Trends in health behaviors over 20 years: findings from the 1998-2018 Korea National Health and Nutrition Examination Survey

OBJECTIVES This study aimed to examine the trends in health behaviors in Korean population using data from the Korea National Health and Nutrition Examination Survey (KNHANES). METHODS The subjects were 96,408 adults aged 19 years or older who participated in the first (1998) through seventh (2016-2018) KNHANES health interview. The prevalence of health behaviors (cigarette smoking, alcohol drinking, and physical activity) and annual percent change (APC) were estimated using SAS and the Joinpoint program. RESULTS The prevalence of current cigarette smoking in men decreased by 2.8%p (APC=-2.8, p<0.001) annually over the 20-year period, and the prevalence of exposure to secondhand smoke at home substantially decreased compared to 2005 (APC=-8.8, p<0.001). Compared to 2005, the prevalence of current alcohol drinking in women, but not men, increased (APC=2.0, p<0.001), and the prevalence of binge drinking decreased in men (APC=-0.7, p<0.001) and increased in women (APC=2.4, p<0.001). The prevalence of aerobic physical activity decreased from 2014 in both gendersd (p<0.001). The prevalence of healthy behaviors practice (non-smoking, alcohol abstinence, and aerobic physical activity) was down-trending (APC=-5.3, p<0.001), especially among women (APC=-6.4, p<0.001). CONCLUSIONS Over the past 20 years, smoking behaviors improved. However, drinking behavior was unchanged and physical activity indicators markedly decreased. More active programs are necessary for improving health behaviors, which are major risk factors linked to chronic diseases.


Health interview
The health interview was divided into face-to-face interview by trained interviewers (medical condition, healthcare utilization, injuries, etc.) and self-administered survey (cigarette smoking, alcohol drinking, etc.). Survey items were designed slightly differently for children (aged 1-11 years), adolescents (aged 12-18 years), and adults (aged ≥ 19 years). In this study, the analysis was limited to the cigarette smoking, alcohol drinking, and PA data collected in the self-administered section of the health interview (PA data has been collected by face-to-face interviews since 2014).
For cigarette smoking, information regarding lifetime history of smoking, current smoking status, and exposure to secondhand smoke at home, work, and public places (added 2013) were assessed. Additionally, in 2013, an instruction was added for survey participants to specifically consider the last 7 days only when responding to the questions on exposure to secondhand smoke. Starting in 2005, data on the frequency and volume of alcohol consumption and frequency of binge drinking using the Alcohol Use Disorder Identification Test (AUDIT) and data on PA using the short form of the International PA Questionnaires (IPAQ short form) began to be collected. The questionnaires consisted of a total of 7 items that assessed the amount of time spent on vigorous-intensity PA, moderate-intensity PA, walking, and sedentary time. In 2014, the Global PA Questionnaire (GPAQ) was intro-duced, allowing for the participation of PA to be examined in the domains of work, leisure, and transportation. The GPAQ consisted of a total of 16 items regarding the amount of time spent on moderate-intensity activity, vigorous intensity activity, and sitting in each of the three categories (work, leisure, and transportation).
Since the questionnaire was difficult to answer, it was changed to face-to-face interview instead of self-administered survey. Other than slight changes to wording, the items regarding muscle strengthening did not undergo any major revisions since they were introduced in 2005.

