatal outcomes for young women? A review of the research literature. Women Birth. 2012;25. 334. Gurillo P, Jauhar S, Murray RM, MacCabe JH. Does tobacco use cause psychosis? Systematic review and meta-analysis. The Lancet Psychiatry. 2015;2(8):718-25. 335. Edwards JH. Contribution of Cigarette Smoking to Respiratory Disease. British Journal of Preventive & Social Medicine. 1957;11(1):10-21. 336. Locke GR, Talley NJ, Fett SL, Zinsmeister AR, Melton LJ. Risk factors associated with symptoms of gastroesophageal reflux. The American Journal of Medicine. 1999;106(6):642-9. 337. Oncken C, Gonzales D, Nides M, et al. Efficacy and safety of the novel selective nicotinic acetylcholine receptor partial agonist, varenicline, for smoking cessation. Archives of Internal Medicine. 2006;166(15):1571-7. 338. Swan GE, Javitz HS, Jack LM, Wessel J, Michel M, Hinds DA, et al. Varenicline for smoking cessation: nausea severity and variation in nicotinic receptor genes. The Pharmacogenomics Journal. 2011;12:349. 339. Arnson Y, Shoenfeld Y, Amital H. Effects of tobacco smoke on immunity, inflammation and autoimmunity. Journal of Autoimmunity. 2010;34(3):J258-J65. 340. Rong L, Frontera AT, Benbadis SR. Tobacco smoking, epilepsy, and seizures. Epilepsy & Behavior. 2014;31:210-8. 341. Vestergaard P. Smoking and thyroid disorders--a meta-analysis. European Journal of Endocrinology. 2002;146(2):153-61. 342. Davidson NL, Wolski P, Callaway LK, Barrett HL, Fagermo N, Lust K, et al. Chronic kidney disease in pregnancy: Maternal and fetal outcomes and progression of kidney disease. Obstetric Medicine. 2015;8(2):92-8.

201

343. Mills EJ, Thorlund K, Eapen S, Wu P, Prochaska JJ. Cardiovascular events associated with smoking cessation pharmacotherapies: a network meta-analysis. Circulation. 2014;129(1):28-41. 344. Chen JS, Roberts C, Simpson J, Ford J. Use of hospitalisation history (lookback) to determine prevalence of chronic diseases: Impact on modelling of risk factors for haemorrhage in pregnancy. BioMed Central Medical Research Methodology. 2011;11(1):68. DOI:10.1186/471-2288-11-68. 345. Bell JC, Ford JB, Cameron CA, Roberts CL. The accuracy of population health data for monitoring trends and outcomes among women with diabetes in pregnancy. Diabetes Research and Clinical Practice. 2008;81(1):105-9. 346. Roberts CL, Algert CS, Morris JM, Ford JB, Henderson-Smart DJ. Hypertensive disorders in pregnancy: a population-based study. The Medical journal of Australia. 2005;182(7):332-5. 347. Roberts CL, Bell JC, Ford JB, Hadfield RM, Algert CS, Morris JM. The accuracy of reporting of the hypertensive disorders of pregnancy in population health data. Hypertens Pregnancy. 2008;27(3):285-97. 348. Pratt NL, Kerr M, Barratt JD, Kemp-Casey A, Kalisch Ellett LM, Ramsay E, et al. The validity of the Rx-Risk Comorbidity Index using medicines mapped to the Anatomical Therapeutic Chemical (ATC) Classification System. BMJ Open. 2018;8(4). 349. Roberts CL, Cameron CA, Bell JC, Algert CS, Morris JM. Measuring Maternal Morbidity in Routinely Collected Health Data: Development and Validation of a Maternal Morbidity Outcome Indicator. Medical Care. 2008;46(8):786-94. 350. Castles A, Adams EK, Melvin CL, Kelsch C, Boulton ML. Effects of smoking during pregnancy. Five meta-analyses. American journal of preventive medicine. 1999;16:208-15. 351. Lilford R, Van Coeverden De Groot H, Moore P, Bingham P. The relative risks of caesarean section (intrapartum and elective) and vaginal delivery: a detailed analysis to exclude the effects of medical disorders and other acute pre-existing physiological disturbances. BJOG: An International Journal of Obstetrics & Gynaecology. 1990;97(10):883-92. 352. Hawkins Joy L, MD, Koonin Lisa M, MN, MPH, Palmer Susan K, MD, Gibbs Charles P, MD. Anesthesia-related Deaths during Obstetric Delivery in the United States, 1979–1990. Anesthesiology: The Journal of the American Society of Anesthesiologists. 1997;86(2):277-84. 353. Lain SJ, Algert CS, Nassar N, Bowen JR, Roberts CL. Incidence of Severe Adverse Neonatal Outcomes: Use of a Composite Indicator in a Population Cohort. Maternal and Child Health Journal. 2012;16(3):600-8. 354. Thorngren-Jerneck K, Herbst A. Low 5-minute Apgar score: a population-based register study of 1 million term births. Obstetrics & Gynecology. 2001;98(1):65-70. 355. Organisation WH. Recommended Definitions, Terminology and Format for Statistical Tables Related to The Perinatal Period And Use of A New Certificate For Cause of Perinatal Deaths. Acta Obstetricia et Gynecologica Scandinavica. 1977;56(3):247-53. 356. Dobbins TA, Sullivan EA, Roberts CL, Simpson JM. Australian national birthweight percentiles by sex and gestational age, 1998–2007. Medical Journal of Australia. 2012;197(5):291-4. 357. Taylor Lee. Hospital procedures for review and reporting of perinatal deaths2006 [cited 2020 9 March]. Available from: https://www1.health.nsw.gov.au/pds/ArchivePDSDocuments/GL2005_013.pdf. 358. Apgar V, Holaday DA, James LS, Weisbrot IM, Berrien C. Evaluation of the newborn infant-second report. Journal of American Medical Association. 1958;168(15):1985-8. 359. Kurti AN, Davis DR, Skelly JM, Redner R, Higgins ST. Comparison of nicotine dependence indicators in predicting quitting among pregnant smokers. Experimental and Clinical Psychopharmacology. 2016;24(1):12-7. 360. Australian Institute of Health and Welfare. Maternal deaths in Australia 2012–2014. Cat. no. PER 92. Canberra: AIHW; 2017. 361. Australian Bureau of Statistics. Australian Statistical Geography Standard (ASGS), Volume 5 - Remoteness Structure, July 2011 Cat. No: 1270.0.55.005. Canberra: ABS; 2013.

