Factors involved in Cancer Screening Participation: Multilevel Mediation Model
FFACTORS INVOLVED IN CANCER SCREENING PARTICIPATION: MULTILEVELMEDIATION MODEL
DONGHYUN (ETHAN) KIMGyeonggi Suwon International School, 451 YeongTong-Ro, YeongTong-Gu, Suwon-Si, Gyeonggi-Do, Republic of Korea
Abstract.
In this paper, we identify the factors associated with cancer screening participation inKorea. We expand upon previous studies through a multilevel mediation model and a composite regionalsocioeconomic status index which combines education level and income level. Results of the modelindicate that education level, nutritional education status and income level are significantly associatedwith cancer screening participation. With our findings in mind, we recommend health authorities toincrease promotional health campaigns toward certain at-risk groups and expand the availability ofnutrition education programs. Introduction
Cancer is the leading cause of death in Korea. In 2018, for every 100,000 people, cancer was responsiblefor 154.3 deaths (26.5% of all deaths), a 0.2% increase from 2017 ([16]).Early detection and diagnosis of cancers significantly increases a patient’s chance of survival andreduces the costs associated with treatment. A study in the United States found that women diagnosedwith cervical cancer had significantly greater 5 year survival rates when the cancer was detected in earlystages compared to diagnosis at advanced stages (92% vs 17%) ([7]).Korea’s National Cancer Screening Program covers 6 major cancers, namely gastric cancer, colorectalcancer, breast cancer, cervical cancer, liver cancer, and lung cancer. For National Health Insurance (NHI)beneficiaries in the lower 50% income bracket, cancer screening tests are completely free of cost. For NHIbeneficiaries in the upper 50% income bracket, 90% of cancer screening costs are covered by the NHIand 10% is left for the individual to pay ([13]).Even with the government’s efforts to increase the availability of cancer screening tests, the participa-tion rate was 55.6 % in 2019; rates did not increase much from 50.1 % in 2015 ([14]). In this study, we aimto expand upon previous research and identify the factors associated with cancer screening participationwith a multilevel mediation model.First, section 2 will discuss factors found from previous studies. Next, section 3 will present the modelwith some preliminary statistical tests. Finally, sections 4 and 5 will discuss the results and recommenda number of actions moving forward. 2.
Factors Involved
General Factors from Korean Study and other papers.
We will first take a look at someprevious studies on the factors involved in cancer screening tests. These factors will serve as a foundationfor the model presented in this paper.In 2012, authors Shin, Ji-Yeon and Lee, Duk-Hee conducted a study on the factors associated withthe use of gastric cancer screening services in Korea. Through multivariate analysis, they found thateducation level, alcohol consumption, and smoking habits are significantly associated with participationin gastric cancer screening. More specifically, individuals who graduated from middle school and highschool (adjusted Odds Ratio = 1.21) and those who graduated from college or higher (adjusted OddsRatio = 1.40) were much more likely to utilize gastric cancer screening services than those of lowereducation level. Binge drinkers (adjusted Odds Ratio = 0.94) and frequent binge drinkers (adjustedOdds Ratio = 0.74) were less likely to utilize gastric cancer screening services than non-binge drinkers.Current smokers (adjusted Odds Ratio = 0.75) and ex-smokers (adjusted Odds Ratio = 0.93) were lesslikely to utilize gastric cancer screening services as well.Monthly household income, however, was found to be positively associated with the use of gastriccancer screening services (4th quartile: adjusted Odds Ratio = 1.31; 3rd quartile: adjusted Odds Ratio= 1.09). a r X i v : . [ q - b i o . O T ] A ug DONGHYUN (ETHAN) KIM
The study also found that dietary habits (”trying to eat more vegetables”) are positively associatedwith cancer screening participation when uni-variate logistic regression was conducted (Very often: OddsRatio = 1.00; Somewhat often: Odds Ratio = 0.98; Not at all: Odds Ratio = 0.80) ([12]).Another study from Korea in 2016 found that moderate level of physical activity (1-4 times a week)is positively associated with cancer screening participation (Odds Ratio = 1.20 for organized screeningand Odds Ratio = 1.41 for opportunistic screening) ([4]).2.2.
