Protocol for a Japanese nationwide repeated cross-sectional study to assess tobacco and nicotine product use behaviour after market introduction of tobacco heating products (THPs)

Introduction

Tobacco is used by more than one billion people globally [MacKay et al., 2006], and combustible tobacco is the most common form of tobacco use. Epidemiological data has concluded that smoking significantly increases the risk of diseases such as cardiovascular disease, COPD and lung cancer [US Department of Health and Human Services, 2014]. Tobacco Harm Reduction is a concept described, amongst others, by the Institute of Medicine (IOM) in their monologue Clearing the Smoke [Stratton et al., 2001], which summised that population health could be improved by replacing high-risk combustible tobacco products like cigarettes with innovative lower risk tobacco and nicotine products. This has led to an increase in alternative tobacco and nicotine products that have the potential to reduce exposure to harmful substances and reduce individual health risk. Some have termed such innovations as potentially reduced risk products, or PRRPs, and include electronic cigarettes (e-cigarettes), tobacco heating products (THPs), and oral tobacco products like Scandinavian style snus, amongst others.

The US Institute of Medicine (IOM) outlined standards for studies assessing the effectiveness of potentially reduced risk products (PRRPs) and implications of their market introduction [IOM, 2012]. Providing data to support the evaluation of behaviour/usage patterns of these new products is one of the important factors in assessing population health risk reduction. For this purpose, not only current tobacco users, but also non-users (never users and former users) must be taken into account. This inclusive approach is pursued to ensure that the population as a whole can benefit from the introduction of PRRPs into a market (not just current combustible tobacco users).

Consumer comprehension of risk perception associated with PRRP use is also crucial for acceptance and continued use, and depends on consumer exposure to product information and peer-to-peer communication [Pepper et al., 2015; Roditis & Halpern-Felsher, 2015]. In addition to risk perception, acceptance and continued use of new products depend on perceived benefits for users. Studies about quality of life (QoL) assessing smoking cessation suggested wellbeing improvements [Piper et al., 2012]. Therefore, it is reasonable for survey based studies to explore the impact of switching to PRRPs on QoL and short-term health effects.

Several groups, tobacco industry [Murphy et al., 2017; Smith et al., 2016] and independent [Berman et al., 2015; Farsalinos & Polosa, 2014], have published scientific assessment frameworks which outline the studies that should be conducted and comparator products that should be used to assess the emissions, exposure to harmful and potential harmful constituents (HPHCs) and risks of using PRRPs relative to combustible tobacco smoking. Each of these frameworks describe the requirement of assessing the risk profile of the PRRP at a population level.

British American Tobacco’s (BAT’s) approach to population risk profiling for PRRPs is informed by leading national and international public health agencies [CDC, 2012; FDA, 2012; WHO, 2012]. Their recommendations on population based studies allow for the evaluation of the effect of introducing new products on consumer perception, behaviour and health [FDA, 2012]. The overarching objectives of BAT’s surveillance programme are to:

Tobacco heating products (THPs) are just one sub-category of PRRP and contain tobacco that is heated (in a device, or with hot air passing through tobacco) but in contrast to cigarettes is not lit therefore cannot burn, thus there is no combustion [Eaton et al., 2018; Simonavicius et al., 2018; Smith et al., 2016]. The lower temperatures of the THPs heating profile (a maximum of 240°C), compared to combustion temperatures within the lit end of a cigarette (greater than 950°C), results in reduced toxicant emissions [Forster et al., 2018a; Mallock et al., 2018; Schaller et al., 2016] which in turn, results in lower biomarker of exposure (BoE) to HPHCs in humans compared to conventional combustible cigarettes [Gale et al., 2017; Schaller et al., 2016].

The aforementioned studies have demonstrated that THPs have potential to reduce harm to the individual user, however, the intrinsic risk of the product should also be contextualised within use behaviour patterns after market introduction to enable population risk assessment. Population use patterns would further be linked to consumer risk perception – such as how do non-users, cigarette smokers, solus THP users and dualist (THP and cigarette) users view the risk of THPs – and ultimately how a consumer switch or gradual transition away from cigarette consumption to solus THP use may lead to significant improvements in smoking related health effects and self-percieved quality of life.