Definitions of health behavior indicators
Current cigarette smokers were defined as individuals with a lifetime cigarette smoking history of 5 packs or more who also currently smoked. The prevalence of exposure to secondhand smoke inside home/public places was estimated by using the number of current non-smokers as the denominator and calculating the rate of non-smokers exposed to secondhand smoke inside home/public places during the last 7 days. When calculating the prevalence of exposure to secondhand smoke in indoor workplaces, the denominator was the number of current workers classified as current non-smokers and the numerator was the number of those exposed to secondhand smoke in indoor workplaces during the last 7 days. Current alcohol drinkers were defined as individuals who consumed alcohol, on average, at least once a month, and binge drinkers as individuals who binge drank (consuming 7 and 5 glasses of alcohol in a single occasion for men and women, respectively) at least once a month on average. The prevalence of aerobic PA was estimated in accordance with the PA Guidelines for Koreans, by calculating the rate of individuals who typically performed moderate and vigorous intensity PA for a total of 150 minutes or more per week. In this calculation, work, leisure, and transportation time were combined and the amount of time spent on vigorous intensity PA was doubled. The prevalence of muscle strengthening was estimated by calculating the rate of individuals who performed muscle strengthening at least twice per week regardless of the type and duration.
Based on the definition of healthy behavior practice as the practice of engaging in aerobic PA, not smoking, and abstaining from alcohol, the prevalence of healthy behaviors could be calculated. Non-smoking was defined as a lifetime history of non-smoking or current non-smoking with a history of smoking. Alcohol abstinence was defined as not drinking or drinking alcohol less than once a month in the previous 1 year. PA was defined as practicing aerobic PA 150 minutes or more in a typical week. To calculate the prevalence of aerobic PA, the data were collected using two different questionnaires, the IPAQ from 2007-2013 and the GPAQ starting in 2014.   Values are present as number or number (%). 1 Calculated as monthly household income divided by the square root of the number of persons in the household, categorized into quantile by age and gender.  Values are presented as weighted %. 1 Age-standardized prevalence was calculated using the 2005 population projections for Korea. 2 Current cigarette smoking: percentage of adults who have smoked at least 100 cigarettes during their lifetime and who are currently smokers. 3 The annual percent change (APC) is significantly different from 0. 4 Calculated as monthly household income divided by the square root of the number of persons in the household, categorized into quantile by age and gender.
***p< 0.001. USA). To represent the Korean population, the sampling weights assigned to participants were applied to all analyses. Sampling weights were generated by considering complex sample design, non-response rate of the target population, and post-stratification. To adjust differences in results from change in age structure of each year, age-standardized prevalence was calculated using the age-and gender-specific structures of estimated population based on the 2005 population projections for Korea. Trends in cigarette smoking, alcohol drinking, PA prevalence by gender and household income level were age-standardized prevalence and were calculated using SAS (PROC SURVEYREG). The prevalence indicators by age were crude rates and were calculated using SAS (PROC SURVEY-MEANS). The estimates and their standard errors obtained from SAS were input into NCI's Joinpoint program, the model was estimated by setting the Joinpoint as 0 or 1, and the annual percent change (APC) was calculated. APC verified that the annual rate of change was "0" under the significance level of 0.05, and the Monte Carlo method in Joinpoint Regression Program was used to test the statistical significance of the optimal model.

RESULTS
The demographic characteristics of the KNHANES participants (1998-2018) are shown in Table 1. The total number of subjects aged 19 years or older was 96,408. Of those, 42,071 were men and 54,337 were women, making women a higher proportion of the sample than men. The proportion of participants aged between 19 years and 49 years decreased, while the proportion of those aged ≥ 50 years increased over time.