202

362. Schmied V, Donovan J, Kruske S, Kemp L, Homer C, Fowler C. Commonalities and challenges: A review of Australian state and territory maternity and child health policies. Contemporary Nurse. 2011;40(1):106-17. 363. Vittinghoff E, McCulloch CE. Relaxing the Rule of Ten Events per Variable in Logistic and Cox Regression. American Journal of Epidemiology. 2006;165(6):710-8. 364. SURE: Secure Unified Research Environment [online]. 2017 [Available from: https://www.saxinstitute.org.au/our-work/sure/. 365. Kearney PM, Baigent C, Godwin J, Halls H, Emberson JR, Patrono C. Do selective cyclo-oxygenase-2 inhibitors and traditional non-steroidal anti-inflammatory drugs increase the risk of atherothrombosis? Meta-analysis of randomised trials. British Medical Journal. 2006;332(7553):1302-8. 366. FitzGerald GA, Patrono C. The Coxibs, Selective Inhibitors of Cyclooxygenase-2. New England Journal of Medicine. 2001;345(6):433-42. 367. Schjerning Olsen A-M, Gislason GH, McGettigan P, Fosbøl E, Sørensen R, Hansen ML, et al. Association of NSAID Use With Risk of Bleeding and Cardiovascular Events in Patients Receiving Antithrombotic Therapy After Myocardial Infarction. Journal of American Medical Association. 2015;313(8):805-14. 368. Mohsin M, Jalaludin B. Influence of previous pregnancy outcomes and continued smoking on subsequent pregnancy outcomes: an exploratory study in Australia. BJOG: An International Journal of Obstetrics & Gynaecology. 2008;115(11):1428-35. 369. Winickoff JP, Healey EA, Regan S, Park ER, Cole C, Friebely J, et al. Using the Postpartum Hospital Stay to Address Mothers' and Fathers' Smoking: The NEWS Study. Pediatrics. 2010;125(3):518-25. 370. Verbiest S, McClain E, Stuebe A, Menard MK. Postpartum Health Services Requested by Mothers with Newborns Receiving Intensive Care. Maternal and Child Health Journal. 2016;20(1):125-31. 371. Borland T, Babayan A, Irfan S, Schwartz R. Exploring the adequacy of smoking cessation support for pregnant and postpartum women. BMC public health. 2013;13(1):1-11. 372. McMillen R, O'Connor KG, Groner J, Tanski S, Park ER, Klein JD. Changes and Factors Associated With Tobacco Counseling: Results From the AAP Periodic Survey. Academic Pediatrics. 2017;17(5):504-14. 373. Bonollo DP, Zapka JG, Stoddard AM, Ma Y, Pbert L, Ockene JK. Treating nicotine dependence during pregnancy and postpartum: understanding clinician knowledge and performance. Patient Education and Counseling. 2002;48(3):265-74. 374. Taylor LK, Travis S, Pym M, Olive E, Henderson-Smart DJ. How useful are hospital morbidity data for monitoring conditions occurring in the perinatal period? Australian and New Zealand Journal of Obstetrics and Gynaecology. 2005;45(1):36-41. 375. Taylor L, Pym M, Bajuk B, Sutton L, Travis S, Banks C. Part 8: Validation Study NSW Midwives Data Collection 1998. New South Wales Public Health Bulletin Supplementary Series. 2000;11(1):97-9. 376. Hess LM, Raebel MA, Conner DA, Malone DC. Measurement of Adherence in Pharmacy Administrative Databases: A Proposal for Standard Definitions and Preferred Measures. Annals of Pharmacotherapy. 2006;40(7-8):1280-8. 377. Fish LJ, Peterson BL, Namenek Brouwer RJ, Lyna P, Oncken CA, Swamy GK, et al. Adherence to nicotine replacement therapy among pregnant smokers. Nicotine & Tobacco Research. 2009;11(5):514-8. 378. Davies NM, Taylor GMJ, Taylor AE, Jones T, Martin RM, Munafò MR, et al. The effects of prescribing varenicline on two-year health outcomes: an observational cohort study using electronic medical records. Addiction. 2018;113(6):1105-16. 379. Itani T, Martin R, Rai D, Jones T, Taylor G, Thomas K, et al. Use of varenicline and nicotine replacement therapy in people with and without general practitioner-recorded dementia: retrospective cohort study of routine electronic medical records. BMJ Open. 2019;9(8):e027569. 380. Roberts CL, Bell JC, Ford JB, Hadfield RM, Algert CS, Morris JM. The Accuracy of Reporting of the Hypertensive Disorders of Pregnancy in Population Health Data. Hypertension in Pregnancy. 2008;27(3):285-97.

