Abstract
Objectives. Smokeless tobacco has many adverse health effects. We analyzed long-term national trends in smokeless tobacco use.
Methods. We used 1987 to 2000 National Health Interview Survey data for adults aged 18 years and older, 1986 to 2003 data from Monitoring the Future surveys of adolescents, and 1991 to 2003 data from the Youth Risk Behavior Survey for 9th- to 12th-grade students to examine overall and demographic-specific trends.
Results. Smokeless tobacco use among adult and adolescent females was low and showed little change. Smokeless tobacco use among men declined slowly (relative decline=26%), with the largest declines among those aged 18 to 24 years or 65 years and older, Blacks, residents of the South, and persons in more rural areas. Overall and demographic-specific data for adolescent boys indicate that smokeless tobacco use increased for 12th-grade students from 1986 until the early 1990s, but has subsequently declined rapidly in all grades since then (range of relative overall declines=43% to 48%).
Conclusions. Smokeless tobacco use has declined sharply, especially among adolescent boys. Ongoing prevention and cessation efforts are needed to continue this trend.
Smokeless tobacco products, which consist of chewing tobacco, moist snuff, and dry snuff, cause many adverse health effects, including cancers of the oral cavity and pharynx, oral soft tissue lesions, gum recession, and nicotine addiction.1–9 One large cohort study also reported an association between smokeless tobacco use and cardiovascular disease,10 although this has not been a consistent finding.11,12 More recent studies suggest that smokeless tobacco use may be associated with adverse pregnancy outcomes13 and pancreatic cancer.14
Studies of US adults have shown that smokeless tobacco use is much higher among certain adult demographic groups, including men, young adults, rural residents, residents of southern or western states, Whites, American Indians/Alaska Natives, and persons with lower levels of education.15–19 For adolescents, smokeless tobacco use is higher among those of older age, boys, Whites, American Indians/Alaska Natives, and residents in some parts of the South, Midwest, and West.20–22
Smokeless tobacco use was widely recognized as a major public health problem beginning in the mid-1980s after the smokeless tobacco industry began a wide-scale advertising and marketing campaign earlier in the decade that succeeded in substantially boosting sales.1,5,7,9,20,23,24 In 1986, Congress enacted legislation banning smokeless tobacco television and radio advertising, and requiring warning labels on such products.20 Since the early 1990s, less attention has been focused on smokeless tobacco issues. Some health scientists in recent years have even suggested smokeless tobacco be actively promoted to cigarette smokers as a safer alternative (i.e., a harm-reduction product) for those having difficulty quitting smoking,25,26 although others have noted that considerably more research and product regulation is necessary before considering such a strategy.4,27
There have been some studies of adult smokeless tobacco use on the basis of selected national, state, and other data sets.16,19,28,29 Descriptive reports suggest that smokeless tobacco use, which increased between 1970 and 1991 among adults,30 may have declined in more recent years.31 There are several studies on the prevalence and correlates of adolescent smokeless tobacco use.32–34 Overall national data from Monitoring the Future (MTF) surveys show that smokeless tobacco prevalence among adolescents declined, beginning in the mid-1990s, among students in grades 8, 10, and 12.35 The Centers for Disease Control and Prevention’s (CDC) national Youth Risk Behavior Survey (YRBS) also reports an overall decline in smokeless tobacco use among 9th- through 12th-grade students since 1991.36
To date, however, formal trend analyses of smokeless tobacco use are rare.18 Overall findings on smokeless tobacco use, although important, may mask differing smokeless tobacco trends among demographic subgroups such as age, gender, race/ethnicity, and place of residence. To better describe and understand national adult and adolescent trends in smokeless tobacco use over the past 2 decades, we used several nationally representative data sets to conduct overall and subgroup trend analyses.
We used the National Health Interview Survey (NHIS) to examine adult smokeless tobacco use trends. (Note: We considered using data from the Census Bureau’s Current Population Survey and the Substance Abuse and Mental Health Administration’s National Household Survey on Drug Abuse [the name was changed to the National Survey on Drug Use and Health in 2002].37 We chose not to do so because of substantial changes in design, modes of interview, or questions in these surveys during the study period.)