The prevalence of current cigarette smoking substantially decreased in men, from 66.3% in 1998 to 36.7% in 2018 (APC = -2.8, p < 0.001). Over the same time period, this prevalence, however, increased slightly in women, from 6.5% to 7.5%, but the change was not statistically significant (APC = 0.6) ( Table 2). In men, the prevalence of current cigarette smoking significantly decreased in every age group and income level (p < 0.001), but the gap between the uppermost and lowest income levels did not decrease after 2010, maintaining approximately a difference of 10%p each year (31.0% and 40.1% in 2018, respectively). In contrast, the prevalence of current cigarette smoking in women increased in the 19-29, 30-39, and 40-49 age groups (p < 0.001), while it decreased in those aged ≥ 60 years. The prevalence of current non-smokers' exposure to secondhand smoke at home, workplace, and in public places greatly diminished, with decreases of 14.5%p for home (18.5% in 2005 to 4.0% in 2018), 25.4%p for indoor workplaces (36.9% in 2008 to 11.5% in 2018), and 41.1%p for indoor public places (58.0% in 2013 to 16.9% in 2018) (p < 0.001) ( Table 3).  Values are presented as weighted %. 1 Age-standardized prevalence was calculated using the 2005 population projections for Korea. 2 The annual percent change (APC) is significantly different from 0. 3 Secondhand smoking at home: percentage of current non-smokers who have been exposed to secondhand smoke at home during the past 7 days. 4 Secondhand smoking in the workplace: percentage of current non-smokers who have been exposed to secondhand smoke in the workplace during the past 7 days. 5 Secondhand smoking in public places: percentage of current non-smokers who have been exposed to secondhand smoke in public places during the past 7 days. ***p<0.001. Age-standardized prevalence was calculated using the 2005 population projections for Korea. 2 Current alcohol drinking: percentage of adults who have had alcoholic drinks 1 or more times a month during the past year. 3 The annual percent change (APC) is significantly different from 0. 4 Calculated as monthly household income divided by the square root of the number of persons in the household, categorized into quantile by age and gender. ***p<0.001.
The prevalence of current alcohol drinking did not change in men from 2005 to 2018, whereas it increased in women, from 37.0% in 2005 to 51.2% in 2018 (APC = 2.0, p < 0.001) ( Table 4). The prevalence among women significantly increased in all age groups except for those aged ≥ 70 years (p < 0.001) and in all income levels (p < 0.001). The prevalence of binge drinking in men decreased from 55.3% in 2005 to 50.8% in 2018, and the extent of the decrease was particularly sizable in the 19-29 age group (APC= -1.4, p< 0.001) ( Table 5). In contrast, the prevalence of binge drinking in women increased from 17.2% in 2005 to 26.9% in 2018 (APC, 2.4, p < 0.001), rising in all age groups except for those aged ≥ 70 years. The increase in the 19-29 age group was particularly substantial (19.1%p), increasing from 30.7% in 2005 to 49.8% in 2018. Additionally, the prevalence increased for all income levels (p < 0.001).
The prevalence of aerobic PA decreased by approximately 25%p in both genders, from 76.3% to 51.0% in men between 2007 and 2018, respectively (APC= -4.2, p< 0.001) and from 68.9% to 44.0% in women during the same period (APC = -4.7, p < 0.001) ( Table  6). The prevalence significantly decreased, especially after 2014, in both genders in every age group and every income level (p< 0.001). The prevalence of muscle strengthening did not change from 2005 to 2018 in men, whereas it increased in women, from 11.2% in 2005 to 14.9% in 2018 (APC= 2.3, p< 0.001). The increasing trend  Age-standardized prevalence was calculated using the 2005 population projections for Korea. 2 Binge drinking: percentage of adults who have drunk ≥ 7 (men) or ≥ 5 (women) alcoholic drinks more than once a week during the past year. 3 The annual percent change (APC) is significantly different from 0. 4 Calculated as monthly household income divided by the square root of the number of persons in the household, categorized into quantile by age and gender. ***p<0.001.
was clear in the 60-69 and 70-79 age groups (data not shown) [11]. The prevalence of healthy behaviors practice dropped by approximately 10%p, from 25.1% in 2007 to 14.5% in 2018 (APC = -5.3, p < 0.001) ( Table 7). The prevalence decreased in both men and women; however, in men, it significantly decreased in those aged ≥ 50 years, while in women, the prevalence significantly decreased in all the age groups. Regarding changes in the prevalence of healthy behaviors practice, the decrease was smaller in men since cigarette smoking and alcohol drinking improved in the 19-29 and 20-29 age groups. In contrast, the prevalence among women decreased by 19.0%p compared to 2007 (APC= -6.4, p< 0.001) and by 8.8%p compared to 2014, because alcohol drinking and PA deteriorated in most age groups.