203

381. Chen JS, Roberts CL, Simpson JM, Ford JB. Prevalence of pre-eclampsia, pregnancy hypertension and gestational diabetes in population-based data: Impact of different ascertainment methods on outcomes. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2012;52(1):91-5. 382. West R, Evins AE, Benowitz NL, Russ C, McRae T, Lawrence D, et al. Factors associated with the efficacy of smoking cessation treatments and predictors of smoking abstinence in EAGLES. Addiction. 2018;113(8):1507-16. 383. Rojewski AM, Nicotine tCWotSfRo, Network TT, Baldassarri S, Nicotine tCWotSfRo, Network TT, et al. Exploring Issues of Comorbid Conditions in People Who Smoke. Nicotine & Tobacco Research. 2016;18(8):1684-96. 384. Ralston S, Roohi M. A randomized, controlled trial of smoking cessation counseling provided during child hospitalization for respiratory illness. Pediatric Pulmonology. 2008;43(6):561-6. 385. Nabi-Burza E, Drehmer JE, Hipple Walters B, Rigotti NA, Ossip DJ, Levy DE, et al. Treating Parents for Tobacco Use in the Pediatric Setting: The Clinical Effort Against Secondhand Smoke Exposure Cluster Randomized Clinical Trial. Journal of American Medical Association Pediatrics. 2019;173(10):931-9. 386. Winickoff JP, Nabi-Burza E, Chang Y, Regan S, Drehmer J, Finch S, et al. Sustainability of a Parental Tobacco Control Intervention in Pediatric Practice. Pediatrics. 2014;134(5):933-41. 387. Hall ES, Venkatesh M, Greenberg JM. A population study of first and subsequent pregnancy smoking behaviors in Ohio. Journal Of Perinatology. 2016;36:948. 388. Hauge LJ, Aarø LE, Torgersen L, Vollrath ME. Smoking During Consecutive Pregnancies Among Primiparous Women in the Population-Based Norwegian Mother and Child Cohort Study. Nicotine & Tobacco Research. 2013;15(2):428-34. 389. Louis JM, Bryant A, Ramos D, Stuebe A, Blackwell SC. Interpregnancy Care. American Journal of Obstetrics and Gynecology. 2019;220(1):B2-B18. 390. Dorney E, Black K. Preconception care. Australian Journal for General Practitioners. 2018;47:424-9. 391. Prepregnancy counseling: Committee Opinion No. 762. Fertility and Sterility. 2019;111(1):32-42. 392. Shiffman S, Di Marino ME, Sweeney CT. Characteristics of selectors of nicotine replacement therapy. Tobacco Control. 2005;14(5):346-55. 393. Cokkinides VE, Ward E, Jemal A, Thun MJ. Under-use of smoking-cessation treatments: Results from the National Health Interview Survey, 2000. American journal of preventive medicine. 2005;28(1):119-22. 394. Shiffman S, Brockwell SE, Pillitteri JL, Gitchell JG. Use of Smoking-Cessation Treatments in the United States. American journal of preventive medicine. 2008;34(2):102-11. 395. Kasza KA, Hyland AJ, Borland R, McNeill AD, Bansal-Travers M, Fix BV, et al. Effectiveness of stop-smoking medications: findings from the International Tobacco Control (ITC) Four Country Survey. Addiction. 2013;108(1):193-202. 396. Kotz D, Brown J, West R. ‘Real-world’ effectiveness of smoking cessation treatments: a population study. Addiction. 2014;109(3):491-9. 397. Walsh RA, Walsh RA. Over-the-counter nicotine replacement therapy: a methodological review of the evidence supporting its effectiveness. Drug and alcohol review. 2008;27(5):529- 47. 398. Conde-Agudelo A, Belizán JM. Maternal morbidity and mortality associated with interpregnancy interval: cross sectional study. British Medical Journal. 2000;321(7271):1255-9. 399. Hanley GE, Hutcheon JA, Kinniburgh BA, Lee L. Interpregnancy Interval and Adverse Pregnancy Outcomes: An Analysis of Successive Pregnancies. Obstetrics & Gynecology. 2017;129(3):408-15. 400. Ahrens KA, Nelson H, Stidd RL, Moskosky S, Hutcheon JA. Short interpregnancy intervals and adverse perinatal outcomes in high-resource settings: An updated systematic review. Paediatric and perinatal epidemiology. 2019;33(1):O25-O47.

204

401. Zhong Z, Zhao S, Zhao Y, Xia S. Combination therapy of varenicline and bupropion in smoking cessation: A meta-analysis of the randomized controlled trials. Comprehensive Psychiatry. 2019;95:152125. 402. Chang P-H, Chiang C-H, Ho W-C, Wu P-Z, Tsai J-S, Guo F-R. Combination therapy of varenicline with nicotine replacement therapy is better than varenicline alone: a systematic review and meta-analysis of randomized controlled trials. BMC public health. 2015;15(1):689. 403. Kotz D, Brown J, West R. Prospective cohort study of the effectiveness of varenicline versus nicotine replacement therapy for smoking cessation in the “real world”. BMC public health. 2014;14(1):1163. 404. Jackson SE, Kotz D, West R, Brown J. Moderators of real-world effectiveness of smoking cessation aids: a population study. Addiction. 2019;114(9):1627-38. 405. McKee SA, Smith PH, Kaufman M, Mazure CM, Weinberger AH. Sex Differences in Varenicline Efficacy for Smoking Cessation: A Meta-Analysis. Nicotine & Tobacco Research. 2015;18(5):1002-11. 406. Benowitz NL, Bernert JT, Foulds J, Hecht SS, Jacob P, III, Jarvis MJ, et al. Biochemical Verification of Tobacco Use and Abstinence: 2019 Update. Nicotine & Tobacco Research. 2019;22(7):1086-97. 407. Dietz PM, Homa D, England LJ, Burley K, Tong VT, Dube SR, et al. Estimates of nondisclosure of cigarette smoking among pregnant and nonpregnant women of reproductive age in the United States. American Journal of Epidemiology. 2011;173. 408. Bar-Zeev Y, Lim LL, Bonevski B, Gruppetta M, Gould GS. Nicotine replacement therapy for smoking cessation during pregnancy. Medical journal of Australia. 2018;208(1):46-51. 409. Ellenberg JH. Selection bias in observational and experimental studies. Statistics in Medicine. 1994;13(5‐7):557-67. 410. Cha S, Erar B, Niaura RS, Graham AL. Baseline Characteristics and Generalizability of Participants in an Internet Smoking Cessation Randomized Trial. Annals of Behavioral Medicine. 2016;50(5):751-61. 411. Piper ME, Smith SS, Schlam TR, Fiore MC, Jorenby DE, Fraser D, et al. A Randomized Placebo-Controlled Clinical Trial of 5 Smoking Cessation Pharmacotherapies. Archives of General Psychiatry. 2009;66(11):1253-62. 412. Härkönen J, Lindberg M, Karlsson L, Karlsson H, Scheinin NM. Education is the strongest socio-economic predictor of smoking in pregnancy. Addiction. 2018;113(6):1117-26. 413. Pullon S, Webster M, McLeod D, Benn C, Morgan S. Smoking cessation and nicotine replacement therapy in current primary maternity care. Aust Fam Physician. 2004;33(1-2):94-6. 414. Grimshaw JM, Eccles MP, Lavis JN, Hill SJ, Squires JE. Knowledge translation of research findings. Implementation Science. 2012;7(1):50. 415. Thomson R, McDaid L, Emery J, Naughton F, Cooper S, Dyas J, et al. Knowledge and Education as Barriers and Facilitators to Nicotine Replacement Therapy Use for Smoking Cessation in Pregnancy: A Qualitative Study with Health Care Professionals. International journal of environmental research and public health. 2019;16(10):1814.

