Changes in the Number of Smokers by State Before and after the US

Epidemic

Yiyang Tian

1,*

and Ruoyu Ying

Department of Letters and Science, University of California, Davis, U.S.A.

Department of Computing, University of Technology Sydney, Sydney, Australia

Keywords: COVID-19 Pandemic, Tobacco Use Trends, Descriptive Statistical Analysis, Regression Analysis.

Abstract: Against the backdrop of the global COVID-19 pandemic, significant changes have occurred in people's

lifestyles and behavioral patterns, including variations in tobacco use. This study employs descriptive

statistical analysis and regression analysis to explore trends in tobacco use among different age groups in the

United States before and during the pandemic, as well as the influencing factors. The research covers

descriptive statistics on tobacco use rates among individuals aged 12 - 17, 18 - 25, and 26+ for the periods

2018 - 2019 and 2021 - 2022, comparing changes and disparities across states and age groups. Regression

models are constructed to analyze the impact of time dummy variables, state, and age on tobacco use rates.

The findings indicate a decline in tobacco use rates across all age groups during the pandemic, with varying

degrees of reduction and distinct influencing factors among different age groups. This study provides evidence

for understanding the pandemic's impact on tobacco use behavior and offers insights for formulating more

effective tobacco control policies.

1 INTRODUCTION

In the field of global public health, tobacco use has

long been a critical issue of concern. Smoking is

closely linked to the development of severe diseases

such as lung cancer and cardiovascular conditions,

posing a persistent threat to population health.

According to the World Health Organization,

smoking causes over 8 million deaths annually

worldwide, including approximately 1.2 million non-

smokers who die from secondhand smoke exposure.

With increasing public health awareness and the

implementation of tobacco control policies globally,

smoking rates have shown a gradual decline over the

past few decades. However, individual smoking

behavior remains dynamically influenced by complex

social and environmental factors.

The COVID-19 pandemic, which emerged in

early 2020 as one of the most severe global public

health crises of the century, has profoundly disrupted

societal production and daily life. During the

pandemic, measures such as social distancing,

economic shutdowns, fear of infection, and abrupt

changes in daily routines created a unique high-stress

environment. In this context, smoking - a complex

habit with both physiological dependence and

psychological regulation functions - inevitably

experiences multifaceted influences. From a

psychological perspective, pandemic-induced anxiety

and depression may have driven some individuals to

increase smoking frequency as a coping mechanism.

A 2020 survey by the American Psychological

Association found that 40% of respondents reported

increased smoking during the early pandemic, with

the 18 - 34 age group showing a 52% increase.

However, from a behavioral restriction standpoint,

measures such as the closure of public spaces,

tightened tobacco retail controls, and reduced social

smoking opportunities due to remote work

objectively constrained smoking behavior. This dual

dynamic of "stress-driven increase" and

"environmentally constrained decrease" led to

significant individual variability in smoking behavior

during the pandemic.

The United States, as one of the hardest-hit

countries, exhibits notable regional heterogeneity

among states in terms of pandemic response

strategies, economic structures, demographic

distributions, and tobacco control policy strictness.

For instance, New York implemented strict

lockdowns and closed all indoor public spaces as

early as March 2020, while Texas lifted restrictions

Tian, Y. and Ying, R.

Changes in the Number of Smokers by State Before and after the US Epidemic.

DOI: 10.5220/0013860800004719

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 2nd International Conference on E-commerce and Modern Logistics (ICEML 2025), pages 779-785

ISBN: 978-989-758-775-7

779

relatively early in 2021. Such differences resulted in

varying levels of pandemic pressure, economic

disruption, and behavioral constraints across states,

leading to complex and diverse patterns in smoking

behavior changes. Some industrial states saw 15% -

20 % fluctuations in smoking rates due to soaring

unemployment and the closure of social venues,

while states with a high concentration of tech

industries, such as California, experienced smaller

changes (around 5%) owing to widespread remote

work and a younger demographic. Given this

complexity, existing research lacks a systematic

explanatory framework, necessitating in-depth

comparative analysis.