DISCUSSION
In this study, changes in major health behaviors of Korean adults were investigated using the KNHANES data. The prevalence of cigarette smoking and exposure to secondhand smoke consistently decreased, as well as the prevalence of PA. Regarding alcohol drinking, there was no considerable change in men, whereas in women, the prevalence of alcohol drinking and binge drinking both increased (Supplementary Material 1).  Age-standardized prevalence was calculated using the 2005 population projections for Korea. 2 Aerobic PA: percentage of adults who have performed 150 minutes of moderate-intensity PA or 75 minutes of vigorous-intensity PA or an equivalent combination of moderate-and vigorous-intensity PA in a typical week. 3 The annual percent change (APC) is significantly different from 0. 4 Calculated as monthly household income divided by the square root of the number of persons in the household, categorized into quantile by age and gender. ***p<0.001.
Awareness of the harm of smoking and the need for smoking cessation has increased through the smoking cessation movement spurred by private sector in the late 1980s [12]. In 1995, the Korean government also established legal foundations for tobacco regulations through the National Health Promotion Act and promoted policies (e.g., reinforcing legal and institutional foundations, a smoking cessation campaign and support services for smoking cessation, and expanding non-smoking areas) to ratify and implement the 2005 World Health Organization Framework Convention on Tobacco Control [13][14][15][16]. Owing to these efforts, the prevalence of current cigarette smoking among men between 2013 and 2014 was approximately 43%, which was more than 20%p less than that in 1998. Since 2015, policies, such as the increase in cigarette prices, smoking cessation campaigning in the mass media, the expansion of support services for smoking cessation, and the introduction of cigarette health warning on cigarette packs (2016) have been implemented [15][16][17], and in 2018, the prevalence of current cigarette smoking among men had further decreased to 36.7%. Additionally, starting in 2012, various enforcements have been implemented for the purpose of preventing harm from secondhand smoke, such as smoking prohibition in all public facilities and the expansion of indoor non-smoking are- Table 7. Trends in the prevalence 1  Age-standardized prevalence was calculated using the 2005 population projections for Korea. 2 Healthy behaviors practice: percentage of adults who met the guidelines for cigarette smoking, alcohol drinking, and physical activity. 3 The annual percent change (APC) is significantly different from 0.  [18][19][20]. Since the decreasing trend in current cigarette smoking prevalence has recently slowed, the use of tobacco products other than cigarettes has increased, and there continues to be a huge gap between the current prevalence and the HP 2020 goal; more powerful policies should be implemented with continuity. In 2018, the use of heated tobacco products and e-cigarettes increased to about 5 times that of 2013, the year when the relevant survey items were introduced to the KNHANES. Additionally, the prevalence of the dual use of conventional cigarettes and vaping products (including heated tobacco products and ecigarettes) increased up to 22.5%. These findings suggest the need to investigate the harms of new tobacco products more actively and monitor use behavior to develop preemptive policies in preparation for a potential increase in prevalence due to these new products. Unlike in men, the prevalence of current cigarette smoking in women was found to be low relative to other countries [9,10,18], but there was an increasing trend in the 19-29, 30-39, and 40-49 age groups. The increase in working women, the availability of flavored and low-nicotine cigarettes, which appeal more to women [21], and the reduction in underreporting as the social stigma of women smokers has lessened are potential factors contributing to the increase in women smokers [22]. Given that, to date, the focus of support services for smoking cessation and related policies have been on men due to their high prevalence of cigarette smoking, in-depth studies, including qualitative research on women's smoking behavior and relevant factors, are needed and effective smoking cessation programs specifically designed for women need to be developed [23].