Relations between factors.
In this section, we will take a look at previous studies that examinedthe relationship between the aforementioned factors.Authors Catherine E. Ross and Chia-ling Wu concluded in their paper that individuals with highereducation levels are more likely to exercise (unstandardized regression coefficient: .295), less likely tosmoke (unstandardized regression coefficient: -.046), and more likely to drink heavily (unstandardizedregression coefficient: .036) ([9]). In another study, authors David M. Cutler and Adriana Lleras-Muneyfound that education is positively associated with one’s diet (regression coefficient: 0.0658) ([2]).Looking at income, author M. Christopher Auld found that drinking is associated with higher income(10% higher for moderate drinking and 12% higher for heavy drinking) and smoking is associated withlower income (smokers earn 8% less than those who do not smoke) ([1]). Another study found that thequality of an individual’s diet/nutrition improved with higher income levels ([5]). Finally, authors BradHumphreys and Jane Ruseski found positive associations between exercise (including swimming, golfing,weight lifting, and running) and income levels ([6]).2.3.
Socioeconomic Status Index.
Regional and individual socioeconomic status indices are oftenbased on income, assets, occupation, and education (or a combination of these variables). These indiceshave varied uses depending on the study in question, including measuring inequality or acting as a mul-tilevel predictor in a model. An example of a composite socioeconomic status index is the ”Kuppuswamysocioeconomic scale,” which combines assets, education level, occupation, and income ([11]).3.
Method
Participants.
This study is based on data obtained from the Seventh Korea National Health andNutrition Examination Survey, which includes data from the years 2016 to 2018. This particular datasetincludes information from 10611 households (24269 individuals) and consists of a survey of health andnutrition and a health examination. Study participants were chosen through stratified cluster sampling;primary sampling units were chosen from the results of the annual Population and Housing Census, fromwhich representative households were drawn ([10]).3.2.
Factors/Variables.
Cancer Screening Participation.
The main variable of interest is whether an individual has par-ticipated in cancer screening in the past 2 years. Possible responses to this question are ”yes” and ”no.”3.2.2.
Level 1 Factors.
Based on the discussion from section 2, several factors are chosen for this study:education level, income level, exercise frequency, drinking frequency, smoking frequency, whether di-etary/nutritional supplements are taken, nutrition education status and nutrition facts label use.Education level is classified into 4 categories: elementary school graduates or below, middle schoolgraduates, high school graduates, and university graduates or higher. Income level is classified into 4categories as well: low, lower-middle, upper-middle, and high.Exercise frequency, drinking frequency, and smoking frequency are all a combination of 2 variablesfrom the dataset, and they are classified into 3 categories. Exercise frequency is classified by often (2-7times a week), occasionally (1-2 times a week), and never/not regularly. Drinking frequency is classifiedby often (2 or more times a week), occasionally (under 5 times a month), and never/not in the past year.Smoking frequency is classified by daily, occasionally/in the past, and never/less than 5 packs in lifetime.Whether dietary/nutritional supplements are taken is a response to the question, ”have you takensupplements regularly for at least 2 weeks in the past year?,” and possible responses are ”yes” and ”no”.Nutrition education status and nutritional facts label use are both binary variables with ”yes” and ”no”for possible responses.
ACTORS INVOLVED IN CANCER SCREENING PARTICIPATION: MULTILEVEL MEDIATION MODEL 3
Level 2 Factors: socioeconomic status.
Continuous Variable.
Note that all variables in the multilevel mediation model are configured tobe continuous variables. Though certain variables are categorical by nature, to aid the interpretation ofresults (especially given the large number of variables and categories), all variables are made continuous.3.3.
Preliminary Statistical Analysis.