Detailed descriptions of the NHIS are available elsewhere and are only briefly summarized here.38,39 The NHIS is designed to represent the civilian, noninstitutionalized adult population of the United States. Data are primarily collected through in-person interviews conducted in households. Smokeless tobacco questions were included in the 1987, 1991, 1992, 1994, 1998, and 2000 surveys. For these years, the total number of households participating ranged from 38 000 to 49 000 per year, with the total number of participants ranging from 98 000 to 128 000; household-level response rates ranged from 87% to 96%.40–45
There were slight variations in smokeless tobacco question wording over the study period. Before 1998, current smokeless tobacco users were defined as persons who had used either chewing tobacco or snuff 20 or more times in their life and who reported that they were now current users of either product. In 1998 and 2000, current smokeless tobacco use was defined as persons ever using either chewing tobacco or snuff 20 or more times in their life and using either smokeless tobacco product every day or some days.
Preliminary analyses indicated that smokeless tobacco use among women was very low; as a result, we limited further analyses to men only, examining trends by age group (18–24, 25–44, 45–64, and ≥65 years), race/ethnicity (non-Hispanic White, non-Hispanic Black, or Hispanic), education level (less than high school, high-school graduate, some college/technical school, or college graduate), marital status (married or not married), geographic region (Northeast, Midwest, South, or West), and population density (metropolitan statistical area [MSA] or non-MSA; i.e., urban/suburban versus more rural areas).
We used the National Institute for Drug Abuse’s MTF surveys, which are conducted by the University of Michigan, and CDC’s national YRBS, to examine national trends in adolescent smokeless tobacco use. More detailed descriptions of these surveys are available elsewhere.35–46 Briefly, the MTF uses a cluster design to select a nationally representative sample of approximately 18000 eighth-grade, 17000 10th-grade, and 16000 12th-grade students each year from between 130 and 150 public and private high schools. Smokeless tobacco use data from MTF surveys were available annually for 12th-grade students from 1986 to 1989 and 1992 to 2003 (no smokeless tobacco questions were included in 1990 and 1991); for students in 8th and 10th grades, data were available annually from 1991 to 2003.
The smokeless tobacco questions were the same for all grades and years: “Have you ever taken or used smokeless tobacco (snuff, plug, dipping tobacco, or chewing tobacco) at least once or twice?” “How frequently have you taken smokeless tobacco during the past 30 days?” Current smokeless tobacco use was defined as ever using such products and using an smokeless tobacco product at least once within the past 30 days.
The YRBS also uses a cluster design; data are collected every 2 years from a nationally representative sample of students from approximately 110 to 160 schools in grades 9 to 12 through self-completed written surveys administered in classrooms.46 Smokeless tobacco data are available from 1991 through 2003. The YRBS student response rates ranged from 83% to 90%, and sample sizes ranged from 10 904 to 16 296 over this period.21,47–52
Although there were a few minor changes over time, questions used on the YRBS to ascertain smokeless tobacco use were generally similar. In 1991 and 1993, smokeless tobacco use was defined as answering “yes” to a question about use of smokeless tobacco products (chewing tobacco or snuff) in the past 30 days. From 1995 through 1999, current use was defined as reporting use of smokeless tobacco products on 1 or more of the past 30 days. In 2001 and 2003, the same definition was used but the question was slightly modified to include the word “dip” in the definition of smokeless tobacco products.
As with adults, prevalence of smokeless tobacco use in the MTF and YRBS was initially analyzed by gender; because smokeless tobacco prevalence was low for girls in all years, the remaining analyses by subgroups were restricted to adolescent boys. Bivariate analyses of MTF and YRBS data were done by race/ethnicity (White, Black, and Hispanic for YRBS; White and Black for MTF) and region (Northeast, Midwest, South, and West). (Note: Regional data were not available in the YRBS for 1997.) Additional YRBS analyses were stratified by grade (9th/10th or 11th/12th; MTF data were analyzed separately for each grade); additional MTF analyses were stratified by population density (MSA vs non-MSA).