205

Appendix Chapter Two Appendix Appendix 2

206

207

208

209

210

211

212

213

Chapter Three Appendices Appendix 3A Email invitation to participate in the survey

214

215

216

217

218

219

220

Appendix 3B WHA website recruitment of survey participants

221

Appendix 3C Full questionnaire

222

223

224

225

226

227

228

229

230

Appendix 3D Ethics approval

231

Chapter Four Appendices Appendix 4A Ascertainment of pre-existing maternal morbidities from hospital and PBS dispensing data Morbidities Hospital ICD-10-AM codes (periods of the PBS dispensing ATC codes (365-day lookback period) admission indicated below) Mental health Lookback and gestation periods: Anxiety: N05BA01 - N05BA12, N05BE01 F31-F34, F38, F39, F40, F41, F44, F48, F20-F25, F28-F30, O99.5 Bipolar: N05AN01*

Depression: N06AA01-N06AG02, N06AX03 - N06AX11, N06AX13 - N06AX18, N06AX21 - N06AX26

Psychotic illness: N05AA01 - N05AB02, N05AB06 - N05AL07, N05AX07 - N05AX13 Chronic airway Look back and gestation periods: J32, J35, J37, J40, Chronic airways disease: R03AC02 - R03DC03, R03DX05 J41, J42, J43, J44, , J47, R05, O99.5

Lookback period only: J45, J46, J98, J99, Gastro-oesophageal Lookback period only: K21.0, K21.9 Gastro-oesophageal reflux disease: A02BA01 - A02BX05 reflux Use of NSAIDS NSAIDS for pain: M01AB01 - M01AH06

Use of steroids Steroid responsive diseases: H02AB01 - H02AB10

Anaemia and Lookback and gestation periods: D56-D57, D65- Anticoagulants: B01AA03 - B01AB06, B01AE07, B01AF01, coagulation D68, D50-D53, D55, D58-D64 B01AF02, B01AX05

Antiplatelets: B01AC04 - B01AC07, B01AC12 - B01AC30, PBS items (05030R, 05035B, 05042J, 10111E, 10117L, 10129D, 10130E, 05751Q, 06456T) †

232

Morbidities Hospital ICD-10-AM codes (periods of the PBS dispensing ATC codes (365-day lookback period) admission indicated below) Drug and alcohol Lookback and gestation period: F10, Z50.2, Z72.1, * Alcohol dependency: N07BB01 - N07BB99 disorder F11, F12, F13, F14, F15, F16, F18, F19, Z50.3, Z72.2 Thyroid Lookback only: E00-E07, E89.0 Hyperthyroidism: H03BA02 - H03BB01

Hypothyroidism: H03AA01 - H03AA02

Cardiovascular Lookback and gestation period: I05-I09, I34-I39, I50, Congestive heart failure: C03DA02 - C03DA99, C07AB07, I20, I25, I27, I28, Q20-Q25, O99.4 C07AB12, C07AG02, [(C03CA01 - C03CC01) and (C09AA01- C09AX99 or C09CA01 - C09CX99)], PBS items (08732N, Lookback period only: I00-I02, I21-I24, I26, I30-I33, 08733P, 08734Q, 08735R) ‡ I40-I43, I44-I49, I51-I52, I60-I64, G45.8, G45.9, I65, I66, I67.2, I70, I73, I74, I77 Ischaemic heart disease-hypertension: C07AA01 - C07AA06, C07AA08 - C07AB01, [C07AB02 if PBS item code is not (08732N, 08733P, 08734Q, 08735R) ‡], C07AB03, C07AG01, C08CA01 - C08DB01, C09BB02 - C09BB10, C09DB01 - C09DB04, C09DX01, C09DX03, C10BX03 § Pre-existing diabetes Lookback and gestation period: E10, E11, E13, E14, O24.0, O24.1, O24.2, O24.3, supplemented by perinatal data £

Pre-existing Lookback and gestation period I10, I11, I12, I13, I15, hypertension supplemented by perinatal data £

Epilepsy Lookback period only: G40, F80.3 Epilepsy: N03AA01 - N03AX99

233

Morbidities Hospital ICD-10-AM codes (periods of the PBS dispensing ATC codes (365-day lookback period) admission indicated below) Chronic renal disease Lookback and gestation period: N02-N08, N1-N12, Renal disease: A11CC01 - A11CC04, B03XA01 - B03XA03, N14-N16, N18-N19, N25-N28, Q60-Q63, N39.1, V03AE02, V03AE03, V03AE05 N39.2, T82.4, T86.1, Z49, Z94.0, Z99.2

Lookback period only: N00, N01, N17

* Lithium (WHO N05AN01 code) was recorded as N06AX in PBS data [323] †: These PBS items are epoprostenol and iloprost. ‡: These PBS items are metoprolol succinate. § Combination product for hyperlipidaemia and ischaemic heart disease: hypertension £: Perinatal record supplements the identification of pre-existing diabetes and hypertension if these conditions were not recorded in hospital records.

