© The Author 2014. Published by Oxford University Press on behalf of the Society for Research on Nicotine and Tobacco. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

doi:10.1093/ntr/ntt223

Brief report

impact of Changing U.S. Demographics on the Decline in Smoking prevalence, 1980–2010

Jamie Tam MPH1, Kenneth E. Warner PhD1, Brenda W. Gillespie PhD2, John A. Gillespie PhD3

1Department of Health Management and Policy, School of Public Health, University of Michigan, Ann Arbor, MI; 2Department of Biostatistics, School of Public Health, University of Michigan, Ann Arbor, MI; 3Department of Mathematics and Statistics, University of Michigan-Dearborn, Dearborn, MI

Corresponding Author: Jamie Tam, MPH, Department of Health Management and Policy, School of Public Health, University of Michigan, 1415 Washington Heights, Ann Arbor, MI 48109-2029, USA. Fax: 734-764-4338; E-mail: jamietam@umich.edu

Received August 25, 2013; accepted December 3, 2013

ABStrACt

introduction: U.S. smoking prevalence has been declining over the last several decades. During this time, the population has also experienced changes in its demographic composition, as Americans are living longer and becoming increasingly racially and ethnically diverse. Since smoking rates vary across age and race/ethnicity groups, demographics alone could contribute to changes in smoking prevalence among the general population. We examined the effect of changing age and race/ethnicity distri-butions on total smoking prevalence from 1980 to 2010.

Methods: Using the National Health Interview Survey weighting scheme, we applied the distribution of smokers across age and race/ethnicity categories for the years 1980 and 2010 to the distribution of adults in those categories for both years. The total number of smokers was summed to determine resulting smoking prevalence.

results: The combined effect of aging and the changing racial/ethnic composition of the U.S. population has contributed 2.1% points to the decline in smoking prevalence. If the age and racial/ethnic demographic composition had not changed since 1980, smoking prevalence would have been 21.3% in 2010 (with rounding)—statistically significantly higher than the reported 19.3%. Of the 3 demographic factors we considered (age, race, and ethnicity), ethnicity—specifically the rising share of Hispanics in the population—is the most important contributor to declines in smoking.

Conclusions: Our changing demographics have had an impact on smoking prevalence over the last 3 decades. Future declines in smoking may be driven even more by the aging of the population and increasing racial and ethnic diversity.

introDUCtion

Declines in U.S. smoking rates have been attributed to success-ful tobacco control initiatives, including tobacco tax increases, smoke-free workplace legislation, and public education cam-paigns (Institute of Medicine, 2007). Since 1980, adult smok-ing prevalence has fallen from 33.2% to 18.0% (Centers for Disease Control and Prevention, 2012b; National Center for Health Statistics, 1980; Schiller, Ward, Freeman, & Peregoy, 2013). However, tobacco control initiatives may not be the only driver behind decreasing smoking prevalence.

One possible explanation for declines in smoking rates could be the changing distribution of individuals across age, race, and ethnicity groups. The U.S. population is aging, and a rising share of the population is non-White and/or Hispanic (United States Census Bureau, 2012). Each of these demo-graphic factors is associated with differences in smoking prevalence. Thus, shifts in demographic composition over time could influence total smoking prevalence.

A previous study demonstrated that California’s lower smoking rates are at least partially a consequence of changing demographics in the state (Warner, Mendez, & Alshanqeety, 2008). In this analysis, we examine whether underlying demo-graphic changes could similarly explain some of the decline in U.S. smoking prevalence from 1980 to 2010.

MethoDS

We relied on the National Health Interview Survey (NHIS), a commonly accepted source for data on smoking for decades. 1980 was the first year the NHIS allowed respondents to select their race from a list. In previous years, race was determined by interviewer observation. Four single-race categories were con-sidered: White, Black, Asian and Pacific Islander (API), and American Indian and Alaska Native (AIAN). Individuals of other, multiple, or unknown racial background (1.6% and 1.8% of the sample in 1980 and 2010, respectively) were excluded

Advance Access publication January 8, 2014

864

nicotine & tobacco research, Volume 16, number 6 (June 2014) 864–866 at U

niversity of Windsor on July 11, 2014

http://ntr.oxfordjournals.org/D

ownloaded from

nicotine & tobacco research

from the analysis. Hispanic or non-Hispanic origin status was considered within each race category, resulting in eight total racial and ethnic origin groups.

Fourteen age groups were considered in 5-year ranges, with the exception of ages 18–24 and ages ≥85. The size of the adult population within each of the 8 race and 14 age categories was determined using the NHIS weighting scheme. NHIS person-level weights adjusted for sex, age, and race/ethnicity using Census control totals. NHIS weights are derived from 1970 and 2000 Census projected population estimates for the years 1980 and 2010, respectively (using current-year Census weighting produced nearly identical results to those presented here). The percent of smokers for each group was then determined using NHIS data for the years 1980 and 2010. Individuals who self-identified as “present” or “occasional” smokers were consid-ered current smokers in 1980, while “every day” or “someday” smokers were considered current smokers in 2010.