Data for all surveys and years were weighted to be nationally representative, and missing or unknown data were excluded from all analyses. For the NHIS and YRBS, smokeless tobacco prevalence estimates, variances, and confidence intervals were calculated using SUDAAN, Version 8.0 (Research Triangle Institute, Research Triangle Park, NC), which accounts for the complex survey designs used in these 2 surveys. For estimating variances on the basis of MTF data, we used published design effects supplied by the University of Michigan.53 The school-based cluster design employed for both of the adolescent surveys, which used geographic area at the first selection stage and specific schools at the second stage, resulted in generally large design effects and wide 95% confidence intervals.
Exploratory and model-fitting data analyses were conducted in SAS, Version 8 (SAS Institute Inc, Cary, NC). Coefficients for trend lines for each survey and demographic domain were estimated using weighted least squares regression.54 The Y-values used in the regressions were the estimated smokeless tobacco prevalence estimates weighted by the inverse of their estimated variances; the X-values were the survey years. Residual plots were used to assess the fit of each model.
For each domain, the estimated smokeless tobacco prevalences were plotted to roughly ascertain the shape of the trend line. Linearity was tested by fitting a second-degree polynomial. The quadratic term was not significantly different than zero for more than 85% of all models. For these analyses, a linear model was fit and the estimated slope of the line used to estimate the average annual percentage point change. For the remaining analyses, we used a second-degree polynomial model. The average annual percentage point change in prevalence for these nonlinear analyses was estimated on the basis of the difference between the ending and starting predicted prevalence values divided by the total number of years.
Smokeless tobacco use among adults overall, and among men only, declined slowly but steadily from 1987 through 2000 (Table 1 and Figure 1). Use among women was rare over this period (< 1% in all years) although prevalence declined slightly; thus, overall trends primarily reflect changes in use among men. Although the average annual change among men was small (0.14 percentage points), the relative decline in smokeless tobacco use was 26% over the study period.
There were differences in smokeless tobacco trends among men by age, race/ethnicity, education level, and population density (Table 1). The largest declines in smokeless tobacco use by age were seen among those aged 18 to 24 years or 65 years and older (about 0.3 percentage points per year); there was a smaller but statistically significant decline among men aged 45 to 64 years, but no decline among men aged 25 to 44 years. Smokeless tobacco use dropped faster among Black compared with White men; no significant change was seen for Hispanic men but smokeless tobacco use was low for this group in all years. Significant declines by education level were seen only for men with a high-school education or less. Significant regional declines in smokeless tobacco use occurred among men in all but the West, with the largest declines in the South and Midwest. A nearly 3-fold greater decline in smokeless tobacco use (0.25 vs 0.08 percentage points per year) was found for men in more rural areas (non-MSAs) compared with men in urban areas, although prevalence in non-MSA areas remained substantially higher.
Overall and gender-specific trends from MTF are included in Table 2 and Figure 2. As with adult women, smokeless tobacco use among adolescent girls was low in all years and did not change; thus, overall trends reflect changes among adolescent boys. Smokeless tobacco use among 8th- and 10th-grade boys has declined since the early 1990s by an average of 0.6 to 0.7 percentage points per year. For 12th-grade boys, smokeless tobacco use increased slightly during the late 1980s, leveled off in the early 1990s, and subsequently declined, with especially rapid annual decreases since the late 1990s. The relative percent declines in smokeless tobacco use were 48% for both 8th- and 10th-grade boys from 1991 to 2003, and 43% for 12th-grade boys from 1986 to 2003.
There were some differences by demographic groups in the MTF surveys (Table 3). Except for Black adolescent boys in grades 8 and 10 (for whom smokeless tobacco use was much lower than for White adolescent boys in all years), significant declines in smokeless tobacco use occurred for the remaining subgroups from 1991 to 2003. No significant decline was found for 12th-grade boys in the Northeast.
Generally similar findings for smokeless tobacco use among adolescents in grades 9 through 12 were seen in the YRBS from 1991 to 2003 (Tables 2 and 3; Figure 2). Once again, smokeless tobacco use among girls was uncommon and did not change over the study period. There was a linear decline of 0.76 percentage points per year among boys (which represented a relative decline of 43% from 1991 to 2003) that was similar across grades. Smokeless tobacco use was much lower among both Black and Hispanic adolescent boys in all years. Significant declines in smokeless tobacco use among boys were seen by grade and among residents in the Midwest and South.