Chi-square test.
We first conduct Chi-square tests to assess the chosen factors and determinewhich are significantly associated with cancer screening participation.Variable Degree of Freedom Chi-squared value P-valueedu 3 147.5246 0.000incm 3 199.9009 0.000nutri edu 1 32.7320 0.000diet suppl 1 271.6104 0.000label use new 1 2.3040 0.129exer freq 2 11.6863 0.003smk freq 2 323.9205 0.000drink freq 2 34.9039 0.000
Table 1.
Chi-squared tests with Cancer Screening variableThough 1 variable appears to be statistically insignificant with uni-variate analysis, we choose to keepall variables for the multilevel mediation model (all variables are of theoretical importance).3.3.2.
ICC Analysis.
We calculate intraclass correlation coefficients in mixed-effect models to identifyclustering in primary survey units for education levels and income levels.Variable Level 2 Level 2 Predictor ICCEducation Level primary survey unit NA .0879755Income Level primary survey unit NA .2242859Education Level primary survey unit (psu) average education level per psu 4.52e-23Income Level primary survey unit (psu) average income level per psu 7.97e-23
Table 2.
Intraclass correlation coefficientsClearly, the addition of level 2 predictors (with random intercepts) significantly reduce the amount ofcorrelation within groups. As such, level 2 predictors will be included in the multilevel mediation model.
DONGHYUN (ETHAN) KIM
Multilevel Mediation Model.
With the above variables, we now propose the following multilevelmediation model.
Figure 1.
Multilevel Mediation Model edu_lvl ε income_lvl ε nutri_edusmk_freq ε exer_freq ε drink_freq ε label_use ε diet_suppl ε screening ε ses Tables 3 and 4 contain the results of the model. Note that the model was run on Stata/IC 16.1
ACTORS INVOLVED IN CANCER SCREENING PARTICIPATION: MULTILEVEL MEDIATION MODEL 5
Table 3.
Multilevel Mediation Model: Coefficients (1) (2) (3)avg income lvl avg edu lvl composite sesscreeningexer freq 0.00831 0.00831 0.00831(1.42) (1.42) (1.42)drink freq -0.0184 ∗∗ -0.0184 ∗∗ -0.0184 ∗∗ (-2.61) (-2.61) (-2.61)smk freq -0.0768 ∗∗∗ -0.0768 ∗∗∗ -0.0768 ∗∗∗ (-11.43) (-11.43) (-11.43)label use -0.0275 ∗∗ -0.0275 ∗∗ -0.0275 ∗∗ (-2.84) (-2.84) (-2.84)income lvl 0.0504 ∗∗∗ ∗∗∗ ∗∗∗ (12.19) (12.19) (12.19)diet suppl 0.123 ∗∗∗ ∗∗∗ ∗∗∗ (13.45) (13.45) (13.45)edu lvl -0.0434 ∗∗∗ -0.0434 ∗∗∗ -0.0434 ∗∗∗ (-8.74) (-8.74) (-8.74)nutri edu 0.0963 ∗∗∗ ∗∗∗ ∗∗∗ (4.98) (4.98) (4.98)cons 0.664 ∗∗∗ ∗∗∗ ∗∗∗ (28.80) (28.80) (28.80)exer freqincome lvl 0.0598 ∗∗∗ ∗∗∗ ∗∗∗ (12.13) (12.13) (12.13)edu lvl 0.121 ∗∗∗ ∗∗∗ ∗∗∗ (24.93) (24.93) (24.93)cons 0.900 ∗∗∗ ∗∗∗ ∗∗∗ (51.70) (51.70) (51.70)drink freqincome lvl -0.0167 ∗∗∗ -0.0167 ∗∗∗ -0.0167 ∗∗∗ (-3.62) (-3.61) (-3.61)edu lvl 0.176 ∗∗∗ ∗∗∗ ∗∗∗ (38.58) (38.62) (38.60)cons 1.448 ∗∗∗ ∗∗∗ ∗∗∗ (89.99) (90.00) (90.03)smk freqincome lvl -0.0597 ∗∗∗ -0.0597 ∗∗∗ -0.0597 ∗∗∗ (-11.65) (-11.62) (-11.64)edu lvl 0.0484 ∗∗∗ ∗∗∗ ∗∗∗ (9.34) (9.35) (9.34)cons 1.346 ∗∗∗ ∗∗∗ ∗∗∗ (73.85) (73.87) (73.88)label useincome lvl -0.00308 -0.00308 -0.00308(-0.84) (-0.84) (-0.84)edu lvl 0.0774 ∗∗∗ ∗∗∗ ∗∗∗ (21.24) (21.25) (21.25)cons 0.119 ∗∗∗ ∗∗∗ ∗∗∗ (8.62) (8.62) (8.62)income lvledu lvl 0.0319 ∗∗∗ ∗∗∗ ∗∗∗ (5.70) (5.15) (4.82)M3[avg income lvl] 1(.)M3[avg edu lvl] 1(.)M3[composite ses] 1(.)cons 2.379 ∗∗∗ ∗∗∗ ∗∗∗ (80.06) (82.18) (81.73)diet supplincome lvl 0.0427 ∗∗∗ ∗∗∗ ∗∗∗ (12.81) (12.80) (12.80)nutri edu 0.0320 ∗∗ ∗∗ ∗∗ (2.61) (2.61) (2.61)cons 0.387 ∗∗∗ ∗∗∗ ∗∗∗ (41.94) (41.93) (41.93)edu lvlM3[avg income lvl] 0.539 ∗∗∗ (30.25)M3[avg edu lvl] 0.662 ∗∗∗ (32.09)M3[composite ses] 0.582 ∗∗∗ (31.66)cons 2.483 ∗∗∗ ∗∗∗ ∗∗∗ (156.12) (137.84) (150.27)nutri eduM3[avg income lvl] 0.0166 ∗∗∗ (3.85)M3[avg edu lvl] 0.0150 ∗∗ (3.19)M3[composite ses] 0.0165 ∗∗∗ (3.77)cons 0.101 ∗∗∗ ∗∗∗ ∗∗∗ (46.12) (46.08) (46.09) N t statistics in parentheses ∗ p < . ∗∗ p < . ∗∗∗ p < . DONGHYUN (ETHAN) KIM
Table 4.
Multilevel Mediation Model: Odds Ratio (1) (2) (3)avg income lvl avg edu lvl composite sesscreeningexer freq 1.008 1.008 1.008(1.42) (1.42) (1.42)drink freq 0.982 ∗∗ ∗∗ ∗∗ (-2.61) (-2.61) (-2.61)smk freq 0.926 ∗∗∗ ∗∗∗ ∗∗∗ (-11.43) (-11.43) (-11.43)label use 0.973 ∗∗ ∗∗ ∗∗ (-2.84) (-2.84) (-2.84)income lvl 1.052 ∗∗∗ ∗∗∗ ∗∗∗ (12.19) (12.19) (12.19)diet suppl 1.131 ∗∗∗ ∗∗∗ ∗∗∗ (13.45) (13.45) (13.45)edu lvl 0.958 ∗∗∗ ∗∗∗ ∗∗∗ (-8.74) (-8.74) (-8.74)nutri edu 1.101 ∗∗∗ ∗∗∗ ∗∗∗ (4.98) (4.98) (4.98)exer freqincome lvl 1.062 ∗∗∗ ∗∗∗ ∗∗∗ (12.13) (12.13) (12.13)edu lvl 1.128 ∗∗∗ ∗∗∗ ∗∗∗ (24.93) (24.93) (24.93)drink freqincome lvl 0.983 ∗∗∗ ∗∗∗ ∗∗∗ (-3.62) (-3.61) (-3.61)edu lvl 1.192 ∗∗∗ ∗∗∗ ∗∗∗ (38.58) (38.62) (38.60)smk freqincome lvl 0.942 ∗∗∗ ∗∗∗ ∗∗∗ (-11.65) (-11.62) (-11.64)edu lvl 1.050 ∗∗∗ ∗∗∗ ∗∗∗ (9.34) (9.35) (9.34)label useincome lvl 0.997 0.997 0.997(-0.84) (-0.84) (-0.84)edu lvl 1.081 ∗∗∗ ∗∗∗ ∗∗∗ (21.24) (21.25) (21.25)income lvledu lvl 1.032 ∗∗∗ ∗∗∗ ∗∗∗ (5.70) (5.15) (4.82)M3[avg income lvl] 2.718(.)M3[avg edu lvl] 2.718(.)M3[composite ses] 2.718(.)diet supplincome lvl 1.044 ∗∗∗ ∗∗∗ ∗∗∗ (12.81) (12.80) (12.80)nutri edu 1.033 ∗∗ ∗∗ ∗∗ (2.61) (2.61) (2.61)edu lvlM3[avg income lvl] 1.714 ∗∗∗ (30.25)M3[avg edu lvl] 1.939 ∗∗∗ (32.09)M3[composite ses] 1.790 ∗∗∗ (31.66)nutri eduM3[avg income lvl] 1.017 ∗∗∗ (3.85)M3[avg edu lvl] 1.015 ∗∗ (3.19)M3[composite ses] 1.017 ∗∗∗ (3.77) N Exponentiated coefficients; t statistics in parentheses ∗ p < . ∗∗ p < . ∗∗∗ p < . ACTORS INVOLVED IN CANCER SCREENING PARTICIPATION: MULTILEVEL MEDIATION MODEL 7 Results
