Ciencia y Salud, Vol. 8, No. 3, septiembre-diciembre, 2024 • ISSN (impreso): 2613-8816 • ISSN (en línea): 2613-8824

IDENTIFYING PATTERNS OF TOBACCO USE IN MIDDLE AND HIGH SCHOOL STUDENTS IN THE DOMINICAN REPUBLIC: A LATENT CLASS ANALYSIS

Identificando patrones de consumo de tabaco en estudiantes de secundaria y preparatoria en la República Dominicana: un análisis de clase latente

DOI: https://doi.org/10.22206/cysa.2024.v8i3.2862

César Arredondo Abreu¹, Kari-Lyn Sakuma²

Recibido: 12 de junio, 2023 • Aceptado: 26 de febrero, 2024

Cómo citar: Arrendondo Abreu C & Sakuma K-L. (2024). Identifying patterns of tobacco use in middle and high school students in the Dominican Republic: a latent class análisis. Ciencia y Salud, 8(3), 45-54. https://doi.org/10.22206/cysa.2024.v8i3.2862

Introduction

The reduction of tobacco use is a health priority worldwide¹. In 2013, the United Nations set a target to reduce the relative prevalence of tobacco by 30% by 2025². Most high-income countries are expected to reach that target; However, low and middle-income countries remain in the early stages of tobacco control¹, ².

Currently, the Dominican Republic is a middle-income tobacco-producing country located in the Caribbean and with limited tobacco control efforts^{3, 4}. In 2018, the overall tobacco smoking prevalence in the Dominican Republic was 9.4%, with 11.2% for men and 7.4% for women reporting current smoking⁵. The Dominican Republic is not on track to meet the goal of reducing tobacco prevalence by 30% by 2025².

One study in the Dominican Republic found that most smokers started before age 16, and the initiation age can be as early as six years old⁶. Adolescents with a younger smoking initiation have elevated risks of negative health outcomes such as nicotine dependence⁷. Prevention interventions that delay the start of tobacco use are an essential tool to fighting needed to fight the tobacco epidemic; yet, little is known about the initiation and use behaviors of Dominican youth⁷.

The introduction of many new tobacco products to the market may mean that youth are more likely to be dual- and poly-tobacco users⁸. In the Dominican Republic, surveillance data suggest that Dominican youth may also be exposed to and use multiple products. In 2011, 19.5% of Dominican students reported using tobacco products other than cigarettes⁹. Understanding the different tobacco use patterns may help to identify various risk groups to target prevention and cessation interventions for youth. In addition, understanding what social and environmental exposures might be associated with these patterns of tobacco use may further identify high-risk groups of youth.

Our study aims to understand the patterns of tobacco use in the Dominican Republic and identify what factors are related to those patterns to inform future prevention efforts.

Methods

Dataset

We analyzed the Dominican Republic’s 2016 Global Youth Tobacco Survey dataset (GYTS) for this study. The GTYS is a school-based cross-sectional survey that employs a two-stage clustered design to obtain a national representative sample (10). The Dominican Republic’s GTYS data was collected by the Ministry of Health in the Dominican Republic. A total of 1532 seventh to twelfth grade students from 15 different schools participated as the sample. The overall response rate was 79.5%.

Measures

Tobacco Products

Cigarette use. Cigarette ever use was assessed by asking participants, “Have you ever tried or experimented with cigarette smoking, even one or two puffs?” and answer choices were 1= “yes” or 0= “no.” Cigarette current use was assessed by asking participants, “During the past 30 days, on how many days did you smoke cigarettes?” Answers were recoded and dichotomized into 1= “1 day to all 30 days” and 0=”0 days.”

Shisha use. Shisha ever use was assessed by asking participants, “Have you ever tried or experimented with shisha smoking, even one or two puffs?” and answer choices were 1= “yes” or 0= “no.” Shisha current use was assessed by asking participants, “During the past 30 days, on how many days did you smoke shisha?” Response choices were “0 days,” “1 or 2 days,” “3 to 5 days,” “6 to 9 days,” “10 to 19 days,” “10 to 19 days,” “20 to 29 days,” and “all 30 days.” Answers were recoded and dichotomized into 1= “1 day to all 30 days” and 0= “0 days.”

Smokeless tobacco use. Smokeless tobacco ever use was assessed by asking participants, “Have you ever tried or experimented with any form of smokeless tobacco products (such as dip, chewing tobacco or snuff)?” and answer choices were 1= “yes” or 0=“no.” Smokeless tobacco current use was assessed by asking participants, “During the past 30 days, did you use any form of smokeless tobacco products (such as dip, chewing tobacco or snuff)?”) and answer choices were 1= “yes” or 0=“no.”

