Trends in Cigarette Smoking Among High School Students: United States, 1991-19

One of the 10 leading health indicatorsthat reflect the major health concerns in the United States is cigarette smokingamong adolescents. To examine changes in cigarette smoking among high schoolstudents in the United States from 1991 to 1999, the Centers for Disease Controland Prevention (CDC) analyzed data from the national Youth Risk Behavior Survey(YRBS). This report summarizes the results of the analysis and indicates thatcurrent smoking among high school students in the United States increasedsignificantly from 27.5% in 1991 to 34.8% in 1999; however, the analysis alsosuggested that later in the decade, current smoking may have leveled or possiblybegun to decline.

The YRBS measures the prevalence of health-risk behaviors amongadolescents through representative biennial national, state, and local surveys.The 1991, 1993, 1995, 1997, and 1999 national surveys used independent,three-stage cluster samples to obtain cross-sectional data representative ofstudents in grades 9 through 12 in the 50 states and the District of Columbia.In 1991, 1993, 1995, 1997, and 1999, the respective sample sizes were 12,272,16,296, 10,904, 16,262, and 15,349; school response rates were 75%, 78%, 70%,79%, and 77%; student response rates were 90%, 90%, 86%, 87%, and 86%; andoverall response rates were 68%, 70%, 60%, 69%, and 66%.

Self-Administered Questionnaire

For each cross-sectional survey, students completed ananonymous, self-administered questionnaire that included identically wordedquestions about cigarette smoking. Lifetime smoking was defined as having eversmoked cigarettes, even one or two puffs. Current smoking was defined as smokingon ³ 1 of the 30 days preceding the survey. Frequentsmoking was defined as smoking on ³ 20 of the 30days preceding the survey. Data are presented only for non-Hispanic black,non-Hispanic white, and Hispanic students because the numbers of students fromother racial/ethnic groups were too small for meaningful analysis.

Data were weighted to provide national estimates; SUDAAN wasused for all data analysis. Secular trends were analyzed using logisticregression analyses that controlled for sex, race/ethnicity, and grade and thatsimultaneously assessed linear and quadratic time effects. Quadratic trendssuggest a significant but nonlinear trend in the data over time. When asignificant quadratic trend accompanies a significant linear trend, the datademonstrate some nonlinear variation (eg, leveling or change in direction) inaddition to a linear trend.

Linear Trend Found for 10th-Grade Students

The prevalence of lifetime smoking remained stable from 1991 to1999 among high school students overall and among all sex, racial/ethnic, andgrade subgroups except 10th-grade students. In 1999, 70.4% (95% confidenceinterval [CI] = ± 3.0%) of all students reported lifetime smoking. Among10th-grade students, lifetime smoking showed a significant linear trend from1991 (68.3% [95% CI = ± 3.31%) to 1999 (73.9% [95% CI = ± 4.11%).

From 1991 to 1999, current smoking exhibited a significantlinear trend among students overall and among all sex, racial/ethnic, and gradesubgroups (Table 1). The overall prevalence of current smoking was 27.5%in 1991 and 34.8% in 1999. A simultaneous quadratic trend was identified forstudents overall, suggesting a leveling or possible decline in current smoking.The male, black, black male, and 9th-grade student subgroups also showed thissimultaneous quadratic trend.

Each year, white students were significantly more likely thanHispanic students, who were significantly more likely than black students, toreport current smoking (except in 1995 when white and Hispanic students wereequally likely to report current smoking, but both were significantly morelikely than black students to report this behavior). In 1991, white studentswere 2.5 times more likely than black students and 1.2 times more likely thanHispanic students to report current smoking. In 1999, white students were 2.0times more likely than black students and 1.2 times more likely than Hispanicstudents to report current smoking.

The prevalence of frequent smoking showed a significant lineartrend from 1991 to 1999 among students overall and in all sex, racial/ethnic,and grade subgroups, except for Hispanic female students. The overall prevalenceof frequent smoking was 12.7% (95% CI = ± 2.2) in 1991 and 16.8% (95% CI = ±2.5) in 1999. Among Hispanic female students, the prevalence of frequent smokingremained stable from 1991 to 1999. For each of the five surveys, white studentswere significantly more likely than black and Hispanic students to report thisbehavior.

Editorial Note From the CDC

Despite a leveling or possible decline in current smoking amongyouth overall during the late 1990s, this trend may have been limited toselected groups (ie, male, black, black male, and 9th-grade students). Inaddition, frequent smoking rates overall and in all sex, racial/ethnic, andgrade subgroups (except Hispanic females) were significantly higher in 1999 thanin 1991 and showed no pattern of leveling or declining.

Additional research is needed to understand how current smokingrates and secular changes in these rates vary among racial/ethnic groups. Forexample, throughout the decade, the YRBS and other national surveys found thatblack high-school students smoked at lower rates than white and Hispanichigh-school students; however, the 1999 National Youth Tobacco Survey reportedthat current smoking rates among black middle-school students were similar torates among white and Hispanic middle-school students.

Among grade subgroups, data for 9th-grade students suggested aleveling or possible decline in current smoking. Current smoking among12th-grade students continued to rise each year. A previous study suggested thatcurrent smoking peaked among 10th- and 12th-grade students in 1996 and 1997,respectively. It is unclear whether future YRBS data will show a delayed peakamong 10th- and 12th-grade students.

Limitations of Findings

The findings in this report are subject to at least threelimitations. First, these data apply only to adolescents who attend high school.In 1998, 5% of persons aged 16 to 17 years were not enrolled in a high schoolprogram and had not completed high school. Second, the extent of underreportingor overreporting in the YRBS cannot be determined, although the survey questionsdemonstrate good test-retest reliability. Finally, using only five data pointsmakes it possible to characterize trends over the decade but difficult toaccurately characterize the direction current smoking will take during the nextdecade.

Reducing the prevalence of current smoking among adolescents to16% is one of the goals of the Leading Health Indicators. Achieving this goal by2010 will require a 54% reduction in current smoking among adolescentsnationwide. Data from Florida, where comprehensive tobacco-control programs havebeen initiated, suggest such declines are possible. From 1998 to 2000 inFlorida, current smoking declined 40% among middle school students and 18% amonghigh school students.

The CDC recommends that communities fully implement its"Best Practices for Comprehensive Tobacco Control Programs" byestablishing comprehensive, sustainable, and accountable tobacco-controlprograms. In addition, communities should follow the CDC’s "Guidelinesfor School Health Programs to Prevent Tobacco Use and Addiction," whichrecommend implementing school-based tobacco-use prevention programs in grades K-12,with intensive instruction in grades 6-8, and supporting cessation efforts fornicotine-dependent students. Finally, comprehensive tobacco-control programsshould also reduce the appeal of tobacco products, implement mass mediacampaigns, increase tobacco excise taxes, implement policy and regulation oftobacco products, and reduce youth access to tobacco products.

This commentary was reported by the Office on Smoking andHealth, and Division of Adolescent and School Health, National Center forChronic Disease Prevention and Health Promotion, CDC. Adapted from Morbidity andMortality Weekly Report 49(33):755-758, 2000.