This study addresses the dual mechanisms of

pandemic impact on smoking behavior and the

significant socioeconomic and policy differences

across U.S. states. It aims to comprehensively analyze

dynamic changes in smoking rates across states from

2018 - 2019 (pre-pandemic) to 2021 - 2022 (during

and post-pandemic). The research focuses on three

core dimensions: (1) using panel data models to

precisely depict smoking rate trends across states by

age, gender, and education level; (2) employing

spatial econometric methods to identify regional

clustering effects in smoking behavior changes; and

(3) constructing structural equation models to explore

the interaction mechanisms among pandemic control

intensity, economic fluctuations, demographic

characteristics, and pre-existing tobacco control

policies. The study seeks to answer: Which states

experienced significant changes in smoking rates

during the pandemic? How do these changes

quantitatively relate to state-level policy variables

and socioeconomic indicators? What lasting effects

did temporary environmental changes during the

pandemic have on long-term smoking behavior

patterns?

Theoretical contributions of this study include

expanding traditional smoking behavior research,

which primarily focuses on individual psychological

factors, by incorporating macro-level public health

events into the analytical framework. Through multi-

level data integration, the study reveals the complex

pathways through which major social crises influence

addictive behaviors. The proposed theory enriches the

situational factors theory in health behavior studies,

offering new analytical dimensions for understanding

individual decision-making during emergencies.

Practically, the findings will inform state-specific

tobacco control policies in the U.S. For states with

significant increases in smoking rates (e.g., Rust Belt

industrial states), targeted interventions such as

psychological support for the unemployed and

community-based cessation programs are

recommended. For states with notable declines (e.g.,

West Coast regions), successful strategies like

"remote work smoking control" and "health

communication for younger demographics" can be

replicated elsewhere. Importantly, the regional

difference analysis framework developed in this

study can serve as a methodological reference for

other countries assessing smoking behavior changes

during public health crises and formulating tailored

intervention strategies, thereby contributing to multi-

objective collaborative governance systems.

2 LITERATURE REVIEW

2.1 Domestic Literature Review

Research on smoking behavior in China has evolved

in stages, reflecting shifting public health priorities.

Early studies focused on establishing the link between

smoking and diseases, laying a scientific foundation

for understanding smoking hazards. Liu Xiaoyan's

(2018) 10-year cohort study in Xuanwei City

provided large-scale evidence that smokers' lung

cancer mortality was 3.2 times higher than non-

smokers, solidifying smoking as a primary risk factor.

Xiao Junling's (2020) cross-sectional survey of

10,000+ residents in Jiayuguan City further

demonstrated that smokers faced 47% higher

respiratory disease risks and 32% higher

cardiovascular risks, establishing dose-response

relationships between smoking and chronic diseases.

Post-2010, studies began examining

sociodemographic disparities. Kang Guorong et al.

(2015) revealed urban-rural divides in smoking rates

(28.7% rural vs. 21.3% urban) and health literacy

gaps (22 percentage points), incorporating

socioeconomic status into smoking behavior models.

Li Shanpeng and Qi Fei (2018) highlighted policy

effects, showing that smoking bans reduced daily

cigarette consumption from 16.2 to 13.5 in Qingdao.

Li Zhongyou et al. (2018) tracked gender disparities

over 25 years in Guangxi, with male smoking rates

declining from 58.3% (1991) to 52.7% (2015) while

female rates remained low (3.2% - 4.1%).

Recent studies emphasize targeted interventions.

Lei Chunping (2012) identified peer influence

(coefficient 0.63) and family environment (0.48) as

key factors for youth smoking. Wang Zhao et al.

(2011) classified college smokers into social (37.2%)

and stress-relief (28.9%) types, informing tailored

campus programs. Zheng Bao (2011) found policy

perception gaps: smokers' support for public smoking

ICEML 2025 - International Conference on E-commerce and Modern Logistics

780

bans (68%) lagged behind non-smokers' (89%), but

72% desired cessation services, guiding local policy

adjustments.

Domestically, research has progressed from

biological foundations (disease links) to sociological

explanations (group difference) to public health

applications (precision strategies). While establishing

an "individual-behavior-policy" framework, gaps

remain in studying pandemic impacts and emerging

products like e-cigarettes.

2.2 International Literature Review

Global research employs multidisciplinary

approaches. Van Gucht et al. (2010) used behavioral

diaries to show how social contexts and moods affect

smoking patterns. Stang et al. (2000) quantified

smoking's contribution to COPD. Noonan et al.