Electronic cigarettes tobacco use. Electronic cigarette ever use was assessed by asking participants, “In total, how many days have you used an electronic cigarette or e-cigarette in your entire life?” Response choices were “0 days,” “1 day,” “2 to 10 days,” “11 to 20 days,” “21 to 50 days,” “51 to 100 days,” and “More than 100 days.” Answers were recoded and dichotomized into 1= “yes and 0= “No.” and answer choices were 1=“yes” or 0=“no.” Electronic cigarette current use was assessed by asking participants, “During the past 30 days, on how many days did you use electronic cigarettes?” Response choices were “0 days,” “1 or 2 days ,” “3 to 5 days ,” “6 to 9 days,” “10 to 19 days,” “10 to 19 days,” “20 to 29 days,” and “all 30 days.” Answers were recoded and dichotomized into 1= “1 day to all 30 days” and 0= “0 days.”

Sociodemographic variables

Gender. Gender was assessed by asking participants, “What is your sex?” Response choices were 1 = “Male” or 0 = “Female.”

Age. Age was measured by asking participants, “How old are you?” Response choices were “11 or younger,” “12 years old,” “13 years old,” “14 years old,” “15 years old,” “16 years old,” and “17 years old or older”. Answers were recoded and dichotomized into 1= “15 to 17 years or older” and 0= ”11 or younger to 14 years old”. We recoded it this way because the Dominican Republic’s secondary education is divided into two cycles. The first one is from seventh to ninth grade, and the second is from tenth to twelfth grade, which roughly corresponds to the ages we have grouped together¹¹.

Weekly spending money. Weekly spending money was measured by asking participants, “During an average week, how much money do you have that you can spend on yourself, however you want?” Response choices were “I usually don’t have any spending money”, “Less than Dominican Pesos (DOP)100.00,” “DOP 100.00 to DOP 199.00,” “DOP 200.00 to DOP 299.00,” “DOP 300.00 to DOP 399.00,” “DOP 500.00 to DOP 999.00,” and “DOP 1000.00 or more.” Answers were recoded and dichotomized into 1= “DOP 200 to DOP 9999” and 0= “No spending money to DOP 199.” The price of the most sold 20 cigarette pack was DOP 150 (equivalent to $7.28) (12). We decided to set the cutoff at DOP 200 since it reflected that all students above DOP 200 had enough money to buy at least one pack each week.

Social modeling variables

Exposure to tobacco inside their house. Exposure to tobacco inside their house was measured by asking participants, “During the past 7 days, on how many days has anyone smoked inside your home, in your presence?” Response choices were “0 days,” “1 to 2 days,” “3 to 4 days,” “5 to 6 days,” and “7 days.” Answers were recoded and dichotomized into 1= “1 to 7 days” and 0= ”0 days.”

Exposure to tobacco inside enclosed spaces. Exposure to tobacco inside enclosed spaces was measured by asking participants, “During the past 7 days, on how many days has anyone smoked in your presence, inside any enclosed public place, other than your home (such as school, stores, shops, restaurants, shopping malls, hair salon, barber shop, liquor store)?” Response choices were “0 days,” “1 to 2 days,” “3 to 4 days,” “5 to 6 days,” and “7 days.” Answers were recoded and dichotomized into 1= “1 to 7 days” and 0= ”0 days.”

Exposure to tobacco outside. Exposure to tobacco outside was measured by asking participants, “During the past 7 days, on how many days has anyone smoked in your presence, at any outdoor public place (such as playgrounds, sidewalks, entrances to buildings, parks, beaches, colmados or mini markets)?” Response choices were “0 days,” “1 to 2 days,” “3 to 4 days,” “5 to 6 days,” and “7 days.” Answers were recoded and dichotomized into 1= “1 to 7 days” and 0= ”0 days.”

Statistical Analyses

Descriptive statistics were calculated in R to summarize participant characteristics and Survey package was used to estimate the weighted proportions and confidence intervals^{13, 14}. Latent class analyses (LCA) were used to identify distinct patterns of tobacco use among a DR youth. Multinomial regression analyses were then used to estimate odds ratios for each covariate to compare the probability of belonging for each class versus a reference class.

Absolute and relative model fit statistics (the G2 likelihood ratio Chi-Square Test, the Bayesian Information Criteria (BIC), Akaike Information Criteria (AIC)), were used to compare model selection. Final model selection depended on fit statistics and interpretability.