234

Appendix 4B Components and ICD10 codes for Maternal Morbidity Outcome Indicator (MMOI), adapted from Roberts et al. (2008)

235

Appendix 4C Components and ICD10 codes for Neonatal Adverse Outcome Indicator (NAOI), adapted from Lain et al. (2012)[353]

Diagnosis Data source ICD10 diagnosis code Gestational age < 32 weeks Birth and hospital data Birthweight < 1,500 g Birth and hospital data Death (within 28 days of birth or before a Birth, hospital and ABS discharge home from hospital) mortality data Respiratory distress syndrome Hospital data P22.0 Seizure Hospital data P90, R56 Intraventricular hemorrhage (grades 2, 3 and Hospital data P52.1, P52.2 4) Cerebral infarction Hospital data I63 Periventricular leukomalacia Hospital data P91.2 Birth trauma (intracranial haemorrhage Hospital data P10.0 to P10.3, P13.0, P13.2, P13.3, P14.0, paralysis due to brachial plexus injury, skull or P14.1 long bone fracture) Hypoxic ischemic encephalopathy Hospital data P91.5, P91.81, P91.6 Necrotising enterocolitis Hospital data P77 Broncho-pulmonary dysplasia Hospital data P27.1 Sepsis/septicaemia (streptococcus, Hospital data P36, A40, A41.5, A41.9, B95.1, B96.2 staphylococcus, E. coli, unspecified Gram- negative) Pneumonia Hospital data P23, J12 to J18

236

Diagnosis Data source ICD10 diagnosis code Other respiratory: primary atelectasis, Hospital data P28.0, P28.5 respiratory failure Procedure Data source ACHI ICD10 procedure codes Resuscitation Hospital data and birth 92052, 92053, 92042–00, 90225 data Ventilatory support (mechanical ventilation Hospital data and birth 13882, 13857–00, 13879–00, 22007, 90179, and/or CPAP) data 92038, 92039 Central venous or arterial catheter Hospital data 38206, 13303–00, 34524–00, 34530–01, 13300–00, 13300–02, 13319–00, 13815 Transfusion of blood or blood products Hospital data 13706–01 to 04, 92206–00, 13306–00 Pneumothorax requiring an intercostal catheter Hospital data 38409–00 Any body cavity surgical procedure Hospital data 30373, 30375, 30378–00, 30562, 30564 to 30566, 30571, 30601, 30615–00, 30617–00, 32123–00, 36516, 36537, 36564, 36579, 38403–00, 38600–00, all codes start with 387, 39015, 39640–00, 40003, 40100–00, 40103–00, 41883, 43801–00, 43807–00, 43816–02, 43837, 43843, 43852–00, 43864, 43867, 43870–00, 43873, 43876, 43900–00, 43915–00, 43930–00, 43945– 00, 43963–00, 43978, 90180, 90224–00 Any intravenous fluids Hospital data 96199

237

Diagnosis Data source ICD10 diagnosis code Gestational age < 32 weeks Birth and hospital data Birthweight < 1,500 g Birth and hospital data Death (within 28 days of birth or before a Birth, hospital and ABS discharge home from hospital) mortality data Respiratory distress syndrome Hospital data P22.0 Seizure Hospital data P90, R56 Intraventricular hemorrhage (grades 2, 3 and Hospital data P52.1, P52.2 4) Cerebral infarction Hospital data I63 Periventricular leukomalacia Hospital data P91.2 Birth trauma (intracranial hemorrhage paralysis Hospital data P10.0 to P10.3, P13.0, P13.2, P13.3, P14.0, due to brachial plexus injury, skull or long bone P14.1 fracture) Hypoxic ischemic encephalopathy Hospital data P91.5, P91.81, P91.6 Necrotising enterocolitis Hospital data P77 Broncho-pulmonary dysplasia Hospital data P27.1 Sepsis/septicaemia (streptococcus, Hospital data P36, A40, A41.5, A41.9, B95.1, B96.2 staphylococcus, E. coli, unspecified Gram- negative) Pneumonia Hospital data P23, J12 to J18 Other respiratory: primary atelectasis, Hospital data P28.0, P28.5 respiratory failure

238

Procedure Data source ACHI ICD10 procedure codes Resuscitation Hospital data and birth 92052, 92053, 92042–00, 90225 data Ventilatory support (mechanical ventilation Hospital data and birth 13882, 13857–00, 13879–00, 22007, 90179, and/or CPAP) data 92038, 92039 Central venous or arterial catheter Hospital data 38206, 13303–00, 34524–00, 34530–01, 13300–00, 13300–02, 13319–00, 13815 Transfusion of blood or blood products Hospital data 13706–01 to 04, 92206–00, 13306–00 Pneumothorax requiring an intercostal catheter Hospital data 38409–00 Any body cavity surgical procedure Hospital data 30373, 30375, 30378–00, 30562, 30564 to 30566, 30571, 30601, 30615–00, 30617–00, 32123–00, 36516, 36537, 36564, 36579, 38403–00, 38600–00, all codes start with 387, 39015, 39640–00, 40003, 40100–00, 40103–00, 41883, 43801–00, 43807–00, 43816–02, 43837, 43843, 43852–00, 43864, 43867, 43870–00, 43873, 43876, 43900–00, 43915–00, 43930–00, 43945– 00, 43963–00, 43978, 90180, 90224–00 Any intravenous fluids Hospital data 96199

239

Appendix 4D Correlation matrix for each potential confounder

Quantity of Indigenous Country of Maternal SES Remoteness Parity Partner cigarettes Status birth age smoked

