Associations between smoking prevalence, stages of change, cigarette consumption, and quit attempts across the United States

Associations between smoking prevalence, stages of change, cigarette consumption, and quit attempts across the United States

Preventive Medicine 38 (2004) 369 – 373 www.elsevier.com/locate/ypmed Associations between smoking prevalence, stages of change, cigarette consumptio...

104KB Sizes 0 Downloads 97 Views

Preventive Medicine 38 (2004) 369 – 373 www.elsevier.com/locate/ypmed

Associations between smoking prevalence, stages of change, cigarette consumption, and quit attempts across the United States Jean-Francßois Etter, Ph.D., M.P.H. * Institute of Social and Preventive Medicine, University of Geneva, Geneva 4, Switzerland

Abstract Background. We tested whether across the 50 U.S. States, smoking prevalence was associated with smoking behavior among smokers. Methods. We used published data on smoking prevalence, cigarette consumption, and quit attempts by State from the Behavioral Risk Factor Surveillance System (BRFSS) for 1993, 1996, and 1999, and published data on the distribution of smokers by stage of change in each State from the Current Population Survey (CPS). Data were weighted for the population size in each State. Results. For 1993, we found no statistically significant association between smoking prevalence and stages of change. For 1996 and 1999, across the 50 States, each additional percentage point in smoking prevalence was associated with, respectively, 0.62% and 1.19% more smokers in the ‘‘precontemplation’’ stage, that is, smokers not intending to quit smoking (both P < 0.01), with, respectively, 0.82% and 1.21% more ‘‘heavy’’ smokers who smoked 21 – 40 cigarettes per day (both P V 0.001), and with, respectively, 1.12% and 0.81% fewer daily smokers who quit smoking for 1+ days in the previous year (both P V 0.04). Conclusions. Across the 50 States, a higher prevalence of cigarette smoking was associated with a lower motivation to quit smoking, fewer quit attempts, and higher cigarette consumption among smokers. D 2003 The Institute For Cancer Prevention and Elsevier Inc. All rights reserved. Keywords: Smoking; Prevalence; Tobacco cessation; Surveillance

Introduction The concept of ‘‘stage of change’’ is a core element of the Transtheoretical model of behavior change and is widely used in smoking prevention and cessation interventions [1]. In this model, smokers are classified in three stages of change (‘‘precontemplation’’, ‘‘contemplation’’, and ‘‘preparation’’) according to their level of motivation to quit smoking and their recent attempts to quit [2]. In the United States in 1998 –1999, 59% of daily smokers were in the ‘‘precontemplation’’ stage, 32% were in the ‘‘contemplation’’ stage, and 9% were in the ‘‘preparation’’ stage [3]. European samples may include more smokers in the ‘‘precontemplation’’ stage and fewer in the ‘‘contemplation’’ and ‘‘preparation’’ stages [4]. The Transtheoretical model of change asserts that before they are able to successfully quit smoking; smokers have to

* Institute of Social and Preventive Medicine, University of Geneva, CMU, 1 rue Michel-Servet, CH-1211 Geneva 4, Switzerland. Fax: +41-22379-59-12. E-mail address: [email protected].

progress from ‘‘precontemplation’’ to ‘‘contemplation’’ and ‘‘preparation’’ [1,2]. Conversely, smokers who live in societies where there are more anti-smoking policies may feel more pressure to quit smoking and move therefore to more advanced stages of change. Both mechanisms suggest that the distribution of smokers by stage of change should be more favorable in societies where smoking prevalence is lower. Previous research based on nine samples from five countries (one sample each from Finland, Netherlands, and Spain, two samples from Switzerland, four samples from the United States) showed that across these countries, a higher prevalence of cigarette smoking was associated with a higher proportion of smokers in the ‘‘precontemplation’’ stage of change [4]. However, this study found no such association within European samples or within U.S. samples, probably because there were too few data points. Thus, associations between smoking prevalence and the distribution of smokers by stage of change, based on ecological data, have not yet been reported within a specific country. Other aspects of smoking behavior could also be associated with smoking prevalence, including cigarette consump-