Missing Data.

Missing data on age, gender, and average weekly spending money was imputed using the using k-nearest neighbor (KNN) method¹⁵. To perform KNN imputation, we used the fancyimputer module in Python 3.6 with k=5 selected¹⁶. Imputations that had a value equal to or higher than .5 were rounded up to 1, while imputations lower than .5 were rounded up to 0.

We computed multinomial logistic regression using posterior probabilities to assign members to latent classes¹⁷. Sociodemographic characteristics and social modeling variables were added as covariates to investigate whether the classes differed by these variables. We used Maximum Likelihood Estimation to estimate all of the LCA models, and multinomial logistic regression LCA was conducted in Mplus 7.7^{18, 19}.

Results

The sociodemographic characteristics of the study sample (N= 1,532) are presented in Table 1. The overall analytic sample consisted of nearly 50% male youth, 77% were age 15 or older, and 49% had spending money between DOP 200 – DOP 9999. The most common space where students were exposed to people smoking tobacco was in outside areas (30%), followed by enclosed spaces (25%), and lastly, inside their houses (16%). The most common tobacco product used was shishas (39% lifetime use; 17% last month use), followed by electronic cigarettes (23% lifetime use; 11% last month use), cigarettes (16% lifetime use; 4% last month use) and smokeless tobacco (7% lifetime use; 3% last month use).

Latent Class Models

Table 2 shows model fit statistics for each model with an increasing number of classes. Latent class models ranging from one class to seven classes were created. A 5-class model was selected due to model statistics, interpretability, and parsimony (Loglikelihood = -3418.977, AIC = 6949.953, BIC = 7248.676).

Latent classes and the item-response probability are presented in Table 3. The Nonusers class captured majority of the sample (58.44%), followed by Shisha experimenters (18.55%), Poly-tobacco experimenters (10.66%), Poly-tobacco users (8.57%), and last was the Smokeless users and cigarette and shisha experimenters (3.79%).

Predictors of class membership

Table 4 presents the multinomial regression analysis results using covariates to identify differences in latent class membership. Relative to the Nonuser class, members of the Shisha Experimenter class had higher odds of being male (OR = 2.670; 95% CI = 1.936 – 3.813), having between DOP$200 - DOP$9999 for spending money (OR = 2.273; 95% CI = 1.355 -2.596), and having observed someone smoking tobacco in their house (OR = 1.725; 95% CI = 1.034-2.876). Relative to Nonusers, members of the Poly-tobacco experimenters class had higher odds of having observed someone smoking tobacco in their house (OR = 2.948; 95% CI = 1.650 - 2.876). Relative to Nonuser , members of the Poly-tobacco users class had higher odds of having observed someone smoking tobacco in their house (OR = 4.773; 95% CI = 1.904 – 11.968) and outside (OR = 2.702; 95% CI = 1.248 – 5.849) Relative to Nonuser, members of the Smokeless users and cigarette and shisha experimenters class had higher odds of having observed someone smoking tobacco in their house (OR = 3.770; 95% CI = 1.454 – 9772) and outside (OR = 6.633; 95% CI = 1.261- 34.886).

Discussion

Understanding groups of people who display similar risk patterns of tobacco use may lead to more effective tobacco prevention and cessation interventions and messaging. The current study identified five distinct groups of tobacco users among Dominican Republic youth. Most students have never experimented with tobacco; however, most students who have ever experimented with tobacco have experimented with multiple substances. This is similar to other studies that have used LCA to classify tobacco users^{20, 22}. Our study, however, found different patterns of tobacco experimentation compared to those in the US, suggesting that subpopulations of tobacco use might vary between countries. Studies using similar datasets and methodologies in the US found four distinct tobacco use groups in high and middle school. In their studies, the most prevalent group, after nonusers, is the group of ever use of cigarettes and cigars. However, in the Dominican Republic, shisha experimenters are the most prevalent group, after non-users.

Shisha is the most prevalent form of nicotine consumption in the youth of the Dominican Republic. Previous research in the country found that shishas are the second most prevalent product after cigarettes in adults⁴. Given the high prevalence of shisha use and the identified patterns of tobacco smoking, shisha use appears to be the main exposure point for tobacco use. Previous research in the US has also identified shishas as an entry point for other tobacco products^{23, 24}. Shishas are generally perceived to be less harmful and more socially acceptable²⁵. Since shisha are used along with other tobacco products, to reduce tobacco use we need public health interventions that focus on multiple concurrent use of tobacco products. The Dominican Republic has made some progress in control of shishas. In 2019 they banned the use of Shishas in enclosed spaces to reduce secondhand smoking²⁶. The Dominican Republic is also one of the eleven countries in the Americas to raise their taxes on tobacco products to lower accessibility by 10%²⁷. This criterion was set by the WHO in 2015²⁸. However, there has been no evaluation to see if these policies have been successful in reducing shisha use among teenagers.