There has been substantial progress in reducing adult and adolescent smokeless tobacco use, with declines especially large among adolescent boys. Fortunately, smokeless tobacco use among women and adolescent girls remained low throughout the study periods. For adult men, significant declines occurred in all age groups except among men aged 25 to 44 years; this group now has the highest prevalence of smokeless tobacco use among all adult age groups. Declines among several adult male subgroups with high levels of use in 1987 (i.e., men with lower education levels, residents of the South or Midwest, residents in more rural areas) also were encouraging, as these are populations that historically have had the highest smokeless tobacco prevalence. After increasing among 12th-grade boys during the late 1980s and early 1990s, smokeless tobacco use has declined substantially among boys in most sub-populations in all grades.
The declines in adult and adolescent prevalence mirror, to some extent, declines in overall smokeless tobacco production and sales volumes.55,56 Smokeless tobacco sales volume decreased by 11% from 1986 to 2003 (from 125.5 to 114.3 million pounds)56; most of this decline resulted from large drops in sales of chewing tobacco and dry snuff (−49% and −67%, respectively). In marked contrast, the use of moist snuff or dip, through brands such as Copenhagen and Skoal, increased dramatically over this same period (+87%); moist snuff now represents 62% of the smokeless tobacco market.56 Thus, there are somewhat paradoxical market trends with a large increase in moist snuff use occurring because of the shift in smokeless tobacco product preference while overall smokeless tobacco production and sales have declined.
Reasons for the declines in prevalence of smokeless tobacco use cannot be directly ascertained from these data; however, several factors are likely to play some role. Economics may play an important role, as increases in smokeless tobacco excise taxes are associated with decreased prevalence of smokeless tobacco use for adults57 and adolescents.58 The number of states with some type of excise tax on smokeless tobacco products not only increased from 24 states in 1986 to 46 states in 2004, but also the level of state excise taxes grew substantially over this time period.59–61 Sociodemographic changes could also be a factor in the overall decline in prevalence and consumption—smokeless tobacco use is highest among Whites and rural residents, but both the percentage of the US population that is White, and the percentage that resides in rural areas, is declining.62,63 However, the significant declines among adult and adolescent males in non-MSA areas, and the decline among White boys in grades 8 and 10, suggest that the smokeless tobacco initiation rate is decreasing among these groups; thus, the decline is not solely a function of changing demographics.
There has been an overall decline in cigarette smoking, smokeless tobacco use, and other types of tobacco use (e.g., bidis, kreteks) among adolescents since the late 1990s.35,36,64 As with their perceptions of cigarette smoking, there was a substantial increase over time in the percentage of 8th-, 10th-, and 12th-grade students who perceived that regular smokeless tobacco use is harmful.35 Our findings of declining adolescent smokeless tobacco use suggest that the extensive youth antitobacco (primarily anti–cigarette-smoking) efforts, which began in California in the late 1980s and extended throughout the country during the 1990s,65 may have helped to increase the perception that smokeless tobacco use is harmful and helped to reduce smokeless tobacco prevalence. The national truth paid media campaign, which began in 1999 and focused on cigarette-smoking prevention, may have played a role, as adolescents were heavily exposed to anti–tobacco-use messages.66 Specific anti–smokeless tobacco efforts, such as the National Spit Tobacco Education Program, begun in 1994, also may have contributed.67
There are limitations to this study. Adult trend data were available only through 2000; thus, national prevalence of smokeless tobacco use on the basis of NHIS data since then is unknown. Data from the National Household Survey on Drug Abuse (now the National Survey on Drug Use and Health) from 2000 through 2003 show a slight decline in smokeless tobacco use among men aged 18 years and older (from 6.8% to 6.5%, on the basis of different questions than the NHIS), and no change among women, over this time period.68–70 The continued decrease in sales volume since 2000 provides further evidence that smokeless tobacco prevalence may be continuing its decline, albeit at a slow rate.56
It should be noted that prevalence estimates for smokeless tobacco use were consistently higher among adolescents than for adults because of different definitions (i.e., in 2000 the 12th-grade prevalence for boys from the MTF survey was 14.2%, compared with 5.5% for men aged 18 to 24 years in the NHIS [MTF data not shown in tables]). Relying on a different definition of current tobacco use for adolescents is a common practice because tobacco use for many in this age group is not established, and there is strong evidence that adolescents who use tobacco products even occasionally are at much greater risk of becoming regular adult users.20 The different definitions for adults and adolescents, however, do not affect trend analyses.