0091-7435/$ - see front matter D 2003 The Institute For Cancer Prevention and Elsevier Inc. All rights reserved. doi:10.1016/j.ypmed.2003.11.006

370

J.-F. Etter / Preventive Medicine 38 (2004) 369–373

tion among smokers, and the frequency of attempts to quit smoking. A previous study showed that across six countries (Austria, Denmark, Finland, France, Poland, and the United Sates), a lower smoking prevalence was associated with higher nicotine dependence among the remaining smokers, which was interpreted as the result of selective quitting of the less dependent smokers in countries where there are more active tobacco control policies [5]. In this situation, the remaining smokers would be ‘‘hard core’’ smokers who could also be less motivated to quit. In this situation, the distribution of smokers by ‘‘stage of change’’ could, in theory, be less favorable in countries where smoking prevalence is low. The aim of this study was to assess whether smoking prevalence was associated with stages of change, cigarette consumption, and quit attempts within the United States, using previously published data.

next 30 days and had made one 24-h quit attempt in the past year [2,3]. Smokers who planned to stop within 30 days but had not made a 24-h quit attempt in the past year were classified in the ‘‘contemplation’’ stage [3]. Analyses

Methods

We used linear regression models to assess the associations between smoking prevalence, cigarette consumption, the proportion of smokers who made a quit attempt, and the proportion of smokers in the ‘‘precontemplation’’ and ‘‘preparation’’ stages of change, across the 50 States and the District of Columbia. We repeated these analyses for each available year. Coefficients from linear regression models indicate the change in the explained variable (e.g., proportion of smokers in each stage of change), for each additional percentage point in the explanatory variable (e.g., smoking prevalence). We used multivariate linear regression models to assess whether these associations were maintained after adjustment for available variables.

Data sources

Change over time

We used data from the Behavioral Risk Factor Surveillance System (BRFSS) on smoking prevalence, for each of the 50 States and the District of Columbia, for the years 1993 (n = 102,263), 1996 (n = 124,085) and 1999 (n = 159,989) [6,7,8]. For 1996 and 1999, we used BRFSS data on the proportion of ‘‘heavy’’ daily smokers who smoked 21 –40 cigarettes per day and on the proportion of daily smokers who quit smoking for 1 day or longer during the past 12 months (we could not obtain these data for 1993) [9]. We used data on the distribution of daily smokers by ‘‘stage of change’’ in each State, from the Tobacco Use Supplement (TUS) of the Current Population Survey, for the periods 1992– 1993 (n = 39,706), 1995– 1996 (n = 34,865), and 1998 –1999 (n = 30,153) [3]. The BRFSS is a state-based, random-digit telephone survey of the non-institutionalized U.S. population aged z 18 years [7]. The Current Population Survey (CPS) is conducted monthly by the Census Bureau in the civilian noninstitutionalized U.S. population ages 15 years or older [10], but only respondents ages 18 and older were included in the reported data [3]. A 40-item Tobacco Use Supplement (TUS) to the CPS was developed by staff at the National Cancer Institute and included in the CPS in September 1992, January 1993, and May 1993, again in September 1995, January 1996, and May 1996, and again in September 1998, January 1999, and May 1999 [3]. The CPS is based on probability samples and response rates for the TUS averaged 85%. Stages of change were assessed only in daily smokers, who were classified in the ‘‘precontemplation’’ stage if they were not seriously considering stopping smoking within the next 6 months, in the ‘‘contemplation’’ stage if they were seriously considering stopping within the next 6 months, and in the ‘‘preparation’’ stage if they were planning to stop within the

We used linear regression models to assess whether shifts in the proportion of smokers in the ‘‘precontemplation’’ stage between 1993 and 1996 were associated with concomitant shifts in smoking prevalence between 1993 and 1996, or with subsequent shifts in smoking prevalence between 1996 and 1999. We also examined whether changes in smoking prevalence between 1993 and 1996 were associated with subsequent shifts between 1996 and 1999 in stages of change, in the proportion of daily smokers who smoked 21– 40 cigarettes/day, and in the proportion of daily smokers who quit for 1+ days during the past year. In all analyses, data were weighted for the size of the population in each State in 1995, using data published by the Bureau of the Census [11]. We used a standardized weight variable (mean = 0 and standard deviation = 1).