Japan is a major market for THPs, in a nation with high smoking prevalence, in particular among men [Levin, 2013]. Of males 15 years and older in Japan in 2015, smoking prevalence (using tobacco daily) was 33.7%; by contrast, prevalence in the UK and US was 19.9% and 19.5% respectively (Tobacco Atlas, 2018). In addition to the individual harm reduction potential already described [Forster et al., 2018a; Gale et al., 2017; Haswell et al., 2018; Jaunky et al., 2018; Mallock et al., 2018; Schaller et al., 2016; Smith et al., 2016], THPs are likely to resonate very well in Japan given the strong cultural values of order, cleanliness, quality and respect for others [Hair et al., 2018], and these could be potential driving factors for Japanese consumers switching from combustible cigarettes to THPs. Thus reduced residual odour on consumers’ clothes and hair [Forster et al., 2018b], reduced teeth staining [Dalrymple et al., 2018], reduced side stream smoke, no ash and improved indoor air quality versus combustible cigarettes (Ichitsubo & Kotaki 2018; Forster et al., 2018b; Mitova et al., 2016; Mottier et al., 2016) may be considered strong additional motivations for some Japanese consumers switching from cigarettes to THPs.

In Japan, several different tobacco heating products are currently available on the market [Tabuchi et al., 2018]. iQOS from Philip Morris [Smith et al., 2016] was launched countrywide in 2014 after having been tested in selected regions. BAT’s THP glo [Eaton et al., 2018] entered the market at the end of 2016 with a city-launch in Sendai. ploom TECH from Japan Tobacco was introduced in central Tokyo in June 2017 [Reuters, 2018]. All three THPs are referred to in the study instrument (Appendix 1). More recently, Korean Tobacco and Ginseng (KT&G) released their THP Lil in Seoul in November 2017 [Tobacco Reporter, 2017]. Lil was not included in the survey at the time of the pilot study due to its nascent release in South Korea, but may still be accessible to consumers in nighbouring Japan. Additionally there are numerous and diverse THPs emerging from other markets (such as China). Despite this rapidly evolving and innovative product category, this study will allow us to explore the effect of the availability of THPs on the Japanese population as well as to investigate use patterns within this specific consumer product category.

The following sections of this manuscript will describe the study details, from instrument design and participant selection, through to ethics approval.

Study aim and design

The aim of this study is to describe the usage patterns of tobacco and nicotine containing products in Japan. This includes the estimation of prevalence rates for overall tobacco use, and by product type as well as usage pattern (solus product use, for example someone who only smokes cigarettes or only uses THPs; dual product use, someone who smokes cigarettes and uses THPs; and multi product use, someone who uses two or more different tobacco and/or nicotine products). In addition, information about frequency and amount of product use is collected as well as intention to quit for cigarette smoking and different THPs. The effect of the market introduction of THPs on the population as a whole will be evaluated for: initiation rates among tobacco non-users and former users; consumer switching rates from cigarettes to THPs, and back again to cigarettes (switch back); initiation of dual use; switching from solus THPs to dual use or solus cigarette smoking (gateway effect) and re-initiation rate of tobacco use among former tobacco users. Risk perception of different tobacco products and health-related QoL are also assessed, as well as smoking related short-term health effects among different tobacco user groups.

The tobacco use behaviour of the Japanese population will be characterised by the collection of data on a regular basis over a defined time period. This allows the assessment of usage behaviour at specific time points (point prevalence) and also to explore changes on an individual level within the last 12 months. A repeated, nationwide, population-representative cross-sectional survey is chosen to fulfil the study requirements. A pilot study (in Sendai, Tokyo and Osaka only) was implemented to test the envisaged approach in the first year (2018). The results of the pilot study are used to optimise the conduct of the first nationally representative wave commencing in 2019.