Each of the surveys relied on self-reports without biochemical verification. Limited research suggests that self-reports slightly underestimate actual use for adults and adolescents when compared with biochemically verified use20,71; however, this limitation is unlikely to affect trends. Unfortunately, because of small numbers we were not able to analyze trends for American Indians/Alaska Natives, which is the racial/ethnic group with the highest adult smokeless tobacco prevalence.68–71
There were some changes in wording to smokeless tobacco questions, especially for the NHIS and YRBS. The most important change occurred with the 1998 NHIS, when the definition of current smokeless tobacco use changed from “use now” to “use every day or some days.” On the basis of additional analyses of 1991 data, in which both types of questions were used, the “use every day or some days” definition increased the smokeless tobacco prevalence estimate by 0.5 percentage points (data not shown). The impact of this wording change strongly implies that our estimates for the adult smokeless tobacco decline were conservative. Finally, both the MTF and YRBS were school-based surveys that used a cluster design with large design effects, which helps explain the absence of statistical significance for some groups despite large declines in point prevalence estimates.
Despite the good news about declining smokeless tobacco use, a disturbing development in recent years has been advocacy by some in the scientific literature and lay press recommending smokeless tobacco use as a harm-reduction product.19,25,27,72–74 Smokeless tobacco is not a benign product; it is a cause of several types of diseases and is a known human carcinogen.1,6,27 There is no evidence or scientific consensus that smokeless tobacco is an effective tobacco-cessation method.4 Proven and safer tobacco-cessation therapies (e.g., behavioral interventions, nicotine-replacement products, nonnicotine medications) are available to cigarette smokers interested in quitting, without the health risks associated with smokeless tobacco use.75,76 Cigarette smokers should use one of these proven therapies, rather than use smokeless tobacco products. Finally, given the rapid decline among adolescents and the slower but steadier decline among adults for both cigarette and smokeless tobacco use,35,36,46,64,77 it would be counterproductive to send messages endorsing smokeless tobacco use under any circumstance.
The declines in smokeless tobacco use that our study showed is excellent news. However, it is essential to remember that the tobacco industry, including the smokeless tobacco industry, has a long history of resilience and resourcefulness.78 An aggressive marketing campaign by the smokeless tobacco industry in the 1970s and 1980s helped fuel a large increase in smokeless tobacco use, with almost all new use occurring among adolescent and young adult males.5,24 Several smokeless tobacco companies were not a party to the 1998 Master Settlement Agreement and are not subject to the same marketing and advertising restrictions as cigarette companies.79
It has been well documented that the major manufacturers of smokeless tobacco products manipulate the nicotine dosing characteristics of their products,80,81 and that they have developed moist snuff products that deliver nicotine at varying rates of absorption.82–84 These products include “starter” products that have relatively slow rates of nicotine delivery, and often incorporate flavorings such as cherry, berry, or mint that mask the tobacco taste, which may appeal to young, novice users. As users develop greater tolerance to nicotine and progress in their level of addiction, they are likely to “graduate” to brands engineered to deliver much higher levels of nicotine with more rapid rates of absorption.24,85,86 Despite recent statements by a major smokeless tobacco manufacturer that its products are intended for adults, particularly smokers looking for an alternative tobacco product, it is clear that young males remain a major marketing target.87 The declining trend in smokeless tobacco use among adolescent males parallels recent declines in smoking among that group.35,36 These patterns suggest that it is possible to reduce initiation of tobacco use among young people without promoting product substitution.
Substantial reductions have occurred in many states’ antitobacco programs as a result of budget difficulties and the diversion of Master Settlement Agreement and tobacco excise tax revenues away from tobacco control and prevention activities.88 Such cuts especially threaten the large gains made in recent years to reduce tobacco use among adolescents.89 Continued and ongoing comprehensive efforts to reduce smokeless tobacco and other tobacco use are needed. Increasing smokeless tobacco excise taxes, mass media countermarketing campaigns, school-based and other educational efforts, and prevention and cessation counseling and other interventions by health care providers are necessary to sustain the declining use of both smokeless tobacco and cigarettes.