Results Prevalence of current cigarette smoking in the US remained stable at 23% between 1993 and 1999 [9]. The proportion of smokers in the ‘‘precontemplation’’ stage of change was 59% in 1993 and 1999, and 64% in 1996 [3]. The proportion of ‘‘heavy’’ daily smokers who smoked 21– 40 cigarettes/day was 19% in 1996 and 17% in 1999. The proportion of daily smokers who quit smoking for 1+ days in the previous year was 45% in 1996 and 51% in 1999 [9]. Smoking prevalence and stages of change There was no statistically significant association between smoking prevalence and the distribution of smokers by stage

J.-F. Etter / Preventive Medicine 38 (2004) 369–373 Table 1 Associations between the prevalence of cigarette smoking and the proportion of daily smokers in the ‘‘precontemplation’’ and ‘‘preparation’’ stages of change across the 50 States and the District of Columbia, USA, 1992 – 1999 Dependent variables

Percent smokers in the ‘‘precontemplation’’ stage in 1992 – 1993 or 1995 – 1996, or 1998 – 1999

Independent variables

Regression R-square P Regression R-square P coefficienta value coefficienta value

Smoking 0.36 prevalence in 1993 0.62 Smoking prevalence in 1996 Smoking 1.19 prevalence in 1999

Percent smokers in the ‘‘preparation’’ stage in 1992 – 1993, or 1995 – 1996, or 1998 – 1999

0.18

0.089

0.08

0.08

0.266

0.38

0.009

0.32

0.55

0.001

0.51

0.001

0.37

0.38

0.008

a Coefficients from linear regression models indicate the change in the percentage of smokers in each stage, for a 1% increase in smoking prevalence. Data were weighted for the size of the population in each State.

of change in 1993 (Table 1). However, in 1996 and 1999, across the 50 States and the District of Columbia, each additional percentage point in smoking prevalence was associated with, respectively, 0.62% and 1.19% more smokers in the ‘‘precontemplation’’ stage (both P < 0.01), and with, respectively, 0.32% and 0.37% fewer smokers in the ‘‘preparation’’ stage (both P < 0.01).

371

Smoking prevalence and quit attempts In 1996 and 1999, across the 50 States and the District of Columbia, each additional percentage point in smoking prevalence was associated with, respectively, 1.12% and 0.81% fewer daily smokers who had quit smoking for 1+ days during the previous 12 months (both P V 0.025). Cigarette consumption among daily smokers and stages of change In 1996 and 1999, across the 50 States and the District of Columbia, each additional percentage point in the proportion of ‘‘heavy’’ daily smokers who smoked 21 –40 cigarettes/day was associated with, respectively, 0.58% and 0.86% more smokers in the ‘‘precontemplation’’ stage of change (both P V 0.007), and with, respectively, 0.21% and 0.30% fewer smokers in the ‘‘preparation’’ stage (both P V 0.03) (Table 2). Quit attempts among daily smokers and stages of change In 1996 and 1999, across the 50 States and the District of Columbia, each additional percentage point in the proportion of daily smokers who had quit smoking for 1+ days during the past year was associated with, respectively, 0.32% and 0.52% fewer smokers in the ‘‘precontemplation’’ stage (both P V 0.04), and with, respectively, 0.17% ( P < 0.001) and 0.16% ( P = 0.08) more smokers in the ‘‘preparation’’ stage (Table 3). Change over time

Smoking prevalence and cigarette consumption among daily smokers In 1996 and 1999, across the 50 States and the District of Columbia, each additional percentage point in smoking prevalence was associated with, respectively, 0.82% and 1.21% more ‘‘heavy’’ daily smokers who smoked 21 –40 cigarettes/day (both P V 0.001).