The social modeling variables that we used as covariates predicted group membership. Students who observed people smoking tobacco in their houses were more likely to belong to the shisha experimenter, poly-tobacco experimenter, poly-tobacco users, and smokeless tobacco with experimentation classes than the non-users class. Likewise, students who had observed people smoking outside were more likely to belong to the poly-tobacco and smokeless tobacco users classes than the non-users class. Interventions that focus on changing social norms have successfully reduced tobacco use, morbidity, and mortality^{29, 30}. By changing the social norms, adolescents will be less exposed to tobacco products inside and outside their homes.

This study is subject to some limitations. The study uses a cross-sectional study design. Therefore, it is only a snapshot in time and not able to capture transitions between classes. However, even with cross-sectional data, we could show that most adolescents who use tobacco in the Dominican Republic use different forms of tobacco. This indicates a need to focus on the prevention of multiple tobacco products to reduce tobacco use.

Further research using longitudinal data is necessary to investigate whether these classes represent different stages of tobacco use or if they represent stable subgroups of users. Future studies should include other variables that have been related to tobacco use, such as attitudes, self-efficacy to resist tobacco, and the price of tobacco products.

Our study advances the limited literature on tobacco use in the Dominican Republic. By using nationally representative data, we have identified groups in tobacco patterns in middle and high school students in the Dominican Republic. By analyzing the prevalence of these groups, we can conclude that most tobacco users in middle and high school use multiple tobacco products. This highlights the need to have prevention efforts focused on multiple tobacco products.

References

1. World Health Organization. WHO REPORT ON THE GLOBAL TOBACCO EPIDEMIC, 2019: Offer Help to quit tobacco use. 2019.

2. Bilano V, Gilmour S, Moffiet T, d’Espaignet ET, Stevens GA, Commar A, et al. Global trends and projections for tobacco use, 1990–2025: an analysis of smoking indicators from the WHO Comprehensive Information Systems for Tobacco Control. The Lancet. 2015 Mar;385(9972):966–76.

3. Dozier AM, Ossip-Klein DJ, Diaz S, Chin NP, Sierra E, Quinones Z, et al. Tobacco use in the Dominican Republic: understanding the culture first. 2006;7.

4. Ossip-Klein D, Fisher S, Diaz S, Quinones Z, Sierra E, Dozier A, et al. Tobacco use in six economically disadvantaged communities in the Dominican Republic. Nicotine Tob Res. 2008 May;10(5):851–60.

5. WHO. Global Health Observatory data repository: Age-standardized estimates of current tobacco use, tobacco smoking and cigarette smoking [Internet]. 2020. Available from: https://apps.who.int/gho/data/node.main.TOBAGESTDCURR?lang=en

6. Vincent AL, Bradham DD, Ureña Rojas CA, Fisher SK. The Dominican Republic and the Marlboro brand: a cigarette smoking survey and status report. Bull Pan Am Health Organ. 1993;27(4):370–81.

7. United States Surgeon General. Preventing Tobacco Use among Youth and young Adults: A Report of the Surgeon General: (603152012-001) [Internet]. American Psychological Association; 2012 [cited 2022 Jul 4]. Available from: http://doi.apa.org/get-pe-doi.cfm?doi=10.1037/e603152012-001

8. Jebai R, Osibogun O, Li W, Gautam P, Bursac Z, Ward KD, et al. Temporal Trends in Tobacco Product Use Among US Middle and High School Students: National Youth Tobacco Survey, 2011-2020. Public Health Rep. 2023;138(3):483–92.

9. World Health Organization. Dominican Republic 2011 (Ages 13-15) Global Youth Tobacco Survey (GYTS) Fact Sheet. 2013.

10. Warren CW, Jones NR, Peruga A, Chauvin J, Baptiste JP, Costa de Silva V, et al. Global youth tobacco surveillance, 2000-2007. Morbidity and mortality weekly report. 2008;57(1):1–26.