1 0.1007 -0.267 0.1564 -0.0398 -0.1005 0.1530 -0.0472 Indigenous Status <.0001 <.0001 <.0001 0.0114 <.0001 <.0001 0.0027 0.1007 1 -0.271 -0.0208 -0.0332 -0.0090 0.0150 -0.0243 SES <.0001 <.0001 0.1877 0.0352 0.5669 0.3424 0.1239 -0.2672 -0.2710 1 -0.2023 0.0565 -0.0312 -0.0840 0.0476 Remoteness <.0001 <.0001 <.0001 0.0003 0.0479 <.0001 0.0026 Country of birth 0.1564 -0.0208 -0.202 1 0.0716 -0.0543 0.1598 -0.1047 <.0001 0.1877 <.0001 <.0001 0.0006 <.0001 <.0001 -0.0398 -0.0332 0.0565 0.0716 1 -0.0855 0.6158* 0.1361 Parity 0.0114 0.0352 0.0003 <.0001 <.0001 <.0001 <.0001 Living with -0.1005 -0.0090 -0.0312 -0.0543 -0.0855 1 -0.1034 0.0918 partner <.0001 0.5669 0.0479 0.0006 <.0001 <.0001 <.0001 0.1530 0.0150 -0.0840 0.1598 0.6158 -0.1034 1 0.0675 Maternal age <.0001 0.3424 <.0001 <.0001 <.0001 <.0001 <.0001 Quantity of -0.0472 -0.0243 0.0476 -0.1047 0.1361 0.0918 0.0675 1 cigarettes smoked 0.0027 0.1239 0.0026 <.0001 <.0001 <.0001 <.0001 Spearman correlation coefficients and p-values are presented *correlations with >0.50 denoted in bold 240

Appendix 4E Association between maternal adverse events in relation to varenicline use in the 12 months postpartum, 2011-2012

Characteristics No pharmaco- Varenicline OR 95% CI aOR 95% CI therapy Total 4,806 % 392 % Model 2 (adjusted for maternal morbidities and potential confounders)

Maternal age < 25 2,009 41.8% 159 40.6% 1.00 ref 1.00 ref 25-<35 2,219 46.2% 203 51.8% 1.16 0.93 1.44 1.13 0.90 1.41 >35 578 12.0% 30 7.7% 0.66 0.44 0.98 0.60 0.40 0.92 Country of birth Australia 4,452 92.6% 379 96.7% 1.00 ref 1.00 ref Overseas 354 7.4% 13 3.3% 0.43 0.25 0.76 0.35 0.19 0.65 Living with a partner Yes 2,129 44.3% 170 43.4% 1.00 ref 1.00 ref No 2,658 55.3% 222 56.6% 0.96 0.78 1.18 0.97 0.78 1.20 Remoteness of residence Major cities 2,490 51.8% 221 56.4% 1.00 ref 1.00 ref Regional & remote 2,290 47.6% 170 43.4% 0.84 0.68 1.03 0.87 0.70 1.09 SES of residence Disadvantaged 2,290 47.6% 160 40.8% 1.00 ref 1.00 ref Advantaged 1,341 27.9% 127 32.4% 1.30 1.00 1.68 1.24 0.95 1.62 Average 1,149 23.9% 104 26.5% 1.36 1.06 1.73 1.27 0.99 1.62

241

Characteristics No pharmaco- Varenicline OR 95% CI aOR 95% CI therapy

Indigenous Status No 3,586 74.6% 313 79.8% 1.00 ref 1.00 ref Yes 1,219 25.4% 79 20.2% 0.74 0.58 0.96 0.73 0.56 0.95 Quantity smoked after 20 weeks of pregnancy < 10 sticks 3,579 74.5% 277 70.7% 1.00 ref 1.00 ref >= 10 sticks 1,145 23.8% 108 27.6% 1.22 0.97 1.54 1.23 0.97 1.56 Maternal morbidity** Hypertension (including gestational) 193 4.0% 22 5.6% 1.42 0.90 2.24 1.39 0.88 2.21 Diabetes (including gestational) 163 3.4% 16 4.1% 1.21 0.72 2.05 1.12 0.64 1.97 GORD 231 4.8% 26 6.6% 1.41 0.93 2.14 1.36 0.88 2.11 Mood 1,027 21.4% 103 26.3% 1.31 1.04 1.66 1.32 1.03 1.69 Anxiety 315 6.6% 26 6.6% 1.01 0.67 1.53 0.99 0.64 1.54 Psychosis 234 4.9% 16 4.1% 0.83 0.50 1.40 0.78 0.46 1.34 Drugs and alcohol 333 6.9% 19 4.8% 0.68 0.43 1.10 0.67 0.41 1.11 Respiratory 754 15.7% 67 17.1% 1.11 0.84 1.46 1.02 0.76 1.37 Use of NSAIDS 353 7.3% 42 10.7% 1.81 1.32 2.49 1.48 1.04 2.10 Epilepsy 88 1.8% 5 1.3% 0.69 0.28 1.72 0.68 0.27 1.71 Use of steroids 195 4.1% 14 3.6% 0.88 0.50 1.52 0.82 0.46 1.46 Blood coagulation 109 2.3% 6 1.5% 0.67 0.29 1.53 0.70 0.30 1.63 Maternal adverse events, composite# 456 9.5% 45 11.5% 1.24 0.89 1.71 1.21 0.86 1.69

#adjusted for maternal morbidities and potential confounders and values denoted in bold *aOR: adjusted odds ratio ** Women who did not have the morbidity were the reference category *** Percentages do not add to 100% due to missing data

242

Appendix 4F Association between neonatal adverse events in relation to varenicline use in the 12 months postpartum, 2011-2012, adjusted for maternal morbidities and potential confounders Characteristics No Varenicline OR 95% CI aOR 95% CI pharmacotherapy Total 4,806 % 392 % Model 3 (adjusted for maternal morbidities and potential confounders)

