Background

Cigarette consumption increased dramatically in South Korea in the latter half of the 20th century; by 1992, an estimated 75% of adult males smoked, the highest prevalence of smoking in the world at that time¹. Prevalence declined only slightly by 2000, when 68% of adult males in South Korea smoked. By 2017, prevalence had declined substantially to 38.1%^2,3. Despite strong social stigma against female smoking in South Korea, smoking prevalence among young Korean females (aged 17–19 years) increased within a decade from 1.6% in 1988 to 13% in 1998⁴. Tobacco market liberalization in 1988 was associated with intense competition between transnational and domestic tobacco companies witnessed and a substantial increase in smoking prevalence in the Korean population⁵. In addition, the 1988 Tobacco Business Act (TBA), under the jurisdiction of the Ministry of Finance (MOF), treated tobacco as a means to contribute to the national economy⁶. In 1995, the National Health Promotion Act (NHPA) was enacted to minimize the harmful public health effects of tobacco and reduce its use. In particular, the NHPA protects young adolescents aged ≤19 years from smoking and regulates tobacco companies’ advertising and promotion, which targeted youth, young adults, and females⁵. More importantly, NHPA imposed National Health Promotion taxes on cigarettes at KRW 2 (US$0.0016) per pack in 1997, KRW 150 (US$0.13) in 2002, KRW 354 (US$0.30) in 2005, and KRW 814 (US$0.69) in 2015⁷. Similar rates were imposed on e-cigarettes (ECs) in 2011 and other tobacco products in 2014 to create the National Health Promotion Fund⁸.

In 2005, the Korean government ratified the WHO FCTC and has since strengthened its national tobacco control policies. In 2013, the Korean Ministry of Health and Welfare (MOHW) asked tobacco control experts to identify priorities for national tobacco control policies; that the expert group identified tax increases, expansion of smoke-free places, a ban on tobacco advertising and promotion, and introduction of pictorial health warnings as key priorities to save lives from diseases caused by tobacco use in South Korea^9,10. Based on these priorities, tobacco control policies continued to improve. In 2015, the National Health Plan 2020 (HP2020) was established to reduce smoking prevalence among adult males to 29% and among adult females to 6% by 2020 where key policies, such as increasing retail price for a pack of cigarettes from KRW 2500 (US$2.09) to KRW 4500 (US$3.77) (Article 6), banning misleading descriptors such as ‘mild’ or ‘light’ on all tobacco packages (Article 11), banning smoking in all restaurants and bars (Article 8), banning tobacco advertising in retail shops (Article 13), and smoking cessation programs and reimbursement of smoking cessation treatment costs by the national health insurance program (Article 14), and on 23 December 2017, graphic warning labels (Article 11) were implemented^11,12. In 2019, MOHW developed its 5-year plan that includes regulating novel tobacco and nicotine products and reducing tobacco use among youth. For the period of 2021–2025, a roadmap was announced in 2019 to gradually expand smoke-free zones by banning smoking in all indoor buildings and eliminating all indoor smoking rooms by 2025¹³.

Despite its history of tobacco control, South Korea continues to have one of the highest smoking prevalence rates among high-income countries. The Korea National Health and Nutrition Examination Survey reported that cigarette smoking prevalence among adult males aged ≥19 years decreased from 47.8% in 2008 to 35.7% in 2019. Among adult females, however, prevalence decreased only slightly, from 7.4% in 2008 to 6.7% in 2019^3,10. The recent introduction of novel nicotine products such as nicotine vaping products (more commonly known as e-cigarettes or ECs) and heated tobacco products (HTPs) has pushed the government to implement stronger and more restrictive tobacco control policies for all nicotine and tobacco products^14-16. Future amendments to the TBA may require tobacco manufacturers and importers to list harmful ingredients contained in vaping and tobacco products, banning flavors in vaping and tobacco products, banning nicotine extracts from tobacco stems and roots, and regulating synthetic nicotine. Future amendments to the NHPA may include banning direct and indirect advertising and promotion, banning the display of tobacco products in retail stores, banning the use of cartoon and animal characters on tobacco products, and displaying smoking cessation advertisements alongside tobacco point-of-sale advertising.