(2015) mapped woodsmoke exposure risks using

GIS, while Hall (2004) analyzed smoke hazards in

fire disasters.

Haenszel et al. (1956) pioneered U.S. smoking

pattern analyses. Shopland et al. (1996) identified

higher smoking rates in Southern/Midwestern states,

linking regional culture to behavior. Jamal et al.

(2015) tracked 2005 - 2014 trends, showing faster

declines among men and variations by

race/education.

2.3 Literature Critique

Existing studies lack dynamic cross-regional

comparisons, especially regarding pandemic impacts.

International work underanalyzes state-level

differences during COVID-19, neglecting policy-

economic-psychological interactions.

Methodologically, traditional surveys dominate, with

limited causal inference. This study addresses these

gaps by integrating multi-source data and causal

methods to examine U.S. state-level smoking

changes, offering theoretical and practical

advancements for tobacco control.

3 METHODOLOGY

This study combines descriptive statistics and

regression analysis to investigate pre-/post-pandemic

tobacco use trends among U.S. age groups.

Descriptive statistics- compare 2018 - 2019 and

2021 - 2022 tobacco use rates across ages (12 - 17, 18

- 25, 26+) and states, calculating changes to identify

pandemic effects.

Regression analysis -models tobacco use rates

(dependent variable) against time dummies (0=pre-

pandemic, 1=pandemic), controlling for state and

age. Coefficient significance tests determine

pandemic impacts, while R2 evaluates model fit. This

reveals underlying mechanisms behind trend shifts.

4 RESULTS

4.1 Descriptive Statistical Analyses

This section explores changes in tobacco use trends

before and after the epidemic with the help of

descriptive statistical analyses of data on tobacco use

rates by age from 2018 to 2019 and from 2021 to

2022.

Figure 1: Changes in tobacco use among 12-17 year olds

before and after the epidemic

Figure 1 shows that in the 12- to 17-year-old age

group, the overall trend in the data shows a general

decline in tobacco use in this age group during the

epidemic, with estimates of tobacco use in Alabama

ranging from 20 in 2018-2019 to 5 in 2021-2022, a

decline of 15, and in California from 77 to 42, a

decline of 35. California, on the other hand, dropped

from 77 to 42, a decline of 35. Such a downward trend

was demonstrated in several states, suggesting that

tobacco use in this age group was controlled to some

extent during the epidemic. In terms of regional

distribution, there was a relatively large difference in

tobacco use rates between states, with several states

like Alaska having a relatively low estimated value of

tobacco use before the epidemic of only 4, while

California had a relatively high rate of 77. This

difference persisted during the epidemic, reflecting

the different foundations of youth tobacco prevention

and control, as well as their effectiveness, in different

regions.

Changes in the Number of Smokers by State Before and after the US Epidemic

781

Figure 2: Changes in tobacco use among 18-25 year 25-

year-olds before and after the epidemic

As can be seen in Figure 2, in the 18 to 25 age

group, tobacco use rates in this age group showed the

same decline during the epidemic, for example, in

Alabama, where the rate of tobacco use decreased by

96 from 155 in 2018-2019 to 59 in 2021-2022, and in

Texas, where the rate decreased from 756 to 299, a

decline of 457, with different There is some variation

in usage rates between states. Before the epidemic,

Florida had a tobacco use rate of 399 compared to 23

in Hawaii, and during the epidemic, it dropped to 164

in Florida and 17 in Hawaii, and this difference

between states has a key impact on the overall

tobacco use trend, with this age group experiencing

limited socialisation and lifestyle changes during the

epidemic, which may be a key factor contributing to

the decline in tobacco use rates.

Figure 3 shows that in the 26 and over age group,

there was a decline in tobacco use during the

epidemic for people aged 26 and over, for example,

Alabama dropped from 1012 to 762, a decrease of

250, and New York dropped from 2469 to 2183, a

decrease of 286, although the overall trend was

downward, although the differences in use between

different states were more striking, with states like

Texas had a high estimate of tobacco use of 4,015

before the epidemic, while Vermont had only 97, and

during the epidemic, Texas dropped to 3,293, and

Vermont dropped to 86, a difference that reflects the

impact of different states' social environments, levels

of economic development, and tobacco control

policies on the tobacco use behaviors of the adult

population.