Changes in smoking prevalence between 1993 and 1996 were neither associated with concomitant shifts in the proportion of smokers in the ‘‘precontemplation’’ stage between 1993 and 1996, nor with subsequent shifts in stages between 1996 and 1999. Conversely, shifts in the distribution of stages of change between 1993 and 1996 were neither associated with concomitant shifts in smoking

Table 2 Associations between the proportion of daily smokers who smoked 21 – 40 cigarettes/day and the proportion of smokers in the ‘‘precontemplation’’ or ‘‘preparation’’ stages of change, across the 50 States and the District of Columbia, USA, 1996 – 1999 Dependent variables

Percent smokers in ‘‘precontemplation’’ stage in 1996 or 1999

Percent smokers in ‘‘preparation’’ stage in 1996 or 1999

Independent variables

Regression coefficienta

R-square

Regression coefficienta

Percent daily smokers who smoked 21 – 40 cigarettes/day in 1996 Percent daily smokers who smoked 21 – 40 cigarettes/day in 1999

0.58

0.39

0.007

0.21

0.28

0.031

0.86

0.62

< 0.001

0.30

0.58

< 0.001

a

P value

R-square

P value

Coefficients from linear regression models indicate the change in the percentage of smokers in the ‘‘precontemplation’’ or ‘‘preparation’’ stages of change, for a 1% increase in the proportion of daily smokers who smoked 21 – 40 cigarettes/day. Data were weighted for the size of the population in each State.

372

J.-F. Etter / Preventive Medicine 38 (2004) 369–373

Table 3 Associations between the proportion of daily smokers who smoked 21 – 40 cigarettes/day and the proportion of smokers in the ‘‘precontemplation’’ or ‘‘preparation’’ stages of change, across the 50 States and the District of Columbia, USA, 1992 – 1999 Dependent variables

Percent smokers in ‘‘precontemplation’’ stage

Independent variables

Regression coefficienta

Percent daily smokers who quit for 1+ days in the past year, in 1996 Percent daily smokers who quit for 1+ days in the past year, in 1999

Percent smokers in ‘‘preparation’’ stage

R-square

P value

Regression coefficienta

R-square

P value

0.32

0.43

0.005

0.17

0.63

< 0.001

0.52

0.26

0.039

0.16

0.20

0.08

a

Coefficients from linear regression models indicate the change in the percentage of smokers in the ‘‘precontemplation’’ or ‘‘preparation’’ stages of change, for a 1% increase in the proportion of daily smokers who quit smoking for 1 or more days in the past year. Data were weighted for the size of the population in each State.

prevalence between 1993 and 1996, nor with subsequent shifts in smoking prevalence between 1996 and 1999. Changes in smoking prevalence between 1996 and 1999 were neither associated with concomitant shifts in the proportion of daily smokers who smoked 21 – 40 cigarettes/day between 1996 and 1999, nor with concomitant shifts in the proportion of daily smokers who quit for 1+ days during the past year between 1996 and 1999. Changes in smoking prevalence between 1993 and 1996 were neither associated with subsequent shifts in the proportion of daily smokers who smoked 21– 40 cigarettes/day between 1996 and 1999, nor with subsequent shifts in the proportion of daily smokers who quit for 1+ days during the past year between 1996 and 1999. Multivariate models These associations were not statistically significant in multivariate models including the proportion of smokers in each stage as dependent variable, whether the independent variables included only smoking prevalence and cigarette consumption, or also quit attempts (data not shown).