Participant selection

A geographically stratified three-stage probability sampling will be applied. The sampling universe comprises all persons aged 20 years and older (which is the legal age to consume tobacco in Japan) living in private households in Japan. Additional inclusion criteria are that the participant is able to speak and read Japanese, and be willing to participate after selection and the information about the study was provided. Pregnant and breast-feeding participants are eligible to participate, as are employees of legal, journalistic, media and tobacco industries; there is no recruitment bias. The only exclusion criterion additional to the legal age of smoking are any persons belonging to the institutionalised population (currently living in prisons, military bases, mental facilities, homes for the aged).

After the pilot study (in Sendai (Miyagi Prefecture, Tohoku region), Tokyo (Kanto region) and Osaka (Kinki region)) for subsequent study waves the nation will be stratified based on all 47 prefectures of Japan across nine regions (Chugoku, Hokkaido, Hokuriku, Kanto, Kinki, Kyushu, Shikoku, Tokai, Tohoku). In each of these nine regions, municipalities are stratified into four urbanisation degrees, based on the population of each municipality from the Basic Resident Registration population data: these are major cities (Government-designated metropolitan areas), large cities (>150,000 population, excluding major cities), medium cities (<150,000 population), and small towns.

First stage selection is based on the primary sampling units (PSU) which is a street block, selected by stratified random sampling. From the Basic Resident Registration population data (published by the Japanese Ministry of Internal Affairs and Communications (MIC)) within each stratum, 500 PSUs are selected in total proportionately to the population density. PSUs within each stratum are listed in ascending order of the municipality codes, and are selected by a randomly chosen starting number (no greater than the skip interval which is calculated as target population in each stratum divided by number of PSUs in the respective stratum). Second stage selection occurs within each PSU, all of the households are listed in ascending order of the street numbers, using the residential map database (ZENRIN Co Ltd. (ZENRIN)). A household starting number is randomly chosen to select the first household. In total, 50 households per PSU are chosen using regular numeric intervals and are listed on the household list for interviewers. Third stage selection occurs within the selected household and the individual respondent is selected by next birthday method (Salmon & Spicer Nichols, 1983). Only one individual is selected from each household.

The 20–24 year old population are identified in our sampling universe as a vulnerable group (more susceptible group) as this is most likely where tobacco product initiation takes place; additionally regulators are concerned that PRRPs may appear more attractive and appealing to this age group [FDA, 2012]. As THPs could be considered ‘high tech’ and aspirational products, youth and young adults could be more interested in such product positioning, and therefore this could raise some concern about youth appeal (Hair et al., 2018). Additionally, based on fieldwork experiences in the past, this age group is underrepresented in the applied sampling method, thus in our study the 20–24 year old population are oversampled to ensure robust estimates for this category. The oversampling quota will be assigned to each PSU. Once the target sample size is reached the interviewer will randomly select the household using the same method as for the main sample and uses the quota sampling method for selecting individual respondents aged 20–24 years. Questionnaires from oversampling are specifically marked (by participant ID only, thus participants are unaware of this) to differentiate them from the main sample as different weighting factors will be applied based on selection probability before merging them to one final sample.

Study measures

Study sample size calculation

This study is descriptive in nature thus no statistical hypotheses are tested. Proportions and rates will be estimated based on the overall population as well as for subgroups of interest like users of tobacco or nicotine-containing products, smokers and users of THPs. In Japan the smoking rate at the time of this study was 19.8%; stratified by gender the proportion of men smoking cigarettes was 31.1% and the proportion of women was 9.5% [Ministry of Health, Labour and Welfare (MHLW) 2016].

To allow for subgroup analyses and further stratification, a sufficient number of observations per cell is needed [Kish, 1995]. Thus, taking into account smoking prevalence and expecting a male to female ratio of approximately 3:1, a sample size of N=5,000 is chosen.