Comparison of the three age groups shows that

tobacco use declined in all age groups during the

epidemic,

but the magnitude of the decline and the

Figure 3: Changes in tobacco use among people aged 26+

before and after the epidemic

trend of change varied, with a more pronounced

decline in the 12- to 17-year-old age group, which

may be related to the strengthening of health

promotion and education in schools during the

epidemic, and the restriction of the scope of activities

of students, etc., and a larger decline in the 18- to 25-

year-old age group, where changes in lifestyle and

social patterns had a greater impact on their tobacco

use behaviour. The decline in the 18-25 age group is

also relatively large, with changes in the pace of life

and socialisation patterns having a greater impact on

their tobacco use behaviour. 26 and older age groups,

although also declining, are experiencing a smaller

decline due to the relative stability of this group's

lifestyles. Economically developed regions may see

more pronounced declines in tobacco use because of

strict prevention and control measures and high levels

of health awareness, while states with higher levels of

tobacco industry dependence have seen relatively

smaller declines.

4.2 Regression Results

4.2.1 12-17 Year

Table 1: Tobacco use among 12-17 year olds

Regression Statistics Value

Multiple R 0.8417

R Square 0.7085

Adjusted R Square 0.6646

Standard Error 7.0690

Observations 56

ICEML 2025 - International Conference on E-commerce and Modern Logistics

782

Table 2: Estimated regression coefficients and test results for tobacco use among 12-17 year olds

Coefficients

Standard

Error

t Stat P-value

Lower 95

per cent

Upper 95

per cent

Lower 95.0

per cent

Upper 95.0

per cent

Intercept 3.7394 6.3194 1.0284 16.1349 31.1270 16.1349 31.1270

After 4.4763 -3.6222 0.0007 -25.2310 -7.1976 -25.2310 -7.1976

Treated 0.2400 -0.1377 0.8906 -0.5152 0.4491 -0.5152 0.4491

Interaction Term

(After*Treated)

0.2864 0.0208 0.9835 -0.5667 0.5786 -0.5667 0.5786

The results of the analyses in Table 1 and Table 2

show that, after controlling for the effects of the

different states, the coefficient of the time dummy

variable "After" in the 12 to 17 year olds shows a

specific characteristic, with a coefficient of -16.2143,

which means that in 2021 to 2022, the rate of tobacco

use in the 12 to 17 year olds declined in comparison

to the rate in 2018 to 2019. 2018 to 2019 is

decreasing, as can be seen from the p-value of 0.0007;

the coefficient is statistically significant, indicating

that the epidemic has had a relatively very prominent

effect on the use of tobacco in this age group. During

the epidemic, this age group's exposure to tobacco

may have been reduced due to the restriction of the

scope of activities of young people and the

strengthening of health education in schools, which

led to a decrease in the rate of tobacco use. Moreover,

Treated and Interaction Term are not significant in

Table 1 and Table 2, suggesting that "state" as a

control variable may have a weaker effect on tobacco

use, and the main effect is due to the time dummy

variable "After". After".

4.2.2 18-25 Years

Table 3: Tobacco use among 18-25 year olds

Regression Statistics Value

Multiple R 0.9733

R Square 0.9474

Adjusted R Square 0.9347

Standard Error 23.9077

Observations 56

Table 4: Estimated regression coefficients and test results for tobacco use among 18-25 year olds

Coefficients

Coefficient

Values

Standard

Error

t Stat P-value

Lower 95

per cent

Upper 95

per cent

Lower 95.0

per cent

Upper 95.0

per cent

Intercept 148.8768 10.8210 13.6836 2.1829 127.0232 170.7304 127.0232 170.7303

After -133.8214 12.9327 -10.3475 3.4105 -159.9722 -107.6706 -159.9723 -107.6705

Treated 0.369643 0.61817 0.5980 0.5523 -0.8662 1.6055 -0.8662 1.6055

Interaction Term

(After*Treated)