Note. CI = confidence interval; NS = not significant; MSA = metropolitan statistical area. aNonlinear trend. Note. CI = confidence interval; NS = not significant. In the MTF survey for eighth and 10th grades, initial year = 1991, middle year = 1997, and final year = 2003. In the MTF survey for 12th grade, initial year = 1986, middle year = 1994, and final year = 2003. In the YRBS, initial year = 1991, middle year = 1997, and final year = 2003. aNonlinear trend. Note. CI = confidence interval; MSA = metropolitan statistical area; NS = not significant. In the MTF survey for eighth and 10th grades, initial year = 1991, middle year = 1997, and final year = 2003. In the MTF survey for 12th grade, initial year = 1986, middle year = 1994, and final year = 2003. In the YRBS, initial year = 1991, middle year = 1997, and final year = 2003. a Nonlinear trend. FIGURE 1— Trends in adult smokeless tobacco use, 1987–2000. Source. Data are from the National Health Interview Survey. FIGURE 2— Trends in smokeless tobacco use among adolescent males, 1986–2003. Source. Data are from the Monitoring the Future (MTF) survey and Youth Risk Behavior Survey (YRBS).
1987 % (CI) 1994 % (CI) 2000 % (CI) Average Annual Percentage Point Change P R2 Overall and by gender Overall 3.2 (2.9, 3.5) 3.0 (2.6, 3.4) 2.3 (2.1, 2.5) −0.08 <.01 0.70 Men 6.1 (5.5, 6.7) 6.0 (5.2, 6.8) 4.5 (4.1, 4.9) −0.14 <.01 0.85 Women 0.6 (0.4, 0.8) 0.5 (0.3, 0.7) 0.3 (0.2, 0.4) −0.03 <.05 0.81 Demographic subgroups among men Age, y 18–24 8.9 (7.4, 10.4) 7.7 (5.4, 10.0) 5.0 (3.5, 6.5) −0.32 <.05 0.74 25–44 5.5 (4.8, 6.2) 6.6 (5.6, 7.6) 5.8 (5.1, 6.5) 0.02 NS 0.07 45–64 5.0 (4.1, 5.9) 4.6 (3.4, 5.8) 3.1 (2.8, 3.4) −0.14 <.05 0.80 ≥65 6.9 (5.7, 8.1) 5.3 (3.9, 6.7) 2.8 (2.0, 3.6) −0.31 <.01 0.95 Race/ethnicity White 6.9 (6.2, 7.6) 7.4 (6.4, 8.4) 5.5 (5.0, 6.0) −0.09a <.01 0.91 Black 3.9 (2.8, 5.0) 2.8 (1.6, 4.0) 1.4 (0.8, 2.0) −0.24 <.05 0.76 Hispanic 1.5 (0.6, 2.4) 0.6 (0.3, 0.9) 0.8 (0.3, 1.3) −0.04 NS 0.26 Education < High school 9.0 (7.8, 10.2) 8.6 (6.6, 10.6) 5.8 (4.5, 7.1) −0.23 <.01 0.89 High school 7.1 (6.1, 8.1) 7.2 (5.8, 8.6) 5.5 (4.8, 6.2) −0.15 <.05 0.71 Some college 5.0 (4.1, 5.9) 5.1 (4.0, 6.2) 4.4 (3.6, 5.2) −0.04 NS 0.20 College graduate 2.2 (1.7, 2.7) 3.6 (2.6, 4.6) 2.2 (1.6, 2.8) 0.01 NS 0.01 Marital Status Married 6.0 (5.3, 6.7) 6.3 (5.4, 7.2) 4.4 (3.9, 