The 95% confidence intervals (CIs) are estimated using a conservative exact method proposed by Clopper and Pearson [Clopper & Pearson, 1934]. We accounted for variance inflation due to clustering in k PSUs and m subjects per sampling unit in terms of effective sample size (ESS). The ESS bases on the intra-class correlation coefficient ρ as a relative measure of homogeneity of sampling units and is calculated as ESS = (m*k)/(1+ ρ *(m-1)) [Killip et al., 2004].

With N=5,000 respondents randomly sampled, i.e. m=10 subjects per sampling unit, k=500 PSUs and ρ=0.02 (intra-class correlation coefficient) [Adams et al., 2004] the resulting ESS is n=4,237. With an anticipated proportion p=0.20 (20%) of tobacco product users in Japan the absolute precision will be estimated as e=±0.012 on a 95%-confidence level.

For the pilot study, as only three areas in Japan were selected, with 200 PSUs, a reduced sample size of N=4,000 was considered sufficient. This pilot enables assessment of the envisaged study procedures and to decide if the collected data answer the study objectives during following nationwide waves. An additional sub-sample of 150 subjects for the pilot will be included for the age group 20–24 years as this group could be paramount to understand nicotine and tobacco product use initiation (as previously discussed).

Data collection process

After identification of the target household by sampling the interviewer will visit the household and deliver a postcard containing an explanation of the purpose of the study, a request to participate and contact information (postcard is proprietary to the study fieldwork provider NRC). The interviewer will visit the household on a different day to make a first personal contact. Once the target person in the household has been identified and has agreed to meet/talk to the interviewer, the interviewer will explain the purpose of the study, hand out the Participant Information Letter (proprietary to NRC), and answer questions if necessary. If the target person agrees to participate in the study, the interviewer will provide the questionnaire and request completion. In cases where the participant is able to fill in the questionnaire immediately the interviewer will wait with the participant until the questionnaire is completed. In cases where the participant is not able to fill in the questionnaire immediately, or another household member is met, the interviewer will come back at a later time to collect the completed questionnaire.

If no member of the household is available the interviewer will put the Participant Information Letter describing the date of visit and the possible date of the next visit in the mail box, and will come back for the maximum of three additional times on different days at different times. If the target person in the household cannot be contacted and the questionnaire cannot be filled in by them, the interviewer proceeds to the next household on their list.

The interviewer will check the returned data for completeness/correctness and in case discrepancies are detected the interviewer will ask the participant to clarify these issues and/or complete missing parts of the questionnaire. Completed questionnaires are returned in person or per secured postal mail to the fieldwork provider’s facility for data entry. For the completion of the questionnaire all study participants (whether they are a tobacco/nicotine consumer or not) will receive a small incentive in the form of a cash voucher (¥1,000 = ^~$9 USD).

Data management

Question routing (skip logic) is applied to guide the participant through the questionnaire. Data of collected paper questionnaires are entered manually by two independent persons into a proprietary electronic data capture (EDC) system (proprietary to Kantar Health GmbH) which is validated according to FDA 21 Code of Federal Regulations Part 11 (FDA 21). All data received will be mapped into a pre-defined format of Clinical Data Interchange Standards Consortium Study Data Tabulation Model (CDISC SDTM). The SDTM will then be transformed into Analysis Data Model (ADaM) data for analysis purposes.

Data analysis

All analyses are exploratory and descriptive. Categorical variables will be analysed by frequency tables (total number of observations, number of missing values as additional categories, absolute and relative frequencies). Continuous variables will be reported as summary statistics (total n, number of non-missing and missing values, mean, standard deviation, median, minimum, maximum, and quartiles). For study objectives, 95% confidence intervals will be displayed for the mean of continuous data or frequencies, respectively.

Weighting will be applied to adjust for probabilities of selection of a respondent and non-response with adjustment according to known characteristics of the population. Data will be presented for unweighted and weighted data.