-0.5982 0.7380 -0.8106 0.4218 -2.0792 0.8828 -2.0792 0.8828

As can be seen from Table 3 and Table 4, the

coefficient of the time dummy variable "After" is -

133.8214 for the age group of 18 to 25 years old,

which indicates that during the epidemic period, the

tobacco use rate of this age group decreased

compared with that before the epidemic. In addition,

the p-value of this coefficient is 3.4105, which is

much smaller than the common level of 0.05, and this

change is statistically significant, which confirms that

the epidemic has had a strong impact on the tobacco

use behaviour of people aged 17 to 25. During the

epidemic, there was a significant change in the way

18-25 year olds socialised, with an increase in online

activities and a decrease in offline gatherings, which

may have led to a corresponding decrease in tobacco

use behaviours that were often seen in social

situations, resulting in a significant decrease in

tobacco use.

Changes in the Number of Smokers by State Before and after the US Epidemic

783

4.2.3 26 Years and Over

Table 5: Tobacco use among people aged 26 and over

Regression Statistics Value

Multiple R 0.9733

R Square 0.9474

Adjusted R Square 0.9347

Standard Error 23.9077

Observations 56

Table 6: Estimated regression coefficients and test results for tobacco use among 26-year-olds

Coefficients

Coefficient

Values

Standard

Error

t Stat P-value

Lower 95

per cent

Upper 95

per cent

Lower 95.0

per cent

Upper 95.0

per cent

Intercept 148.8769 10.8800 13.6836 2.1828 127.0232 170.7303 127.0232 170.7303

After -133.8214 12.9327 -10.3476 3.4105 -159.972 -107.6705 -159.972 -107.6706

Treated 0.3696 0.6181 0.5980 0.5523 -0.8662 1.6055 -0.8662 1.6055

Interaction Term

(After*Treated)

-0.5982 0.7380 -0.8106 0.4218 -2.0792 0.8827 -2.0792 0.8828

The analyses in Table 5 and Table 6 show that the

coefficient of the time dummy variable "After" is -

133.8214 for the age group of 26 and above, which

also shows a decrease in tobacco use during the

epidemic, and its p-value is 3.4105, which means that

the epidemic has also had a certain impact on this age

group. The coefficient of "After" is -133.8214, which

also shows a decrease in tobacco use during the

epidemic, with a p-value of 3.4105. During the

epidemic, people in this age group may have reduced

their tobacco consumption due to many factors such

as changes in their work patterns and the

implementation of controls in public places, which

led to a decrease in the rate of tobacco use. Although

the social and life patterns of people aged 26 and

above are relatively more stable, the changes in the

overall social environment brought about by the

epidemic still have an undeniable impact on their

tobacco use behaviour. Although the social and

lifestyle patterns of people aged 26 and over are more

stable, the changes in the social environment brought

about by the epidemic have had a significant impact

on their tobacco use behaviour.

Regression analyses clearly show that the

epidemic had a strong impact on tobacco use trends

in different age groups, and that tobacco use declined

in all age groups during the epidemic. Although the

magnitude of the decline and the factors influencing

it varied by age group, it is undeniable that the

epidemic, as a powerful external variable, reshaped

people's lives and consumption patterns, and

ultimately had a negligible effect on tobacco use

behaviour. use behaviours.

5 CONCLUSION

This study focuses on the changes in tobacco use

trends among different age groups before and after

the epidemic, using regression analyses to deeply

analyse the impact of time variables and factors such

as different states and ages on tobacco use, and the

results show that the epidemic had an impact on the

use of tobacco among people aged 12-17, 17-25, and

26 and older, and that tobacco use among all age

groups showed a declining trend during the epidemic.

The results show that the epidemic had an impact on

tobacco use among 12 to 17, 17 to 25, and 26-year-

olds and older, and that tobacco use among all age

groups decreased during the epidemic.

This result gives insights into many aspects. From

the perspective of public health, the preventive and

control measures implemented during the epidemic,

such as social restriction and control of public places,

have to a certain extent limited the access to and use

of tobacco, resulting in a reduction in the rate of

tobacco use among people of all ages, which provides

new ideas and references for the subsequent

formulation of tobacco control policies, such as

making reference to the effective control model

during the epidemic, strengthening the regulation of

youth smoking behaviour, and restricting smoking

ICEML 2025 - International Conference on E-commerce and Modern Logistics

784

behaviour in public places. This provides new ideas

and references for the subsequent formulation of

tobacco control policies, for example, by referring to

the control model that worked well during the

epidemic, strengthening the regulation of youth

smoking behaviour, and restricting smoking

behaviour in public places.

