PRE VE NTIVE ME DICINE 27, A19±A28 (1998)
ARTICLE NO. PM980379
Self-Initiated Quitting among Adolescent Smokers1
Steve Sussman, Ph.D.,*,2Clyde W. Dent, Ph.D.,* Herbert Severson, Ph.D.,² Dee Burton, Ph.D.,³
and Brian R. Flay, D.Phil.³
*Institute for Health Promotion and Disease Prevention Research, University of Southern California, Los Angeles, California 90033;
²Oregon Research Institute and University of Oregon, Eugene, Oregon 97403; and ³Prevention Research Center,
University of Illinois at Chicago, Chicago, Illinois 60607
INT R ODUCT ION
Objectives. T his paper reviews the literature regard-
ing predictors of adolescent self-initiated smoking ces-
sation and investigates self-initiated smoking cessa-
tion among a large sample of alternative high school
youth in southern California.Youth transfer to alterna-
tive schools because of academic or behavioral prob-
lems, and they are at relatively high risk for ciga-
Methods. Several demographic (e.g., gender), behav-
ioral (e.g., level of smoking), and psychosocial (e.g.,
risk-taking) predictors of adolescent smoking cessa-
tion were investigated. T he alternative high school co-
hort provided a sufficient sample size of quitters (de-
fined as no use in the past 30 days, measured after a
1-year period) to permit a prospective examination of
adolescent smoking cessation.
R esults. Although nine demographic, behavioral,
or psychosocial variables discriminated among quit-
ters and nonquitters in univariate analyses, only
level of baseline smoking, smoking intention, and
perceived stress were predictors in a final multivari-
Conclusions. Based on the literature review and find-
ings among the cohort, smoking cessation programs for
adolescents should include counteraction of problem-
prone attitudes, support of wellness attitudes, provi-
sion of motivation to quit strategies, and assistance
with overcoming withdrawal symptoms.
The recent upturn in smoking among high school
seniors underscores the importance of continuing re-
search todetermine the factors that influence smoking
cessation among adolescents. Therateof daily smoking
decreased among high school seniors from 29 to 20%
between 1977 and 1981, declined only 2% from 1981 to
1991, then increased nearly 1% per year thereafter
[1,2]. As theearlier a person begins tosmoke cigarettes
the more likely it is that he/she will continue using
them as an adult [e.g., 3±5], effectivesmoking cessation
programs for adolescents arevery much needed. Unfor-
tunately, smoking cessation in this age group is poor:
approximately 75%ofteenagedaily smokers will smoke
asadults[1,5].Sadly, somephysical damagefromsmok-
ing begins in adolescence[6,7], making cessation efforts
even more important.
Many adolescents who smoke regularly want to quit
[7,8]. Indeed, between 55 and 65% of smokers ages
12 to 18 years report having tried to stop. Percentages
in this range have been reported in studies of compre-
hensive (regular) high school youth in the United
States [9±12], in a study of smokers in U.S. youth
detention centers , and in a national survey of
Canadian youth . A national survey found that
approximately 45% of adolescent smokers wanted to
quit ªsoonº , but this desire may not be grounded
in a true intent to change behavior. For example,
although 49% of a sample of 130 adolescent weekly
smokers (9th through 12th grade) reported wanting
to quit in the next 6 months, only 18% of the sample
indicated they were ready to take action and quit in
the next 30 days . Identifying relevant variables
to facilitate quit attempts has been considered im-
portant for some time [5,7,16,17] but needed research
has not been completed.
Key Words: self-initiated smoking cessation.
1This research was supported by the National Cancer Institute
Grant CA44907, the National Institute on Drug Abuse Grant
DA07601, and the California Tobacco-Related Disease Research Pro-
gram Grant GRT-0182.
2Towhom reprint requests should be addressed at IPR-USC, 1540
Alcazar Street, CHP-209, Los Angeles, CA 90033. Fax: (213) 342-
2601. E-mail: firstname.lastname@example.org.
Copyright ? 1998 by American Medical Association
All rights reserved.
SUSSMAN ET AL.
Quit Attempts: Perceptions of Difficulty in Quitting
was not found in several studies of self-initiated cessa-
tion among adults . Not surprisingly, intention to
smoke in the future has been found to be inversely
related to quitting . Reports of relatively pleasant
physiological reactions totobacco, which predict escala-
tion in use, also may be related to a smaller likelihood
of cessation [21,25].