4.9) −0.13 <.05 0.78 Not married 6.3 (5.4, 7.2) 5.5 (4.4, 6.6) 4.4 (3.7, 5.1) −0.16 <.05 0.74 Region Northeast 2.6 (1.3, 3.9) 2.8 (1.6, 4.0) 2.2 (1.7, 2.7) −0.06 <.05 0.73 Midwest 6.3 (5.1, 7.5) 6.6 (5.2, 8.0) 4.4 (3.7, 5.1) −0.17 <.05 0.74 South 9.2 (7.9, 10.5) 9.1 (7.3 10.9) 6.7 (5.9, 7.5) −0.24 <.01 0.85 West 4.2 (3.1, 5.3 4.1 (2.8, 5.4) 2.6 (1.7, 3.5) −0.09 NS 0.32 Population density MSA 4.2 (3.6, 4.8) 4.0 (3.4, 4.6) 3.3 (2.9, 3.7) −0.09 <.05 0.79 Non-MSA 12.2 (10.8, 13.6) 13.6 (11.0, 16.2) 9.0 (7.7, 10.3) −0.26 <.05 0.75 Initial Year % (CI) Middle Year % (CI) Final Year % (CI) Average Annual Percentage Point Change P R2 Monitoring the Future Eighth grade Overall 7.2 (6.1, 8.3) 5.6 (4.6, 6.6) 4.2 (3.3, 5.1) −0.36 <.01 0.85 Boys 13.2 (11.6, 14.8) 9.9 (8.4, 11.4) 6.9 (5.6, 8.2) −0.62 <.01 0.77 Girls 1.4 (0.7, 2.1) 1.6 (0.9, 2.3) 1.8 (1.0, 2.6) −0.05 NS 0.12 10th grade Overall 10.0 (8.6, 11.4) 8.8 (7.5, 10.1) 6.1 (5.0, 7.2) −0.38 <.01 0.85 Boys 18.6 (16.6, 20.6) 14.9 (13.1, 16.7) 9.6 (8.1, 11.1) −0.74 <.01 0.89 Girls 1.3 (0.6, 2.0) 2.7 (1.7, 3.7) 1.3 (0.6, 2.0) −0.03 NS 0.06 12th grade Overall 11.5 (10.3, 12.7) 11.3 (8.9, 13.7) 6.7 (4.8, 8.6) −0.24a <.01 0.75 Boys 22.2 (20.4, 24.0) 20.5 (16.9, 24.1) 12.6 (9.7, 15.5) −0.50a <.01 0.71 Girls 1.6 (1.0, 2.2) 2.8 (1.1, 4.5) 1.0 (0.0, 2.0) 0.01 NS 0.00 Youth Risk Behavior Surveys Ninth–12th grade Overall 10.5 (9.6, 11.4) 9.3 (7.1, 11.5) 6.7 (5.2, 8.2) −0.38 <.01 0.82 Boys 19.2 (17.8, 20.6) 15.8 (12.1, 19.5) 11.0 (8.7, 13.3) −0.76 <.01 0.87 Girls 1.3 (0.7, 1.9) 1.5 (0.8, 2.2) 2.2 (1.0, 3.4) 0.02 NS 0.05 Year Initial % (CI) Middle % (CI) Final % (CI) Average Annual Percentage Point Change P R2 Eighth grade (MTF) Race/ethnicity White 15.2 (12.5, 17.9) 12.0 (9.4, 14.6) 7.3 (5.1, 9.5) −0.83 <.01 0.87 Black 2.0 (0.0, 5.0) 3.7 (0.8, 6.6) 6.0 (1.5, 10.5) −0.01 NS 0.00 Region Northeast 11.8 (7.3, 16.3) 5.2 (1.9, 8.5) 5.6 (2.5, 8.7) −0.44 <.05 0.42 Midwest 12.8 (11.0, 14.6) 12.3 (7.6, 17.0) 5.4 (2.3, 8.5) −0.73 <.01 0.85 South 18.4 (13.9, 22.9) 12.6 (9.3, 15.9) 10.8 (7.5, 14.1) −0.79 <.01 0.81 West 5.0 (3.4, 6.6) 6.3 (3.1, 9.5) 2.3 (0.0, 4.6) −0.48a <.01 0.68 Population density MSA 10.5 (9.4, 11.6 7.3 (5.4, 9.2) 5.2 (3.6, 6.8) −0.60 <.01 0.88 Non-MSA 20.5 (17.8, 23.2) 16.6 (12.1, 21.1) 12.2 (7.8, 16.6) −0.79 <.01 0.76 10th grade (MTF) Race/ethnicity White 22.0 (18.7, 25.3) 