However, it is also important to note the

limitations of this study in that the model, although it

takes into account factors such as time and state, may

have an impact on tobacco use as well as other

variables that were not included, such as changes in

economic conditions, shifts in cultural attitudes, etc.,

and that the data from the study only covered two

specific periods, making it difficult to fully reflect

changes in tobacco use trends over time.

Future research can broaden the scope of data,

incorporate more influencing factors, and construct a

more complete model to gain a more comprehensive

understanding of the changing patterns of tobacco use

behaviours, provide a solid theoretical basis and data

support for the development of more effective

tobacco control strategies, and promote public health.

AUTHORS CONTRIBUTION

All the authors contributed equally and their names

were listed in alphabetical order.

REFERENCES

Fan, Z., Xiong, H., & Wang, Z., 2011. Survey and research

on smoking among university students in Kunming,

Yunnan Province. Social Sciences (New Theory),

26(03), 154-155.

Haenszel, W., Miller, H. P., & Shimkin, M. B., 1956.

Tobacco smoking patterns in the United States. US

Government Printing Office.

Hall, J. R., 2004. How many people are exposed to sublethal

fire smoke? Fire Technology, 40, 101-116.

He, X., 2021. Relationship between passive smoking and

respiratory symptoms among rural women in typical

counties and districts in the Huaihe River Basin and

analysis of influencing factors. Lanzhou University.

Jamal, A., Homa, D. M., O'Connor, E., et al., 2015. Current

cigarette smoking among adults-United States, 2005-

2014. MMWR Morb Mortal Wkly Rep, 64(44), 1233-

1240.

Kang, G., Lu, P., Qian, G., et al., 2015. Study on tobacco

prevalence among urban and rural residents aged 15-69

years in Gansu Province. China Health Education,

31(10), 919-924.

Lei, C., 2012. Study on the current situation of smoking

behaviour and influencing factors of students in a

medical college in Chongqing. Third Military Medical

University.

Li, S., & Qi, F., 2014. 2014 Qingdao Adult Tobacco

Epidemic Survey Report. People's Health Press, 169.

Li, Z., Tang, Z., Fang, Z., et al., 2018. Analysis of smoking

status and trend of change among residents aged 18

years and above in 6 cities (counties) of Guangxi from

1991 to 2015. China Health Education, 34(03), 195-

199.

Liu, X., 2018. Estimation of prevalence characteristics of

lung cancer deaths among smokers and non-smokers

aged 30-69 years in six townships of Xuanwei City,

2014-2016. Peking Union Medical College.

Noonan, C. W., Ward, T. J., & Semmens, E. O., 2015.

Estimating the number of vulnerable people in the

United States exposed to residential wood smoke.

Environmental Health Perspectives, 123(2), A30-A30.

Shopland, D. R., Hartman, A. M., Gibson, J. T., et al., 1996.

Cigarette smoking among US adults by state and

region: estimates from the current population survey.

JNCI: Journal of the National Cancer Institute, 88(23),

1748-1758.

Stang, P., Lydick, E., Silberman, C., et al., 2000. The

prevalence of COPD: using smoking rates to estimate

disease frequency in the general population. Chest,

117(5), 354S-359S.

Van Gucht, D., Van den Bergh, O., Beckers, T., et al., 2010.

Smoking behaviour in context: where and when do

people smoke? Journal of behaviour therapy and

experimental psychiatry, 41(2), 172-177.

Xiao, J., 2020. Survey and analysis of community residents'

hygiene and health status in Jiayuguan City. Lanzhou

University.

Xu, L., Liu, T., Li, L., et al., 2014. Study on the prevalence

of tobacco among adults in Guizhou Province. China

Chronic Disease Prevention and Control, 22(06), 675-

678. doi:10.16386/j.cjpccd.issn.1004-

6194.2014.06.008.

Zheng, B., 2011. A survey of smoking attitudes and related

behaviours among key populations for tobacco control

in Henan Province. Zhengzhou University.

Changes in the Number of Smokers by State Before and after the US Epidemic

785