In 2017, tobacco companies introduced heated tobacco products (HTPs) in South Korea and Japan as their test markets, including IQOS by Philip Morris International, glo by British American Tobacco, and lil by Korean Tobacco & Ginseng (KT&G)^4,6. In 2018, the prevalence of current HTP use (use in the last 30 days) among Korean adults was 4.4% overall, 7.8% among males and 0.9% among females. Approximately 90% of current HTPs users also used cigarettes or e-cigarettes, while some used all three products concurrently^14-16. With the introduction and increased prevalence of these alternative nicotine products in the South Korean market, the government of South Korea began to harmonize tobacco control policies so that they applied to HTPs and ECs in addition to combustible cigarettes. For example, HTPs are banned in public places and pictorial health warnings have been added to HTP tobacco packs^4,6,14-16.

In summary, the history of tobacco control in South Korea can be divided into four distinct periods: 1) before the enactment of Tobacco Act Business Act in 1988, a period where most people were not aware of the harms of tobacco use and secondhand smoke; 2) from 1988, the beginning of tobacco control movement with the birth of Korean Association on Smoking or Health (KASH), to 1995 with the enactment of the National Health Promotion Act (NHPA); 3) from 1995, the setting of the stage with NHPA to promote health, to 2015 with the implementation of higher tobacco taxes and expansion smoke-free places; and 4) since 2015, the creation of a roadmap towards a comprehensive tobacco control policy^5-9,17.

International Tobacco Control (ITC) Korea Project

The International Tobacco Control Policy Evaluation (ITC) Project is a global project with 31 countries including South Korea. The ITC Project’s central objective is to evaluate the impact of World Health Organization Framework Convention on Tobacco Control (WHO FCTC) policies. Similar to 30 other ITC countries, the ITC Korea Project uses the same three strategies to rigorously evaluate the effects of tobacco control policies: 1) a quasi-experimental research design (i.e. ‘natural experiments’)¹⁸, in which one group exposed to a policy is compared to another, unexposed group; 2) the use of longitudinal cohort designs¹⁹ in which individuals are measured on the same key outcome variables over time^20,21; and 3) the measurement of appropriate policy-specific variables that are conceptually close to the policy being evaluated and less likely to be affected by other factors. These strategies, including other explanatory variables (covariates), are unparalleled in the study of population-level interventions and produce a research design with the potential to make strong inferences about policy impact^{12,18,19,22-26}. The ITC Korea Surveys serve as an evaluation system for measuring the impact of WHO FCTC implemented in South Korea.

The ITC Korea Project was first established in 2005, the same year South Korea ratified the WHO FCTC. The first ITC Korea Survey was a longitudinal cohort study consisting of three waves conducted in 2005, 2008, and 2010 among 1002, 1818, and 1753 adult smokers aged ≥19 years, respectively^19,27-30. The second ITC Korea Survey consisted of a single wave conducted in 2016 among 2000 adult smokers³⁰. The data from these two cohorts were collected through telephone interviews by commercial survey firms, Gallup Korea and Hankook Research Co. Ltd, respectively. These two cohort studies were a collaboration between researchers from the Korea National Cancer Center and the ITC Project at the University of Waterloo, Canada.

The third ITC cohort survey, the 2020 ITC Korea (KRA1) Survey, was conducted from 18–28 June 2020 during the COVID-19 pandemic. The first wave of this third cohort survey sampled 4794 adult tobacco and/or nicotine users and non-users to examine patterns of use, reasons for use, risk perceptions, and the effects of current tobacco control policies meant to inform implementation of future tobacco control policies. The second wave (KRA2) fieldwork was conducted from November– December 2021. This third cohort survey is a collaboration between researchers from the Korean Health Promotion Institute, Korea National Cancer Center, Seoul National University, and other leading researchers in Korea, and the ITC Project at the University of Waterloo, Canada³¹.