11. Consejo Nacional de la Educación. Ordenanza no.03-2013. Mediante la gual se modifica la estructura académica del sistema educativo dominicano, estableciendo tres niveles educativos de seis (6) años cada uno, subdividio en dos (2) ciclos de tres (3) años, que entraran en vigencia por esta. 03–2013.

12. WHO. Situation of tobacco control by country: Dominican Republic [Internet]. Available from: https://www.paho.org/sites/default/files/DominicanRepublic.pdf

13. R Core Team. R: A language and environment for statistical computing [Internet]. R Foundation for Statistical Computing; 2020. Available from: https://www.R-project.org/

14. Lumley T. Survey: analysis of complex survey samples. 2020.

15. Taunk K, De S, Verma S, Swetapadma A. A Brief Review of Nearest Neighbor Algorithm for Learning and Classification. In: 2019 International Conference on Intelligent Computing and Control Systems (ICCS) [Internet]. Madurai, India: IEEE; [cited 2023 Jun 12] 2019;1255–60. Available from: https://ieeexplore.ieee.org/document/9065747/

16. Rubinsteyn A, Feldman S. Fancyimpute: Matrix completion and feature imputation algorithms. 2021.

17. Collins LM, Lanza ST. Latent class and latent transition analysis: with applications in the social behavioral, and health sciences. Hoboken, N.J: Wiley; 2010:285. (Wiley series in probability and statistics).

18. Sung YJ, Geyer CJ. Monte Carlo likelihood inference for missing data models. Ann Stat [Internet]. 2007 Jul 1 [cited 2021 Apr 22];35(3). Available from: https://projecteuclid.org/journals/annals-of-statistics/volume-35/issue-3/Monte-Carlo-likelihood-inference-for-missing-data-models/10.1214/009053606000001389.full

19. Muthén LK, Muthén BO. MPLUS. Los Angeles, CA; 2020.

20. Gilreath TD, Leventhal A, Barrington-Trimis JL, Unger JB, Cruz TB, Berhane K, et al. Patterns of Alternative Tobacco Product Use: Emergence of Hookah and E-cigarettes as Preferred Products Amongst Youth. J Adolesc Health. 2016 Feb;58(2):181–5.

21. Lisha NE, Thrul J, Ling PM. Latent Class Analysis to Examine Patterns of Smoking and Other Tobacco Products in Young Adult Bar Patrons. J Adolesc Health. 2019 Jan;64(1):93–8.

22. Yu M, Sacco P, Choi HJ, Wintemberg J. Identifying patterns of tobacco use among US middle and high school students: A latent class analysis. Addict Behav. 2018 Apr;79:1–7.

23. Hampson SE, Tildesley E, Andrews JA, Barckley M, Peterson M. Smoking Trajectories Across High School: Sensation Seeking and Hookah Use. Nicotine Tob Res. 2013 Aug 1;15(8):1400–8.

24. Huh J, Leventhal AM. Progression of Poly-tobacco Product Use Patterns in Adolescents. Am J Prev Med. 2016 Oct;51(4):513–7.

25. Smith JR, Novotny TE, Edland SD, Hofstetter CR, Lindsay SP, Al-Delaimy WK. Determinants of Hookah Use among High School Students. Nicotine Tob Res. 2011 Jul 1;13(7):565–72.

26. Nuñez M. Ley que prohibe el use de hookah en lugares públicos y privados. Ley núm. 16-19 2019.

27. Sandoval RC, Bacelar Gomes A, Roche M, Parra N, Armada F. Avances en el control del tabaco en la Región de las Américas 2020. Rev Panam Salud Pública. 2021 Aug 12;45:1.

28. World Health Organization. WHO Report on the Global Tobacco Epidemic, 2015: Raising Taxes on Tobacco [Internet]. Luxemburg; 2015. Available from: https://apps.who.int/iris/bitstream/handle/10665/178574/9789240694606_eng.pdf?sequence=1&isAllowed=y

29. Francis JA, Abramsohn EM, Park HY. Policy-driven tobacco control. Tob Control. 2010 Apr 1;19(Supplement 1):16–20.

30. Sheikh A, Vadera S, Ravey M, Lovatt G, Kelly G. A social norms approach to changing school children’s perceptions of tobacco usage. Health Educ. 2017 Oct 2;117(6):530–9.

_______________________________

¹ Oregon State University. ORCID: https://orcid.org/0000-0003-4874-9355, email: arredoce@oregonstate.edu

² Oregon State University. ORCID: https://orcid.org/0000-0002-4722-1041, email: KariLyn.Sakuma@oregonstate.edu