A wide range of reports have been obtained in which
youth were asked about the ease of quitting smoking.
Some studies [e.g., 18,19] indicate that youth believe
smoking cessation to be quite easy; others show young
people to be less sanguine. For example, in a Missouri
state-widesurvey , 23%ofthosewhoreportedsmok-
ing in the past week reported it would be very difficult
to quit. Similarly, in the CDC's 1993 second Teenage
Attitudes and Practices Survey, 18% of 10- to 18-year-
old monthly smokers and 74% of daily users reported
that it was ªreally hard toquitº . A recent study 
revealed that only 43% of a sample of 1,430 southern
California adolescent smokers reportedwith confidence
that they would ever quit smoking.
How ability to quit matches up with intentions to
quit or beliefs about the ease of quitting are questions
worth pursuing. In one study, approximately 5% of
young smokers believed they would be smoking 5
years later, but at 8-year follow-up 75% were smoking
. It seems likely that differences in smoking behav-
ior will influence perceptions of ease of quitting. Occa-
sional, younger users could well imagine that quitting
is easy since they experience either no withdrawal
symptoms or only a few. On the other hand, regular
smokers may express pessimism about the likelihood
of stopping smoking [5,21]. The fact that withdrawal
symptoms are present even among youth who smoke
seven or fewer cigarettes per day [17,21,22] provides
ample evidence of the strain of quitting for many
Demographic variables and alcohol use.
that have been studied in relation to quit rates among
adolescents include age of initiation, gender, ethnicity,
socioeconomic status, and alcohol consumption. Gener-
ally, if a wide range of ages (e.g., 12 to 20 years) is
included, then younger age of initiation generally is
associated with lower quit rates . Female gender has
been associated with lower smoking quit rates in some
studies [8,20,26], but not in others [3,23,27].
White ethnicity has been associated with lower quit
rates [1,2], but this relationship may be attributable to
younger age of initiation . In any event, by middle
adulthood (35 to 60 years) the rates for the white sub-
population are higher than for other groups (e.g., Afri-
can Americans ). Finally, lower socioeconomic sta-
tus has been associated with lower quit rates [e.g., 26].
A study on drug use found that quitters are more
likely than those whodonot quit toreport not drinking
alcohol in the past month . However, another study
failed to find any relationship between use of drugs
other than nicotine and quit status [e.g., 9]. One may
conjecture that persons who are less reliant on drug
use for functioning will be more likely toquit smoking;
however, those who quit use of other drugs also may
be relatively likely to remain a smoker in recoveryÐ
perhaps considering it an immediately more safe
Predictors or Correlates of Quitting
Some self-report cessation studies have investigated
quitter±smoker differences; others have examined pro-
spectivepredictors ofquit status.In Table1, a summary
of the characteristics of 10 major prospective studies of
self-initiated smoking cessation is presented. Pre-
dictors investigated in one or more of these studies
includesmoking history, demographic variables and al-
cohol use, reasons for quitting, and psychosocial vari-
and antismoking attitudes.
Reasons for quitting.
quently cited in retrospective reports of adolescents as
reasons for quitting [8,13,31,32] or as a reason for want-
ing to quit [12,33]. Social reasons including perceived
peer and family pressures to quit also are mentioned
frequently [12,13,32]. Intrapersonal reasons, which in-
clude personal estimates of cost and addiction, are also
cited frequently [12,32,33].
Various health effects are fre-
ceived use among friends, peer approval if one smokes,
enforcement of peer norms, estimates of the prevalence
of peer smoking, and endorsement of smoking-related
social images (e.g., smoking to show one is a sensation
seeker) have all been examined for their effects on quit
rates. It appears that direct social influence may vary
by age as a predictor of smoking considering that
greater motivation to comply with peer requests ap-
pears negatively related to quitting among middle
school youth, as might be expected, but it is positively
Frequency of cigarette offers, per-
Smoking history and intention to smoke in the fu-
ture.Heavier smokers are less likely to quit than
lighter smokers, and those whohave smoked for a rela-
tively long period are less likely toquit than those with
a briefer experience [8,9,16]. Furthermore, even con-
trolling for self-reports of degreeof tobaccodependence,
those who begin smoking at a younger age are less
likely to quit than those who started later . Among
adolescents,agreater number ofquit attemptsisassoci-
ated with higher quit rates [16,23], but this correlation
SELF-INITIATED QUITTING AMONG ADOLESCENT SMOKERS
TABL E 1
Summary of Methods of Prospective Studies of Self-Initiated Quitting among Adolescents
collection first and
Number of Percentage Program
quitters of quitters exposureAge or grade rangeDefinition of smoker
3 months 1981 3215 to 16.5 years 392 Identify as a ªsmokerº 129
yes or no; mean ?