We see in all 3 versions of the model that drinking frequency (Odds Ratio = 0.982) and smokingfrequency (Odds Ratio = 0.926) are both negatively correlated with cancer screening participation.We also observe that income level (Odds Ratio = 1.052) is positively correlated with cancer screeningparticipation. Nutrition education status (Odds Ratio = 1.101) and the consumption of dietary supple-ments (Odds Ratio = 1.131) are shown to be positively associated with cancer screening participationas well.Furthermore, education level, nutrition education status, and income level appear to have significanteffects on other variables. Income level is positively correlated with exercise frequency (Odds Ratio =1.062) and negatively correlated with drinking frequency (Odds Ratio = 0.983) and smoking frequency(Odds Ratio = 0.942). On the other hand, education level is positively correlated with exercise frequency(Odds Ratio = 1.128), drinking frequency (Odds Ratio = 1.192), smoking frequency (Odds Ratio =1.050), and income level (Odds Ratio = 1.032).Moreover, an individual’s income level (Odds Ratio = 1.044) and nutrition education status (OddsRatio = 1.033) are both positively correlated with the use of diet supplements. Education level is shownto be positively associated with nutrition label usage as well (Odds Ratio = 1.081).Finally, we see that all variations of the level 2 predictor (socioeconomic status index) are significantlyassociated with an individual’s education level (Odds Ratio with average income level: 1.714; averageeducation level: 1.939; composite index: 1.790) and nutrition education status (Odds Ratio with averageincome level: 1.017; averaged education level: 1.015; composite index: 1.017).We now address some unexpected results of the proposed model. Education level was found to benegatively correlated with cancer screening participation (Odds Ratio: 0.958). To examine the effect ofeducation level more closely, we conduct logistic regression with a categorical education level variable ofgreater specificity (7 categories).(1)screeningscreeningNo Education 0(.)Elementary School 0.698 ∗∗∗ (7.83)Middle School 0.829 ∗∗∗ (8.88)High School 0.539 ∗∗∗ (6.40)2/3 Year College -0.0449(-0.51)4 Year College -0.0189(-0.22)Graduate School 0.636 ∗∗∗ (6.12)cons 0.0277(0.35) N t statistics in parentheses ∗ p < . ∗∗ p < . ∗∗∗ p < . Table 5.