Maternal age < 25 2,009 41.8% 159 40.6% 1.00 ref 1.00 ref 25-<35 2,219 46.2% 203 51.8% 1.16 0.93 1.44 1.12 0.90 1.40 >35 578 12.0% 30 7.7% 0.66 0.44 0.98 0.60 0.40 0.92 Country of birth Australia 4,452 92.6% 379 96.7% 1.00 ref 1.00 ref Overseas 354 7.4% 13 3.3% 0.43 0.25 0.76 0.35 0.19 0.65 Living with a partner Yes 2,129 44.3% 170 43.4% 1.00 ref 1.00 ref No 2,658 55.3% 222 56.6% 0.96 0.78 1.18 0.96 0.78 1.20 Remoteness of residence Major cities 2,490 51.8% 221 56.4% 1.00 ref 1.00 ref Regional & remote 2,290 47.6% 170 43.4% 0.84 0.68 1.03 0.87 0.70 1.09 SES of residence Disadvantaged 2,290 47.6% 160 40.8% 1.00 ref 1.00 ref Advantaged 1,341 27.9% 127 32.4% 1.30 1.00 1.68 1.24 0.95 1.63 Average 1,149 23.9% 104 26.5% 1.36 1.06 1.73 1.27 0.99 1.62 Indigenous Status No 3,586 74.6% 313 79.8% 1.00 ref 1.00 ref Yes 1,219 25.4% 79 20.2% 0.74 0.58 0.96 0.73 0.56 0.95 243

Characteristics No Varenicline OR 95% CI aOR 95% CI pharmacotherapy

Quantity smoked after 20 weeks of pregnancy < 10 sticks 3,579 74.5% 277 70.7% 1.00 ref 1.00 ref >= 10 sticks 1,145 23.8% 108 27.6% 1.22 0.97 1.54 1.23 0.97 1.56 Maternal morbidity** Hypertension (including gestational) 193 4.0% 22 5.6% 1.42 0.90 2.24 1.42 0.90 2.25 Diabetes (including gestational) 163 3.4% 16 4.1% 1.21 0.72 2.05 1.13 0.64 1.98 GORD 231 4.8% 26 6.6% 1.41 0.93 2.14 1.37 0.88 2.12 Mood 1,027 21.4% 103 26.3% 1.31 1.04 1.66 1.32 1.03 1.69 Anxiety 315 6.6% 26 6.6% 1.01 0.67 1.53 0.99 0.64 1.54 Psychosis 234 4.9% 16 4.1% 0.83 0.50 1.40 0.78 0.46 1.35 Drugs and alcohol 333 6.9% 19 4.8% 0.68 0.43 1.10 0.68 0.41 1.12 Respiratory 754 15.7% 67 17.1% 1.11 0.84 1.46 1.02 0.76 1.37 Use of NSAIDS 353 7.3% 42 10.7% 1.81 1.32 2.49 1.48 1.04 2.10 Epilepsy 88 1.8% 5 1.3% 0.69 0.28 1.72 0.69 0.28 1.73 Use of steroids 195 4.1% 14 3.6% 0.88 0.50 1.52 0.83 0.47 1.46 Blood coagulation 109 2.3% 6 1.5% 0.67 0.29 1.53 0.72 0.31 1.67 Neonatal adverse events, composite** 1,649 34.3% 130 33.2% 0.95 0.76 1.18 0.97 0.78 1.22

244

Appendix 4G Association between maternal adverse events in relation to NRT patches use in the 12 months postpartum, 2011-2012, adjusted for maternal morbidities and potential confounders Characteristics No pharmacotherapy NRT OR 95% CI aOR 95% CI

Total 4,806 % 173 % Model 2 (adjusted for maternal morbidities and potential confounders) Maternal age < 25 2,009 41.8% 52 30.1% 1.00 ref 1.00 ref 25-<35 2,219 46.2% 79 45.7% 1.38 0.96 1.96 1.28 0.89 1.84 >35 578 12.0% 42 24.3% 2.81 1.85 4.26 2.54 1.63 3.95 Country of birth Australia 4,452 92.6% 162 93.6% 1.00 ref 1.00 ref Overseas 354 7.4% 11 6.4% 0.85 0.46 1.59 0.73 0.38 1.40 Living with a partner Yes 2,129 44.3% 76 43.9% 1.00 ref 1.00 ref No 2,658 55.3% 97 56.1% 0.98 0.72 1.33 0.95 0.69 1.30 Remoteness of residence Major cities 2,490 51.8% 95 54.9% 1.00 ref 1.00 ref Regional & remote 2,290 47.6% 78 45.1% 0.89 0.66 1.21 0.97 0.70 1.35 SES of residence Disadvantaged 2,290 47.6% 71 41.0% 1.00 ref 1.00 ref Advantaged 1,341 27.9% 53 30.6% 1.38 0.95 1.99 1.31 0.89 1.94 Average 1,149 23.9% 49 28.3% 1.28 0.89 1.83 1.24 0.86 1.80 Indigenous Status No 3,586 74.6% 136 78.6% 1.00 ref 1.00 ref Yes 1,219 25.4% 37 21.4% 0.80 0.55 1.16 0.89 0.61 1.31 245

Characteristics No pharmacotherapy NRT OR 95% CI aOR 95% CI

Quantity smoked after 20 weeks of pregnancy < 10 sticks 3,579 74.5% 118 68.2% 1.00 ref 1.00 ref >= 10 sticks 1,145 23.8% 54 31.2% 1.43 1.03 1.99 1.26 0.90 1.77 Maternal morbidities Hypertension (including gestational) 193 4.0% 6 3.5% 0.86 0.38 1.96 0.76 0.33 1.75 Diabetes (including gestational) 163 3.4% 6 3.5% 1.02 0.45 2.35 0.86 0.37 1.99 GORD 231 4.8% 9 5.2% 1.09 0.55 2.15 0.82 0.40 1.67 Mood 1,027 21.4% 64 37.0% 2.16 1.58 2.96 1.96 1.40 2.74 Anxiety 315 6.6% 17 9.8% 1.55 0.93 2.60 1.04 0.60 1.79 Psychosis 234 4.9% 15 8.7% 1.86 1.08 3.20 1.28 0.71 2.30 Drugs and alcohol 333 6.9% 10 5.8% 0.82 0.43 1.58 0.62 0.32 1.22 Respiratory 754 15.7% 35 20.2% 1.36 0.93 1.99 1.15 0.77 1.73 Use of NSAIDS 353 7.3% 17 9.8% 1.93 1.17 3.18 1.20 0.71 2.04 Epilepsy 88 1.8% 7 4.0% 2.26 1.03 4.96 1.89 0.83 4.29 Use of steroids 195 4.1% 12 6.9% 1.76 0.96 3.22 1.38 0.73 2.63 Blood coagulation# 109 2.3% ------Maternal adverse events, composite# 456 9.5% 45 12.4 0.78 0.44 1.38 0.73 0.40 1.33