Ethical study conduct and participant confidentiality

This study is an epidemiological population survey and is conducted in accordance with the most current versions of the Declaration of Helsinki [World Medical Association, 2013], the guidelines for Good Epidemiological Practice [Public Policy Committee ISoP, 2016] and local laws and regulations. The study is conducted in accordance with the most current version of Marketing Research Guideline and Personal Information Protection Guideline that JMRA (Japan Marketing Research Association) have established. Those guidelines also comply with ICC/ESOMAR codes of conduct.

At the beginning of April 2018, an independent ethics committee (IEC) in Japan consisting of 9 members (clinical and legal) were informed about the intended data collection and all required documents were provided, including protocol, participant information letter and questionnaire. The IEC from Kitamachi Clinic, Tokyo, reviewed the study documents and approved study conduct on 17^th April 2018 before fieldwork commenced (study reference: Repeated Cross-Sectional Study to Assess Tobacco and Nicotine Usage Patterns and Behaviour After Market Introduction of glo in Japan).

During fieldwork, potential participants will be informed about the study purpose, their requested tasks, time of involvement, data confidentiality and data protection, incentive (¥1,000 voucher) and contact details in case of questions. By completing the paper questionnaire, the participant implicitly gives consent to participate in this study.

All participating research parties will ensure compliance with national and international data protection regulations. Every effort will be made to protect participant confidentiality according to Directive 95/46/EC and the Japanese Act on Protection of Personal Information (APPI).

Adverse events, product complaints and/or misuse

To comply with BAT’s product surveillance obligations, there was study monitoring for adverse effects, product complaints, and product misuse for the THP glo (this monitoring occurs for all BAT PRRPs in each market they are sold). The study questionnaire does not actively ask for any product related adverse events (AEs) or product complaints which occurred by using glo. But in case study participants report anything to the interviewer or the interviewer detects any AEs participants experienced during or after the use of glo, this information will be transferred to BAT’s local and central Quality Performance Teams. AEs, complaints or misuse regarding other THPs or cigarettes are out of scope and will not be recorded. In case of glo product complaints the interviewer will advise the participant to call the designated Customer Care Hotline. Frequency and type of product misuse is assessed for glo users in terms of if they ever tried to light a tobacco consumable and smoke it like a conventional cigarette, ever tried to use the device with another different consumable, and/or ever tired to reuse once more an already heated tobacco consumable.

Limitations of the research methods

It could be assumed that in studies on self-reported smoking behaviour that respondents might hesitate to reveal information about their tobacco consumption in the context of a study, and instead provide answers according to social desirability. In order to address this issue, the questionnaire is designed for self-administration to minimise the effect of providing socially desirable responses. Using a non-computer-assisted data collection method in combination with a self-administered questionnaire bears the risk of incomplete and inconsistent data. This will be mitigated by giving clear instructions on how to complete the questionnaire, and by formatting the questionnaire in an intuitive way with clear routing. In addition, finished questionnaires are checked by the interviewer for completeness and consistency of the provided answers during collection at the household.

The results of this study will be representative for the Japanese population taking into account the local cultural and regulatory environment. Thus, transferability of study outcomes to other countries might be limited.

Dissemination of information

Results from this study will be shared in publication and at conference(s).

Study status

The pilot study was fully executed as of April 2019 and the pilot data are currently being drafted into publication. Fieldwork for the first nationwide wave also finished in April 2019; after completion of the data life cycle and reporting the nationwide results will be published.

Conclusion

Population studies, including post-market surveillance and cross sectional surveys such as the study protocol described herein, support the evaluation of the introduction of PRRPs at a population level, in terms of consumer perceptions of product risks, behaviours and impact on public health. Analytical assessments of toxicant emissions (in vapour), pre-clinical assessments and clinical study data have all provided evidence that suggests THPs have the potential to be reduced risk alternatives compared to continued smoking. The post-market assessment of the product gives a more complete understanding of population/public health impact (uptake, switching, quitting, misuse, risk perception, quality of life, etc). Monitoring these behaviours is a regulatory requirement in certain markets, and prevalence data and observations from this study will contribute to new understanding of THP use in Japan, as well as support the areas of epidemiological science for assessing long-term usage of THPs.

Declarations

Data availability