Logistic Regres-sion: Coefficients (1)screeningscreeningNo Education 1(.)Elementary School 2.010 ∗∗∗ (7.83)Middle School 2.290 ∗∗∗ (8.88)High School 1.714 ∗∗∗ (6.40)2/3 Year College 0.956(-0.51)4 Year College 0.981(-0.22)Graduate School 1.889 ∗∗∗ (6.12) N Exponentiated coefficients; t statistics in parentheses ∗ p < . ∗∗ p < . ∗∗∗ p < . Table 6.
Logistic Regres-sion: Odds Ratio
DONGHYUN (ETHAN) KIM
We see that compared to lack of schooling, having attended elementary school, middle school, highschool, or graduate school all play a significant role (positive correlation) in cancer screening participation.However, having attended high school or college is not correlated with cancer screening participation( p ≥ . p < . Discussion
Consistent with results from past studies, drinking frequency and smoking frequency are shown to benegatively correlated with cancer screening participation ([12]). These 2 factors are generally consideredcauses or risk factors for certain cancers. Hence, smokers and drinkers would benefit from increasedcancer screening test participation, especially for lung cancer, larynx cancer, throat cancer, and otherhigh-risk cancers. Given the large number of smokers and drinkers in Korea, cancer screening campaignsfocused on such groups should be set up by health authorities.Also, like past studies, income in found to be positively correlated with cancer screening participation([12]). Even with national health insurance and subsidized screening costs, income is still a relevantfactor. Possible causes include the following: not all screening tests are offered at lowered costs andhigher income often results in more available time for health check-ups.As past studies have shown, dietary factors (use of dietary supplements) are positively correlated withcancer screening participation ([12]). This can be explained by the fact that individuals who care abouttheir health are willing to participate in regular check-ups.However, statistically significant relationships between exercise frequency and cancer screening par-ticipation are not found in this study; contrary to past studies ([4]). Unlike dietary factors, exercise iseasily accessible to most individuals; therefore, correlations are less likely to be observed.Furthermore, the relationship between education levels and exercise frequency, drinking frequency, anddietary factors (use of nutrition labels) are in line with previous studies ([9], [2]). Also, the relationshipbetween education levels and income levels is consistent with past surveys ([8]). However, unlike pastresearch, smoking frequency is found to have positive associations with education ([9]). This can beexplained simply by the prevalence of smoking in Korea ([3]). To reduce smoking rates, health authoritiesshould increase smoking campaigns, especially in high schools around the nation.Looking at income, the relationship between income levels and exercise frequency, dietary factors(use of dietary supplements) and smoking frequency is consistent with past studies ([1], [5],[6]). However,unlike past studies, drinking frequency is shown to have negative associations with income ([1]). Increasedawareness of the risks of heavy drinking may be related with these results. To reduce heavy drinking,health authorities should consider increasing health campaigns focused on lower income regions.The relationship between nutrition education status and dietary supplement usage is consistent withprevious findings as well ([15]).However, no statistically significant relationship between income and nutrition label use was found.We believe that this may be related with the fact that individuals of lower income need to plan out theirmeals, often with nutrition in mind.Taking a look at education levels, we see that its effect on cancer screening participation are contraryto previous studies–even with univariate logistic regression, we find that having attended college or highschool is not correlated with cancer screening participation ([12]). This can be explained by the prevalenceof nutrition education programs that do not require higher levels of education. With such programs,individuals can realize the need for and availability of cancer screening tests by visiting local public