7 cigarettes per day
Monthly; 77% were
Smoked in past week
in a whileº
Chassin, Presson, 1 year
1981±198250Grades 6±11 17833 18.5No
Ary, Biglan, 1988 1 year
Laoye, Cresswell, 2 years
Grades 7, 9, and 10
446 of full
26 regular No
37 of full
3 years 1980±1982 50 Grades 7±12642; perhaps Smoked once or
1056Monthly smoking or
greater; 27% of the
551Smoked in past
of the sample had
smoked in the past
month at pretest
of the sample were
weekly smokers at
766Smoked in past
smoked within the
last month at
18 regularEver tried smoking
259 triers or
Chassin, Presson, 7 years
1980±1988 50Grades 6th through
(mean age at last
wave ? 21.8
10- to 12-year-old
Alexander et al.,
1 year 1979±198050242 44Yes
Sussman et al.,
1 year 1991±199250 8th to 9th 2837Yes
Sussman et al.,
1 month1994±19956015 to 18 year olds
(mean age ?
12 years 1976±198829Grades 7±9 through
(mean age at last
wave ? 25.5
related to quitting among high school youth . Per-
haps fewer peers makedirect offers of cigarettes in high
school, or more youth may verbally discourage use. It
is known that those youth who report receiving a rela-
tively greater number of cigarette offers are less likely
to quit .
Quit rates are lower among adolescents who report
use among friends [e.g., 9,16,26,29,34±37], with one ex-
ception in theliterature . Correspondingly, peer ap-
proval of smoking is related tolower quit rates for mid-
dle school youth ; the effect on quit rate for high
school youth remains tobe determined. The perception
that friends are relatively less strict about standards
of behavior has been found to be marginally predictive
SUSSMAN ET AL.
of quitting , which may seem counterintuitive.
Overestimates of smoking among one's peers has been
hypothesized to be inversely related to quitting, but
this hypothesis has not been confirmed . Finally,
higher quit rates or decreases in smoking have been
found in some studies for those holding less favorable
smoking-related social images , but not in others
of thinking that cigarette and alcohol use were wrong
had higher quit rates . A third study found that
disapproval of cigarette advertising was associated
with greater quit rates . Finally, other researchers
have found that youth who self-identify as members of
high-risk groups (in which 50% or more are regular
smokers) are less likely to smoke, themselves, if they
place importance on health as a value .
Health knowledge does not predict quitting [e.g.,
26,32], even though quitters hold more negative beliefs
than current smokers about the psychological and
health consequences of smoking . In summary, in
the few studies that have been completed, attitudes
unfavorabletocontinued smoking, and negativebeliefs
about the health effects of smoking, appear as consis-
tent predictors of quitting among adolescents.
use, and family social support have all been studied as
correlates of cessation. Family use is associated with
lower quit rates among young adult smokers ,
among daily or monthly adolescent smokers [26,29,38],
and, when father's use only was considered, among ex-
perimental adolescent smokers .On theother hand,
parental smoking was associated with greater quitting
among adolescent experimental smokers in the 1982
Surgeon General's report .
Parental disapproval of smoking predicts quitting,
especially among younger smokers . Finally, per-
ceivedparental social support has been foundtopredict
quitting among middleschool youth, a finding that still
held at 7-year follow-up [34,38]. Attachment to father
and greater parental supervision are associated with
higher quit rates , and greater perceived parental
expectationfor one's academic
Family use, family disapproval of
Self-Quitting among High-Risk Youth: Project TND
The present empirical study involves a longitudinal
cohort of youth from 21 continuation high schools, 1
from each of 21 southern California school districts,
whowereadministeredboth a baselineassessment and
a follow-upassessment 1year later (Project Towards No
Drug Abuse, Project TND). Continuation high schools
enroll youth who have transferred out of the regular
system due to academic or behavioral problems (e.g.,
lack of credits, drug use) [44,45]. Approximately two-
thirds ofthestudents in thecurrent study wereexposed
to drug abuse classroom prevention programming as
part ofother researchbut nonewereexposedtosmoking
cessation programming . For this analysis, self-
reported quitting was examined while controlling for
other program involvement.