Discussion This study showed that across the 50 States, a higher prevalence of cigarette smoking was associated with more smokers in the ‘‘precontemplation’’ stage of change (i.e., not motivated to quit smoking). An association between smoking prevalence and the distribution of smokers by stage of change was already reported in a comparative study of five countries [4], but the present study showed that this association also holds within the United States. Contrary to a previous report that showed that smoking prevalence and nicotine dependence among smokers were inversely related across six countries [5], the present study showed that across the 50 States, a higher smoking prevalence was associated with a higher level of dependence and fewer quit attempts among smokers. Thus, our results do not confirm the hypothesis that when smoking prevalence decreases, light smokers quit first and the remaining smok-

ers are highly dependent [5]. In fact, recent research shows that there is no epidemiologic evidence that where smoking prevalence has decreased, dependence levels have increased among the remaining smokers [12]. Research also shows that the target is not hardening, that is, that smokers are not less likely to quit now than in the past [13]. A first possible explanation for associations between smoking prevalence and smoking behavior is that in States where the prevalence of cigarette smoking is low, there is more social pressure upon smokers, and thus smokers in these States are more motivated to quit smoking, smoke fewer cigarettes, and make more quit attempts. Repeated studies (time series) can be used to assess whether progress in stage distributions, cigarette consumption, and quit attempts follow, rather than precede, a decrease in smoking prevalence. In the data used in this study, we found no association between shifts in smoking prevalence and subsequent or concomitant shifts in stages of change, cigarette consumption, and quit attempts. All these variables remained stable during the 1990s in the US, which may explain why no association was detected in prospective analyses. A second possible explanation is that a shift in the distribution of smokers by stage of change or an increase in the number of smokers who make a quit attempt cause a decrease in smoking prevalence, as these shifts prepare more smokers to progress to the ‘‘action’’ stage of change, that is, to quit smoking. A decrease in cigarette consumption among smokers could also cause a decrease in smoking prevalence, as it is easier to quit for light smokers than for heavy smokers [14]. However, in prospective analyses, we observed no association between shifts in stages of change, cigarette consumption, or quit attempts and shifts in smoking prevalence. Again, changes over time were probably too small in these data to observe such a link in prospective analyses. But if this hypothesis is nevertheless true, then shifting the stage distribution of current smokers to the right and reducing cigarette consumption among smokers should be explicit goals of public policies [15]. Across the 50 States, the distributions of smokers by stage of change was positively associated with cigarette consumption and negatively associated with the proportion of daily smokers who made a recent quit attempt. It is

J.-F. Etter / Preventive Medicine 38 (2004) 369–373

possible that a positive shift in stages, quit attempts, or cigarette consumption is followed by a decrease in smoking prevalence, which in turn causes a shift in smoking behavior among the remaining smokers. In this ideal situation, a reinforcing feedback loop (or ‘‘virtuous circle’’) would drive a self-sustained decrease in smoking prevalence. Such multidirectional associations between variables can hardly be entangled using ecological data. The associations between smoking prevalence, stages of change, cigarette consumption, and quit attempts could also be explained by information bias, if people living in States where there is more anti-smoking social pressure tended to declare that they do not smoke, to overstate their level of motivation to quit smoking and understate their cigarette consumption. Finally, these associations could also be explained by confounding variables, including expenditures for tobacco control by State. Previous research showed that increases in funding for State tobacco control programs reduced tobacco use, across U.S. States [16], and that States with higher smoking prevalence invested less per capita in tobacco control programs [17]. We could not test whether there was a confounding effect by State expenditures, since CDC’s data on tobacco control expenditures by State for the years 1993, 1996, and 1999 are not publicly available. There was no statistically significant association between smoking prevalence and stages of change in 1993, and the regression coefficients were larger in 1996 than in 1993, and they were also larger in 1999 than in 1996 (Table 1). This could suggest that the strength of this association grows stronger over time. The analyses conducted in this paper should be repeated in a few years from now, to assess whether this trend continues. Because of the cross-sectional nature of available data, and because of the small changes over time during the observation period, our results do not indicate whether the associations between smoking prevalence, stages of change, cigarette consumption, and quit attempts are causal or which is the nature and the direction of a potential causal link. A confirmation of a causal link between smoking prevalence and the behavior of smokers would have important implications for policy makers, clinicians, and researchers.