Objectives of the 2020 ITC Korea Survey

The third cohort of ITC Korea Survey has three main objectives. First, it will assess the effectiveness of existing FCTC policies in Korea, where tobacco control measures were recently strengthened (e.g. large graphic warnings, stronger smoke-free laws, higher taxes), and compare the effectiveness of these policies between (a) Korea and Japan and (b) across the 20+ ITC countries. Second, it will conduct evaluation studies of tobacco control policies implemented at future dates (e.g. stronger smoke-free laws, revision of graphic warnings, tax increases, changes in regulations on ECs and HTPs). Third, it will measure and assess patterns of HTP use among demographic subgroups and assess cigarette smoking history and nicotine dependence. It will also examine correlates of interest in, trial of, and regular use of HTPs such as reasons for use. Furthermore, it will examine characteristics of and precursors relating to smoking cessation, the relationship between HTP use and dual use and cigarette smoking cessation or relapse to cigarette smoking over time³¹.

Sampling design and method of recruitment

The KRA1 Survey used Rakuten Insight’s existing proprietary online panel in the Republic of Korea (see https://insight.rakuten.com) for its sampling frame. Recruitment for the panel is conducted daily, tapping into affiliated online resources and referrals to maintain a panel as consistent as possible with the general population. Panelists are pre-profiled using a series of questions which can then be used as pre-targeting variables (e.g. smoking, HTP usage, EC usage) for sampling quotas. Panelists received email invitations and had the option of logging into their proprietary panel site to access the KRA1 Survey.

For cigarette-only smokers and non-users, sex-specific age quotas within geographical regions were used for sampling purposes. Six age groups were defined for these quotas (19–29, 30–39, 40–49, 50–59, 60–69, ≥70 years). Data from the 2018 Korea National Health & Nutrition Examination Survey (KNHANES)ⁱ were used to obtain smoking prevalence estimates by region and sex-specific age groups. These estimates were combined with population estimates to obtain the various quotas for cigarette-only smokers and non-users. Unfortunately, the KNHANES did not permit estimation of the proportion of individuals who were exclusive cigarette smokers. Hence, the quotas used cigarette smoking as a proxy for exclusive cigarette smoking. This implies that the quotas for exclusive cigarette smokers were not completely accurate; the resulting small biases, however, were corrected during the construction of sampling weights. This process used data from the 2019 Korean Community Health Survey^31,32,ii which allowed for proper estimation of the proportion of individuals who exclusively smoked cigarettes. Quotas for other user groups were not applied because the desired sample size was too small to use the quotas efficiently or because of insufficient prevalence data³¹.

Invitation emails were sent strategically to Rakuten’s panelists identified as probable tobacco users to fill the five user group subsample quotas. Tobacco user quotas (cigarette-only smokers, HTP-only users, EC-only users, cigarette-HTP dual users, cigarette-EC dual users) were filled first before the non-user quotas. Respondents from the group of probable tobacco users who were non-users were to be accepted as part of the non-user quota. After the five user group subsamples were filled, invitations strategically targeted non-users to fill the remaining open positions in the non-user quotas.