Coping and rebelliousness.
ers often describe the habit as a stress reducer, the
relationship of adolescents' stress management skills
to their smoking behavior is an important area for in-
vestigation. Unfortunately, most of theresearch on cop-
ing with stress and smoking cessation has been con-
ductedamong adults.Among young people, researchers
have found that lower levels of emotional distress and
greater coping and social skills are related to higher
quit rates[41,42].Correspondingly, levelsofself-esteem
are related to quitting among adolescents .
Risk-taking, not believing in obeying the law, and
lack of religiosity all consistently predict who becomes
and remains a long-term smoker [e.g., 36,43]. In addi-
tion, failure to participate in organized activities at
school or elsewhere in the community has been found
to be associated with lower quit rates later in life .
ME T HOD
The continuation schools were all located in a five-
county region of southern California. Thetotal baseline
sample of 2,001 consisted of 423 students who under-
went an anonymous collection procedureand1,578 stu-
dents whoweretobetrackedconfidentially but individ-
ually over time. The anonymous subjects, 217 (51%) of
whom reported smoking in the past month at baseline,
werethosefor whom parental consent was not obtained
for individual-level longitudinal measurement. The
present longitudinal study included only those who re-
ceivedtheconfidential collection procedure;anonymous
and confidential collection students did not differ on
baseline characteristics .
All participating subjects were asked at baseline
whether they had smoked in the past month (30 days);
of theconfidential collection baselinesample, 889 (56%)
reported they had. Thepresent study further examined
593 (67%) of these smokers, all of whom were followed
Attitudes about smoking.
inedtheimpact of attitudes about themorality of smok-
ing, the importance that adolescents place on health as
a value, and knowledge and beliefs about consequences
of smoking. One prospective study found that smokers
who agreed that society has the right to do something
about smoking (e.g., ªteachers should set a good exam-
ple by not smoking cigarettesº) were more likely toquit
. Another study found that adolescents whose be-
liefs changed relatively more over time in the direction
Researchers have exam-
SELF-INITIATED QUITTING AMONG ADOLESCENT SMOKERS
up 1 year later. An attempt was made to follow up
at 1 year all 889 current smokers in the confidential
collection group, often making repeated attempts. Suc-
cess was achieved in reaching at school 136 subjects
(approximately 15% of the targeted sample). Success
was achieved in reaching by telephone the homes of
61% of the targeted sample (after a mean of seven at-
tempts to reach them, SD ? 7.5, to an average of 1.6
telephone numbers, SD ? 1.0). In all, 3% of the youth
or their parents refused to continue participation after
they were finally reached by telephone. Another 6% of
the students in the targeted sample were not available
for interview (either they or their family had been
reached at least once but the subject was not available
at those times for the interview). Thus, success was
achieved in following up 67% of the targeted sample, a
rateonly slightly lower thanthat obtainedwithsamples
from traditional schools at 1-year follow-up (75%) .
drugs in the last month? (In the last 30 days).º Defini-
tions of cessation must include a measure of the dura-
tion of quitting. Generally, adult and adolescent smok-
ers smoking less than one cigarette in the past month
areclassified as ex-smokers [8,16,24]. Thebaselinecur-
rent smoking item and the 1-year follow-up time points
were used to create two smoking status categories:
those who reported having smoked in the past 30 days
at baselinebut not at follow-up wereclassified as ªquit-
ters,º and those who reported smoking at both time
of quitters and nonquitters were large enough to exe-
cute a series of regression models that predicted quit
status from various of baseline variables.
For the 593 subjects who reported smoking at base-
line and who were surveyed 1 year later (the analysis
sample), ages varied from 14 to 19 years; 93% were 16
to 18 years of age (mean age ? 16.7 years, SD ? 0.8).