Acknowledgment This study was initiated by the author, who is currently employed by the University of Geneva, Switzerland.

373

References [1] Prochaska JO, DiClemente CC, Norcross JC. In search of how people change. Applications to addictive behaviors. Am Psychol14 1992;47: 1102. [2] DiClemente CC, Prochaska JO, Fairhurst SK, Velicer WF, Velasquez MM, Rossi JS. The process of smoking cessation: an analysis of precontemplation, contemplation, and preparation stages of change. J Consult Clin Psychol 1991;59:295 – 304. [3] Wewers ME, Stillman FA, Hartman AM, Shopland DR. Distribution of daily smokers by stage of change: current population survey results. Prev Med 2003;36:710 – 20. [4] Etter JF, Perneger TV, Ronchi A. Distributions of smokers by stage: international comparison and association with smoking prevalence. Prev Med 1997;26:580 – 5. [5] Fagerstrom KO, Kunze M, Schoberberger R, Breslau N, Hughes JR, Hurt RD, et al. Nicotine dependence versus smoking prevalence: comparisons among countries and categories of smokers. Tob Control 1996;5:52 – 6. [6] Frazier EL, Okoro CA, Smith C, McQueen DV. State- and sex-specific prevalence of selected characteristics—Behavioral Risk Factor Surveillance System, 1992 and 1993. MMWR CDC Surveill Summ 1996;45:1 – 36. [7] Holtzman D, Powell-Griner E, Bolen JC, Rhodes L. State- and sexspecific prevalence of selected characteristics—Behavioral Risk Factor Surveillance System, 1996 and 1997. MMWR CDC Surveill Summ 2000;49:1 – 39. [8] CDC. State-specific prevalence of current cigarette smoking among adults and the proportion of adults who work in a smoke-free environment—United States, 1999. MMWR CDC Surveill Summ 2000;49: 978 – 82. [9] http:// apps.nccd.cdc.gov/brfss/ page.asp?cat = TU&yr = 1996&state = US#TU, accessed October 23, 2003. [10] Hansen RH. The Current Population Survey: design and methodology. Technical Paper no. 40. Washington, DC: U.S. Department of Commerce, Bureau of the Census, July 1985. [11] Campbell PR. Population projections for States by age, sex, race and Hispanic origin, PPL-47. Washington, DC: U.S. Department of Commerce, Bureau of the Census, Population Division, 1996. [12] Warner KE, Burns DM. Hardening and the hard-core smoker: concepts, evidence, and implications. Nicotine Tob Res 2003;5:37 – 48. [13] National Cancer Institute. Those who continue to smoke. Is achieving abstinence harder and do we need to change our interventions. Smoking and Tobacco Control Monograph No.15. Bethesda, MD: U.S. Department of Health and Human Services, National Institutes of Health, National Cancer Institute, 2003. [14] Ockene JK, Emmons KM, Mermelstein RJ, Perkins KA, Bonollo DS, Voorhees CC, et al. Relapse and maintenance issues for smoking cessation. Health Psychol 2000;19:17 – 31. [15] Tonnesen P. Smoking reduction for smokers not able or motivated to quit? Respiration 2002;69:475 – 8. [16] Farrelly MC, Pechacek TF, Chaloupka FJ. The impact of tobacco control program expenditures on aggregate cigarette sales: 1981 – 2000. J Health Econ 2003;22:843 – 59. [17] Gross CP, Soffer B, Bach PB, Rajkumar R, Forman HP. State expenditures for tobacco-control programs and the tobacco settlement. N Engl J Med 2002;3(347):1080 – 6.