The prevalence of current alcohol drinking and binge drinking in men were 70% and 50%, respectively, without much change over time. Although it is difficult to make accurate between-country comparisons because the definitions of the indicators are slightly different, these rates are higher compared with other countries [24]. One possible explanation why these rates were high could be that programs have primarily been designed for education, and the screening and treatment of high-risk drinking and alcohol addiction, while access to alcohol itself had not been focused [25,26]. Currently, stronger policies are being implemented for the prevention of alcohol-induced harm, such as prohibiting alcohol consumption in public places and establishing guidelines regarding scenes of drinking in the media. Thus, changes in alcohol consumption trends should continue to be followed [27][28][29]. In women, the prevalence of current alcohol drinking and binge drinking both increased significantly, with considerable increases in the 19-29, 30-39, and 40-49 age groups. Like the increase in cigarette smoking, an increase in binge drinking among women may be explained by the increase in working women, increased access due to the development and availability of different types of alcohol products that appeal to women (such as low-alcohol and fruit-flavored soju products), and the decrease in negative perceptions of women drinkers [21]. In the United States, the prevalence of high-risk drinking and alcohol use disorder has increased among women in comparison to men, and it has been suggested that, like Korea, these increases may be associated with increased opportunities for women through education and work and an increased permissive to drinking in the social atmosphere [24,30,31].
The prevalence of aerobic PA consistently showed a decreasing trend. In 2018, the prevalence was 51% in men and 44% in women. The analysis conducted to examine the decreasing trend by area (work, transportation, and leisure activity) showed a 6.6%p increase between 2014 and 2018 in those not engaging in transport-related PA (38.5 and 45.1%, respectively). However, no sizable increase was noted in either the prevalence of those not performing work-related PA (97.4% in 2014 and 98.6% in 2018) or those not performing leisure-related PA (90.2% in 2014 and 91.7% in 2018). Accordingly, the decrease in the prevalence of aerobic PA seemed to be the result of reduced PA for transportation. Since PA practice is closely linked to personal, socioeconom-ic, and environmental factors, to increase PA, accessibility should be enhanced through providing knowledge, motivating individuals, and establishing an appropriate environment [32][33][34][35]. PA indicators have not improved, even though the importance of PA has been emphasized through the PA Guidelines for Koreans since 2013. Additionally, several individual programs to promote PA have been executed as part of the integrated health promotion program in the local community. Accordingly, it seems necessary to identify the causes of the failure by conducting in-depth analyses and developing strategies to increase PA.
The prevalence of healthy behaviors practice (non-smoking, alcohol abstinence, and aerobic PA), which was 25.1% in 2007, decreased to 14.5% in 2018, which was a decrease of 10.6%p. The prevalence was greatly reduced in women. Whereas in men, the prevalence of healthy behaviors practice increased by 5.1%p (12.3% in 2014 to 17.4% in 2018) owing to improved cigarette smoking and alcohol drinking behaviors in the 19-29 age group, women's health behaviors deteriorated overall, which was revealed in the decreased prevalence of healthy behaviors practice in every age group and income level. Healthy behaviors, such as smoking cessation, drinking reduction, and PA have been associated with a lower risk cancer and cardiovascular disease-related death, a lower risk of overall death, and an increased healthy life expectancy [8][9][10][36][37][38]. In this study, the level of healthy behavior practice was very low. Only 1 out of 10 men and 2 out of 10 women practiced healthy behaviors, without desirable diet and healthy weight considered. This finding suggests an urgent need for policies and programs to improve individual health behaviors. It is important to note that this study aimed to identify 20-year trends in health behaviors, which are the risk factors of chronic diseases. However, while these trends were examined for smoking behavior, it was not possible to examine 20-year trends for alcohol drinking and PA because new questionnaires were added or old ones were replaced during the study period.
This KNHANES data analysis covering a span of 20 years found that, except for cigarette smoking in men, health behaviors did not improve in Korea, as shown by decreased levels of PA and increased levels of alcohol drinking and binge drinking in women. This finding is believed to be linked to the increased prevalence of chronic diseases like obesity, dyslipidemia, and diabetes. In the future, effective programs need to be implemented to improve health behaviors to prevent the increasing prevalence of chronic diseases and reduce the burden of disease.