The percentage of smokers by race/ethnicity and age for 1980 and 2010 were first applied separately to the distribu-tion of total adults in those categories for both years. Total smokers were summed and used to estimate U.S.  smoking prevalence. We also applied the smoker age distribution by race and by race/ethnicity to total adults in those categories. The difference between the smoking prevalence in 2010 and a hypothetical 2010 prevalence given the 1980 demographic composition is reported prior to rounding. Statistical signifi-cance was calculated using an upper bound on the variance (further explanation is available at https://sites.google.com/site/demographicssmoking/).

reSUltS

Table 1 shows the calculated 2010 smoking prevalence using the following demographic distributions from 2010 and 1980: age, age by race, race/ethnicity, and age by race/ethnicity. When calculated with each 2010 demographic distribution, total adult smoking prevalence was 19.3%, which matches reported 2010 NHIS prevalence (National Center for Health Statistics, 2010). Applying the age distribution from 1980 increases the smoking prevalence by 0.3% points to 19.7%; this change was not statis-tically significant (p = .30) (the difference appears to be .4 but is .3 due to rounding). However, applying the age by race dis-tribution from 1980 results in a significant 0.8% point increase in smoking (p = .02). Smoking prevalence in 2010 increases by 1.2% points to 20.5% when weighted with the race/ethnicity demographics of 1980 (p < .001). Applying the 1980 distribu-tion of age by race/ethnicity increases smoking prevalence in 2010 even further by 2.1% points to 21.3% (p < .001). Detailed tables for each of these calculations are available online at https://sites.google.com/site/demographicssmoking/.

ConClUSionS

We find that changing demographics in the United States have contributed to the decrease in adult smoking prevalence from 1980 to 2010, particularly the rising share of Hispanics in the population with their lower-than-average smoking prevalence (Centers for Disease Control and Prevention, 2012a). However, the demographic shifts explain only a very small part of the overall decline. Tobacco control interventions, and changing social norms, almost certainly remain the dominant drivers behind decreasing smoking.

The demographic impact of population aging on smoking from 1980 to 2010 was minor. However, current projections indicate that the over-65 population—the group of adults with the lowest smoking prevalence—will increase from 13% to 19% of the population by 2030 (Vincent & Velkoff, 2010). Thus, population aging alone is likely to have an impact on national smoking prevalence. Similarly, if Hispanics continue to smoke at lower rates than the rest of the population, pro-jected growth in the Hispanic share of the population suggests a further decline in smoking prevalence. Over the next few decades, the implications of our shifting demographics will need to be factored into evaluations of the impact of tobacco control.

DeClArAtion of intereStS

None declared.

referenCeS

Centers for Disease Control and Prevention. (2012a). Current cigarette smoking among adults - United States, 2011. Morbidity and Mortality Weekly Report, 61, 889–894. Retrieved June 11, 2013, from http://www.cdc.gov/mmwr/pdf/wk/mm6144.pdf

Centers for Disease Control and Prevention. (2012b). Trends in current cigarette smoking among high school students and adults, United States, 1965–2011. Retrieved July 7, 2013, from http://www.cdc.gov/tobacco/data_statistics/tables/trends/ cig_smoking/index.htm

Institute of Medicine. (2007). Ending the tobacco problem: A blueprint for the nation. Washington, DC: The National Academies Press.

National Center for Health Statistics. (1980). National Health Interview Survey, 1980. Public-use data file and documen-tation. Retrieved March 5, 2013 from ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/

National Center for Health Statistics. (2010). National Health Interview Survey, 2010. Public-use data file and documen-tation. Retrieved March 5, 2013 from ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/

table 1. U.S. Smoking Prevalence Based on Demographic Distributions in 1980 and 2010

2010 smoking prevalence weighted by Age (%) Age–race (%) Race–ethnicity (%) Age–race–ethnicity (%)

2010 demographic distribution 19.3 19.3 19.3 19.31980 demographic distribution 19.7 20.0* 20.5** 21.3**Change in smoking prevalence 0.3 0.8 1.2 2.1

Note. Changes in smoking prevalence calculated prior to rounding. Data from National Health Interview Survey (NHIS) 1980 and 2010.*p < .05. **p < .001.

865

at University of W

indsor on July 11, 2014http://ntr.oxfordjournals.org/

Dow

nloaded from

Demographic impact on smoking prevalence

Schiller, J. S., Ward, B. W., Freeman, G., & Peregoy J. A. (2013). Early release of selected estimates based on data from the 2012 National Health Interview Survey. National Center for Health Statistics. Retrieved June 20, 2013, from http://www.cdc.gov/nchs/nhis.htm

United States Census Bureau. (2012, December 12). U.S. Census Bureau projections show a slower growing, older, more diverse nation a half century from now. Retrieved May 31, 2012, from http://www.census.gov/newsroom/releases/archives/population/cb12-243.html

Vincent, G. K., & Velkoff, V. A. (2010). The next four dec-ades: The older population in the U.S. 2010 to 2050, popula-tion estimates and projections. Current Population Reports. Retrieved May 31, 2013, from http://www.census.gov/prod/2010pubs/p25-1138.pdf

Warner, K. E., Mendez, D., & Alshanqeety, O. (2008). Tobacco control success versus demographic destiny: Examining the causes of the low smoking prevalence in California. American Journal of Public Health, 98, 268–269. doi:10.2105/AJPH.2007.112318

866

at University of W

indsor on July 11, 2014http://ntr.oxfordjournals.org/

Dow

nloaded from