6% Asian American, 4% African American, 3% Native
American, and 4% ªotherº; 45% used both English and
another language, although only 1% reported a prefer-
ence for a second language; and 46% lived with both
parents. Of the youths' parents, 65% had completed
Current Smoking and Quit Measure
Thosewhoreportedany cigarettesmoking within the
past 30 days were classified as current smokers [e.g.,
38]. An 11-category rating scale was used to assess
smoking andother drug use(seeTable2).Subjects were
asked, ªHow many times have you used each of these
TABL E 2
Measures Used in Project TND
Number of items
Number of response options
for each item in measure
? or r of
Live with both parents
Current alcohol use
Current marijuana use
Current hard drug use
Friends' cigarette use
Peer approval of drug use
Prevalence estimates of peer smoking
Fear of victimization
Morality of drug use
Health as a value
Program success expectancies
Derived from birth date
6 to 9
r ? 0.63
r ? 0.59
aIndividual items were examined if ? ? 0.50; Ð , not applicable to this item.
SUSSMAN ET AL.
high school; modal occupations among the fathers were
skilled or semiskilled labor (42%), among the mothers,
unskilled labor or housework (32%). Use of alcohol,
marijuana, and hard drugs in the past month was re-
ported by 78, 71, and 45% of the sample, respectively,
indicating that this group of adolescents had a higher
prevalence of drug use than high school youth from
nearby schools or high school youth nationwide .
conflict , and fear of victimization (adapted from
J ensen and Brownfield ).
The fourth class of baseline predictors included vari-
ables related toindividual differences: morality of drug
subscale ), health as a value , perceived stress
(derived from the Perceived Stress Scale ), depres-
sion , and program success expectancies .
Health as a value was originally tested by Lau et al.
. Depression was measured as the mean of
the scores on the 20-item Center for Epidemiological
StudiesÐ Depression Scale .
A description of the measures used for Project TND
is provided in Table2. For measures with threeor more
items, a coefficient ? is provided. For those with two
items, a correlation (r) is given.
At baseline, all subjects wereadministered a 20-page
self-report questionnaire, which had a core section at
the front, containing items designed to elicit demo-
graphic and behavioral information, followed by knowl-
nairewas administered1 year later. Tobeconservative,
a biochemical validation protocol (a ªpipelineº proce-
dure) was used . Because self-reports of cigarette
use of students receiving this procedure did not differ
from those of anonymously surveyed students , the
self-reports were used in the analysis.
During the 1-year follow-up data collection, for sub-
jects who were still enrolled at their high school, data
collectors traveled to the school to survey them there
(23% of those surveyed). Most subjects (77%), however,
weresurveyed by telephone. Whether or not thesubject
could be reached, the maximum time spent trying to
follow up a subject from a given school was 4 months,
which included first tracking the subject at school and
then outside of school.
ANALYSIS AND R E SULT S
To assess the potential bias introduced by failure to
obtain parental consent for tracking and inability to
follow up some of the consented subjects, the baseline
means for theanalysis subsampleof smokers werecom-
pared with those for the full measured baseline sample
of smokers by using a series of single-sample t tests.
There were few statistically significant differences
found between the means of those measured at both
occasions(n ?593) andthemeans ofall thosemeasured
only at baseline(anonymous and confidential collection
combined n ? 1,106). Regarding the measures exam-
ined in this study, the only differences were that sub-
jects in the analysis sample were slightly less likely to
be from single-parent homes and slightly more likely
to be hard drug users, but these differences were very
small in magnitude (less than 0.04 standard deviation
units). Thus, theprimary results probably wouldgener-
alize to the general population of continuation high
school smokers. To test whether the sample selection
restrictions on these twovariables led tomodel specifi-
cation bias, theadjustment method suggested by Heck-
man was used . All models described below were
run with andwithout theadjustment factor, but results
failed to differ after this adjustment.
Four classes of baseline predictors were examined in
models that predicted nonquitters versus quitters. The
first class consisted of demographic measures: binary-
coded ethnic comparisons (each group compared with
all others), age, gender, socioeconomic status, living sit-
uation, and acculturation. A parental socioeconomic
status (SES) index (across mother and father) was cre-
ated; the index adapted the occupational and educa-
tional categories of Hollingshead and Redlich [32,47].