Sample size

The survey was administered as an online web survey to 4794 adults aged ≥19 years (4234 tobacco and/or nicotine users and 560 non-users). The median survey interview length was 36 minutes³¹. Original target quotas were: 1) 2000 cigarette-only smokers; 2) 500 HTP-only users (also including HTP-EC dual users); 3) 400 EC-only users; 4) 800 cigarette-HTP dual users; and 5) 500 cigarette-EC dual users (also including triple users) and 500 non-users (including never users and former users of any of the products)³¹. For cigarette-only smokers and non-users, additional quotas based on region, sex, and age groups were applied to obtain final sample sizes proportional to stratum sizes based on smoking prevalence estimates in combination with Korean census data^31,32. During data collection, the target sample sizes for HTP-only users, and EC-only users were more difficult to fill than for other user groups. Therefore, adjustments were made to reduce the target sample sizes for these two groups and to increase the size of the dual user group. The sample sizes of EC-only users and HTP-only users were reduced to 150 each, the sample size of dual users of cigarettes and ECs was increased to 750, and the sample size of dual users of cigarettes and HTPs was increased to 1150. In addition, for the cigarette smoker and non-user groups, the oldest three age groups (50–59, 60–69, and ≥70 years) were combined into a single group as ≥50 years. More details about the target adjustment are provided in a technical report³¹. The final achieved sample size of 4794 respondents is presented in Table 1.

Demographic characteristics of respondents

Demographic characteristics of the ITC KRA1 sample are presented in Supplementary file Tables S1 and S2, by tobacco product used (cigarettes, ECs, and HTPs) and frequency of use (daily, non-daily, non-users in Table S1), and combination of products used (exclusive use, dual use, or poly-use of cigarettes, ECs, and HTPs in Table S2). Unweighted percentages show the distribution of respondents within each product user group while the weighted estimates represent the population distribution within each product user group. Geographical region, sex, age, and education were used directly in the raking algorithm to compute the sampling weights; as a result, differences exist between unweighted and weighted estimates for some geographical regions and for some sex, age, and education categories.

Survey measures

All ITC surveys were developed using the same conceptual framework¹⁸ and methods¹⁹. They include validated measures of smoking behavior, psychosocial measures, and policy-relevant measures in all major FCTC policy domains including health warnings, marketing restrictions, tax and price, smoke-free laws, and cessation. Survey measures are designed to be identical or functionally equivalent across all ITC countries, enabling comparisons over time and across different countries. All ITC surveys are publicly accessible on the ITC Project website (https://itcproject.org/surveys/). Table 2 lists the key measures included in the ITC KRA1 Survey.

Survey content was initially developed in English in collaboration between Korean and Canadian research team members. The final English survey was then translated into Korean by a professional translator at the survey firm. The Korean translation was checked and verified by Korean researchers to meet the standards for the highest possible degree of clarity and accuracy and have the closest equivalence to the English survey content. The full surveys in all three languages for ITC KRA1 Survey can be found at https://itcproject.org/surveys/republic-korea/kra1-cohort3/.

The survey protocols and all materials, including the survey questionnaire, received ethical approval from the Office of Research Ethics, University of Waterloo, Canada (ORE# 41512) and the Institutional Review Board of Korea Health Promotion Institute (#120160811107AN01-2004-HR-042-02). All participants provided consent to participate. A full description of the survey methods can be found elsewhere³¹.

Response and cooperation rates

Two key survey statistics were computed. The response rate was defined as the number of eligible respondents who completed the survey divided by the estimated number of eligible respondents that were selected/contacted [the American Association for Public Opinion Research (AAPOR) Response Rate 4 (RR4)]. The cooperation rate was defined as the proportion of eligible respondents (i.e. those who have completed all eligibility questions and have been found to be eligible) who completed the survey [AAPOR Cooperation Rate 4 (COOP4)]³³. For the ITC KRA1 Survey, the response rate was 15.2% and the cooperation rate was 97.4%³¹. It should be noted that the response rate includes in the denominator people who were invited but may never have seen or noticed the invitation, especially with a relatively short fieldwork period. This results in a low response rate, typical of response rates for surveys employing web panels but not indicative of a general response rate for the participating survey firm³¹. Because of this, the cooperation rate may be a better indicator of the potential sampling bias due to the content of the survey (tobacco use). The very high cooperation rate is a sign that the potential bias due to refusal on the basis of survey content was very low³³.

Data quality control

Data quality controls were applied to identify respondents who may not have taken the survey seriously.