ACTORS INVOLVED IN CANCER SCREENING PARTICIPATION: MULTILEVEL MEDIATION MODEL 9 health centers, community centers and welfare facilities. Given the significance of nutrition educationstatus (as shown with its high odds ratio) with cancer screening participation, local health authoritiesshould consider expanding the availability of these programs.Finally, because regional socioeconomic status factors are found to be significantly correlated witheducation levels, nutrition education statuses, and income levels (variables which are associated withother relevant variables as well), improving access to education opportunities and nutritional educationcenters should be prioritized by local governments.Overall, we found in this study that education and income play a key role in cancer screening par-ticipation. Our findings are significant in that we expanded upon previous research, identifying the roleof education and income not only on cancer screening participation but also on other relevant factorsthrough a multilevel mediation model (with considerations to regional socioeconomic status indices).We were also successful in developing a composite socioeconomic status index which accurately cap-tured differences in each primary survey unit for education and income.Nevertheless, the study has some limitations. First, the data obtained from the Korea National Healthand Nutrition Examination Survey was, for the most part, a self-report study. Thus, the data usedmay have contained some inaccuracies which could’ve impacted the results of the model. Second, as isthe nature of cross-sectional data, we cannot establish causal relationships or infer causality with ourmediation model.We hope that related studies continue in the future, expanding upon our results and improving onaforementioned flaws. Causal research, for instance, could help determine causality. Another optionwould be to incorporate more variables like working hours and distance to public health centers into themediation model.Even with these limitations, our novel findings and policy recommendations (special attention shouldbe paid to increasing accessibility of nutrition education programs) may be of use for health authoritiesaiming to increase participation in cancer screening tests.
References
1. Auld, M. C. (2005). Smoking, drinking, and income.
Journal of Human Resources , 40(2), 505-518.2. Cutler, D. M., & Lleras-Muney, A. (2006).
Education and health: evaluating theories and evidence (No. w12352).National bureau of economic research.3. Gunter, R., Szeto, E., Jeong, S. H., Suh, S., & Waters, A. J. (2020). Cigarette Smoking in South Korea: A NarrativeReview. Korean journal of family medicine, 41(1), 3–13. https://doi.org/10.4082/kjfm.18.00154. Hahm MI, Chen HF, Miller T, O’Neill L, Lee HY. Why Do Some People Choose Opportunistic Rather Than OrganizedCancer Screening? The Korean National Health and Nutrition Examination Survey (KNHANES) 2010-2012.
CancerRes Treat.
Journal of the Academy of Nutrition and Dietetics , 113(2), 297-306.6. Humphreys, B. R., & Ruseski, J. E. (2015). The Economic Choice of Participation and Time Spent in Physical Activityand Sport in Canada.
International Journal of Sport Finance
OECD Publishing , Paris,https://doi.org/10.1787/f8d7880d-en.9. Ross, C., & Wu, C. (1995). The Links Between Education and Health.
American Sociological Review
Indian J Public Health , 56(1), 103-104.12. Shin, J.-Y., & Lee, D.-H. (2012). Factors Associated with the Use of Gastric Cancer Screening Services in Korea: TheFourth Korea National Health and Nutrition Examination Survey 2008 (KNHANES IV).
Asian Pacific Journal ofCancer Prevention , 13(8), 3773–3779. https://doi.org/10.7314/apjcp.2012.13.8.377313. 국 가 암 검 진 비 용 지 원 . (n.d.). 정 부 국 민 건강 보 험 공 단 . (2019). 국 가 암 조 기 검 진 사 업 수 검 률 이윤 정 , 강 민 지 , 백 희 영 , & 송 윤 주 . (2017). 한 국 성 인의 성 별 에 따 른 식 이 보 충 제 섭 취 및 관 련 요 인 에 대 한 연 구 : 제 기 국 민 건강 영 양 조 사 (2007-2009) 자 료 를 이 용 하 여 . 대 한 지 역 사 회 영 양 학 회 지 , 22(4), 347-355.16. 통 계 청 . (2019). 사 망 원 인 별 사 망률 추 이 Gyeonggi Suwon International School, 451 YeongTong-Ro, YeongTong-Gu, Suwon-Si, Gyeonggi-Do, Re-public of Korea
E-mail address ::