18.0 (15.1, 20.9) 11.6 (9.0, 14.2) −0.99 <.01 0.92 Black 6.2 (1.7, 10.7) 5.0 (0.6, 9.4) 5.6 (1.6, 9.6) −0.02 NS 0.00 Region Northeast 14.3 (8.9, 19.7) 14.3 (9.9, 18.7) 8.0 (4.3, 11.7) −0.57 <.01 0.55 Midwest 20.7 (15.8, 25.6) 12.4 (8.5, 16.3) 8.8 (5.2, 12.4) −0.96 <.01 0.88 South 22.3 (17.4, 27.2) 18.8 (14.2, 23.4) 14.1 (10.1, 18.1) −0.78 <.01 0.82 West 14.7 (9.8, 19.6) 12.3 (7.0, 17.6) 5.5 (2.3, 8.7) −1.10 <.01 0.76 Population density MSA 15.2 (12.617.8) 11.5 (9.2, 13.8) 7.8 (5.9, 9.7) −0.78 <.01 0.85 Non-MSA 27.5 (21.1, 33.9) 25.4 (19.6, 31.2) 7.1 (1.2, 13.0) −0.99 <.01 0.61 12th grade (MTF) Race White 25.6 (20.3, 30.9) 24.6 (18.8, 30.4) 15.1 (10.1, 20.1) −0.42a <.01 0.62 Black 0.8 (0.0, 3.8) 2.4 (0.0, 7.6) 2.5 (0.0, 5.0) −0.04 <.05 0.64 Region Northeast 16.7 (8.2, 25.2) 20.5 (9.7, 31.3) 10.8 (3.1, 18.5) −0.22 NS 0.06 Midwest 25.0 (16.4, 33.6) 27.6 (17.4, 37.8) 10.2 (3.5, 16.9) −0.41a <.05 0.58 South 25.6 (17.0, 34.2) 17.1 (9.7, 9.24.5) 16.5 (8.9, 24.1) −0.64 <.01 0.61 West 19.2 (9.3, 29.1) 17.7 (8.7, 26.7) 11.5 (3.6, 19.4) −0.64 <.01 0.68 Population density MSA 17.8 (13.1, 22.5) 18.8 (13.8, 23.8) 10.3 (6.4, 14.2) −0.44a <.01 0.69 Non-MSA 30.9 (20.8, 41.0) 24.8 (14.4, 35.2) 18.9 (9.0, 28.8) −0.53 <.05 0.32 9th–12th grade (YRBS) Grade 9th and 10th 17.4 (14.3, 20.5) 14.0 (10.4, 17.6) 9.3 (7.0, 11.6) −0.77 <.01 0.85 11th and 12th 21.0 (17.8, 24.2) 17.4 (13.0, 21.8) 13.0 (10.1, 15.9) −0.71 <.01 0.88 Race/ethnicity White 23.6 (20.3, 26.9) 20.6 (16.6, 24.6) 13.2 (9.9, 16.5) −0.66 NS 0.80 Black 3.6 (2.1, 5.1) 3.2 (1.5, 4.9) 4.1 (2.3, 5.9) −0.05 NS 0.12 Hispanic 10.7 (5.0, 16.4) 8.4 (5.0, 11.8) 6.1 (2.6, 9.6) −0.18 NS 0.17 Region Northeast 16.1 (10.2, 22.0) … 5.9 (3.6, 8.2) −1.46 NS 0.60 Midwest 22.2 (16.5, 27.9) … 15.6 (10.8, 20.4) −0.88 <.05 0.69 South 16.9 (12.9, 20.9) … 12.6 (9.4, 15.8) −0.28 NS 0.28 West 20.3 (13.8, 26.8) … 6.1 (2.0, 10.2) −0.93 <.05 0.68 
We thank Hank Wells for his statistical assistance.
Human Participant Protection No protocol or approval was needed for this study, because analyses were conducted using publicly available data sets. Institutional review board approval was previously obtained for the Monitoring the Future surveys from the University of Michigan, and for the Youth Risk Behavior Surveys and the National Health Interview Surveys from the Centers for Disease Control and Prevention.