Respondents who completed the survey in an extremely short time and/or skipped (selected ‘Refused’ or ‘Don't know’ as answers) to a certain number of questions were considered ‘speeders’. To identify potential ‘speeders’, data from all respondents were used to calculate the percentiles for the time taken to complete each question and the percentage of non-response. Two measures were used: 1) SecperQ, defined as the total time taken to complete the survey divided by the number of questions answered by the respondent, and 2) %RDK, defined as the number of questions answered with either ‘Refused’ or ‘Don't know’ responses divided by the number of questions answered by the respondent. Points were assigned based on percentiles within each of the user groups, the 10th percentile of SecperQ (‘min_value’) and the 90th percentile of %RDK (‘max_value’) were used as independent cut points¹⁸:

Respondents’ SecperQ and %RDK points were then summed (range: 0–8). If the final sum was >4, the respondent was classified as a ‘speeder’ and removed from the data³¹.

It was discovered during the data cleaning process that some respondents reported conflicting information about the types of tobacco or nicotine products they used. By cross-referencing the brand(s) of product(s) they reported, it was determined that some respondents were confused about which product(s) they were using. In Korea, HTP and EC devices are both known as ‘electronic cigarettes’ and HTP heat sticks and combustible cigarette sticks are both known as ‘cigarettes’. For example, some respondents reported that they were cigarette smokers when in fact they were IQOS (HTP) users and other respondents reported they used ECs when they actually used glo, sens, or lil HTP devices. In these cases, it was decided to adjust the user type to match the product brand(s) reported. Therefore, all the weights were recalculated based on the new user types assigned. Overall, 54 records had their user groups adjusted in the final dataset: 35 were reclassified from dual cigarette and HTP users to exclusive HTP users; 13 were reclassified from triple users to dual cigarette and HTP users; three were reclassified from triple users to dual HTP and EC users; one was reclassified from dual cigarette and EC to exclusive cigarette smoker; one was reclassified from HTP and EC user to exclusive EC user; and one was reclassified from a triple user to dual cigarette and EC user³¹.

Survey weights

Cross-sectional survey weights for different analytic purposes were constructed for the final data set. For the cross-sectional weights, respondents were divided into 12 user types: 1) cigarette-only users, 2) HTP-only users, 3) EC-only, 4) HTP-cigarette dual users, 5) EC-cigarette dual users, 6) HTP- EC dual users, 7) triple users, 8) recent quitters using HTPs but not ECs, 9) recent quitters using ECs but not HTPs, 10) recent quitters using both HTPs and ECs, 11) recent quitters using neither HTPs nor ECs, and 12) non-users including long-term quitters³¹.

Data from the 2019 Korea Community Health Survey were used to obtain calibration targets^31,ii. For user groups 1–7 and 12, a raking procedure was applied to calibrate the weights based on gender, age, geographical region, and education; this raking procedure was applied separately for each user group³¹. Some cells were collapsed because they contained too few respondents. Recent quitters (user groups 8–11) were combined in a single raking procedure and their weights were calibrated based on gender, age, geographical region, education, and HTP and/or EC use status. These weights were designed to make these 4794 respondents representative of the adult population aged ≥19 years with the exclusion of: 1) ineligible respondents (current or past cigarette smokers who smoked less than 100 cigarettes in their lifetime); 2) those who quit using HTPs, had never used ECs, and were not smoking cigarettes at least weekly or quit smoking cigarettes within the last two years; 3) those who quit using ECs, used HTPs less than weekly or not at all, and were not smoking cigarettes at least weekly or quit smoking cigarettes within the last two years; and 4) those who quit both HTPs and ECs, and were not smoking cigarettes at least weekly or quit smoking cigarettes within the last two years. These weights were also designed to make respondents in each of the 12 user groups representative (with respect to gender, age, geographical region, and education) of the corresponding population at the time of Wave 1 data collection. A full description of the survey weights can be found elsewhere³¹.