246

Appendix 4H Association between neonatal adverse events in relation to NRT patches use in the 12 months postpartum, 2011-2012, adjusted for maternal morbidities and potential confounders Characteristics No pharmacotherapy NRT OR 95% CI aOR 95% CI

Total 4,806 % 173 % Model 3 (adjusted for maternal morbidities and potential confounders) Maternal age < 25 2,009 41.8% 52 30.1% 1.00 ref 1.00 ref 25-<35 2,219 46.2% 79 45.7% 1.38 0.96 1.96 1.28 0.89 1.86 >35 578 12.0% 42 24.3% 2.81 1.85 4.26 2.58 1.65 4.02 Country of birth Australia 4,452 92.6% 162 93.6% 1.00 ref 1.00 ref Overseas 354 7.4% 11 6.4% 0.85 0.46 1.59 0.73 0.38 1.40 Living with a partner Yes 2,129 44.3% 76 43.9% 1.00 ref 1.00 ref No 2,658 55.3% 97 56.1% 0.98 0.72 1.33 0.95 0.69 1.30 Remoteness of residence Major cities 2,490 51.8% 95 54.9% 1.00 ref 1.00 ref Regional & remote 2,290 47.6% 78 45.1% 0.89 0.66 1.21 0.97 0.69 1.34 SES of residence Disadvantaged 2,290 47.6% 71 41.0% 1.00 ref 1.00 ref Advantaged 1,341 27.9% 53 30.6% 1.38 0.95 1.99 1.32 0.89 1.95 Average 1,149 23.9% 49 28.3% 1.28 0.89 1.83 1.24 0.86 1.80 Indigenous Status No 3,586 74.6% 136 78.6% 1.00 ref 1.00 ref Yes 1,219 25.4% 37 21.4% 0.80 0.55 1.16 0.90 0.61 1.32 247

Characteristics No pharmacotherapy NRT OR 95% CI aOR 95% CI

Quantity smoked after 20 weeks of pregnancy < 10 sticks 3,579 74.5% 118 68.2% 1.00 ref 1.00 ref >= 10 sticks 1,145 23.8% 54 31.2% 1.43 1.03 1.99 1.27 0.91 1.78 Maternal morbidities Hypertension (including gestational) 193 4.0% 6 3.5% 0.86 0.38 1.96 0.75 0.32 1.74 Diabetes (including gestational) 163 3.4% 6 3.5% 1.02 0.45 2.35 0.86 0.37 2.00 GORD 231 4.8% 9 5.2% 1.09 0.55 2.15 0.82 0.40 1.66 Mood 1,027 21.4% 64 37.0% 2.16 1.58 2.96 1.95 1.39 2.74 Anxiety 315 6.6% 17 9.8% 1.55 0.93 2.60 1.04 0.60 1.79 Psychosis 234 4.9% 15 8.7% 1.86 1.08 3.20 1.29 0.71 2.31 Drugs and alcohol 333 6.9% 10 5.8% 0.82 0.43 1.58 0.64 0.32 1.26 Respiratory 754 15.7% 35 20.2% 1.36 0.93 1.99 1.15 0.77 1.73 Use of NSAIDS 353 7.3% 17 9.8% 1.93 1.17 3.18 1.20 0.71 2.03 Epilepsy 88 1.8% 7 4.0% 2.26 1.03 4.96 1.85 0.82 4.21 Use of steroids 195 4.1% 12 6.9% 1.76 0.96 3.22 1.37 0.72 2.61 Blood coagulation# 109 2.3% ------Neonatal adverse events, 1,649 34.3% 130 33.2 0.97 0.70 1.33 0.88 0.63 1.23 composite# #adjusted for maternal morbidities and potential confounders and values denoted in bold *aOR: adjusted odds ratio ** Women who did not have the morbidity were the reference category *** Percentages do not add to 100% due to missing data

248

Chapter Five Appendix Appendix 5 Comparison between women who were excluded and included in the study cohort based on missing information on the quantity of cigarettes smoked during the second half of their first pregnancy Women who were Characteristics Excluded Included (N=547) (N=3,783) n n Maternal age, years < 25 388 48.1% 2268 52.2% 25-<35 437 47.3% 1857 41.5% >= 35 70 4.6% 270 6.3% Living with a partnerb 571 63.8% 2349 53.4% Australia-born 783 87.5% 4031 91.7% b Remoteness of residence Major cities 645 72.1% 2278 51.8% Regional & remote 246 27.5% 2058 46.8% b Socio-economic status of residence disadvantaged 327 36.5% 1959 44.6% average 225 25.1% 1166 26.5% advantaged 339 37.9% 1211 27.6% Indigenous statusb 118 13.2% 974 22.2% Quantity of cigarettes smoked in the second half of pregnancy < 10 NA NA 2734 62.2% >= 10 NA NA 1661 37.8%

Parity Nulliparous 467 52.2% 1808 41.1% Multiparous 411 45.9% 2401 54.6% Grand multiparous 17 1.9% 186 4.2% Year of first delivery 2008 361 40.3% 1993 45.3% 2009 316 35.3% 1735 39.5% 2010 208 23.2% 970 22.1% 2011 10 1.1% 231 5.3% Inter-pregnancy interval, years 0.5-<1 263 29.4% 1429 32.5% 1-<2 397 44.4% 1852 42.1% 2-<3 192 21.5% 830 18.9% >=3 43 4.8% 284 6.5% Any maternal adverse events 127 14.2% 508 11.6% Any neonatal adverse events 252 28.2% 1,449 33.0% Hypertension 70 7.8% 179 4.1% Diabetes 30 3.4% 106 2.4%

249