The assigned SES index was the mean across the four
items (two for each parent). An acculturation index,
based primarily on language of preference and devel-
oped by Marin and colleagues , was modified
slightly for use in this study.
Thesecond class of baselinepredictors included mea-
sures related to current drug use: current cigarette
smoking (previously defined), smoking intention, alco-
hol use, marijuana use, hard drug use , and addic-
tion concern . Included in thethird class of baseline
predictors wereperceivedsocial variables: friends'ciga-
rette use, peer approval of drug use, prevalence esti-
mates of peer smoking , general assertiveness (de-
rived from Gambrill and Richey's measure ), family
Prediction of Quitting at Follow-up
The overall quit rate at follow-up (percentage that
did not smoke in the past 30 days) for the final study
sample of 593 baseline smokers was 21%. Quit status
(yes or no) was predicted with a random regression
model (PROC MIXED) procedure . Linear regres-
sion results are equivalent tologistic regression in this
analysis because more than 20% of the sample were
quitters. The intraclass correlation for schools was
SELF-INITIATED QUITTING AMONG ADOLESCENT SMOKERS
small (0.002). A two-stage analysis protocol was com-
pleted. Thedependent variable was always quit status.
The first set of models examined the prediction of quit
status from single predictors while controlling for nui-
sance factors (prevention study condition , schools,
and collection method). In the second stage, all signifi-
cant predictors from the previous stage were placed
in the same simultaneous model (significance for this
purpose was considered to be P ? 0.1).
Among the drug use-related variables, baseline
smoking and smoking intention (both P ? 0.05) and
addiction concern (P ? 0.1) were significant. The mean
score of baseline cigarette smoking among quitters was
3.7 (30±40 cigarettes in the past month; SD ? 3.7);
among nonquitters the score was 6.9 (approximately
70cigarettesinthepast month;SD ?3.6).Onlikelihood
of smoking in thenext year, quitters (mean ? 3.5, rated
approximately as ªa littlelikelyº; SD ? 1.4) had a lower
mean score than nonquitters, whose mean score of 4.4
(SD ? 1.1) was in theªsomewhatº toªvery likelyº range.
The mean score of baseline addiction concern, slightly
lower among quitters, was 1.2 (SD ? 0.4); among non-
quitters the mean score was 1.4 (SD ? 0.4).
Ofthefiveperceivedsocial variables, only thefriends'
cigarette use item was a significant predictor of quit-
ting. Quitters reported a mean of 3.7 (of 5 closest)
friends who smoked (SD ? 1.6), nonquitters, a mean
of 4.1 (SD ? 1.3). Four of the six individual difference
variablesÐ morality of drug use, health as a value, per-
ceived stress, and program success expectanciesÐ were
significant. Quitters had a lower mean score (mean ?
2.4, SD ? 1.0) on morality of drug usethan did nonquit-
ters (mean ? 2.6, SD ? 1.0); a lower score indicates a
stronger belief in the immorality of drug use. Quitters
also scored slightly higher on health as a value
(mean ? 1.5, SD ? 0.3) than nonquitters (mean ? 1.4;
SD ? 0.2). In addition, quitters had a slightly lower
perceived stress score (mean ? 1.4, SD ? 0.4) than
nonquitters (mean ? 1.5, SD ? 0.4). Finally, while the
scores of both groups indicated that many of the sub-
jects learnedfromschool programs that year something
they believed would help prevent their future drug use,
quitters reported a higher mean level of confidence
(mean ? 2.0, SD ? 0.6) than nonquitters (mean ? 2.2,
SD ? 0.6); a lower mean indicates greater confidence.
Theseresults areshown in Table3. For ethnicity, two
of the six regression models were significant: Latinos
were more likely to quit than other groups (58% of the
quitters were Latino, and 39% of the nonquitters were
Latino), and whites were less likely toquit than others
(24%ofthequitters werewhite, and38%ofthenonquit-
ters were white).
TABL E 3
Project TND Self-Initiated Quitting Results
Measure Partial r square
Live with both parents
Hard drug use
Perceived social variables
Friends' cigarette use
Peer approval of drug use
Prevalence estimates of peer smoking
Fear of victimization
Individual difference variables
Morality of drug use
Health as a value
Program success expectancies
Second Stage: Multivariable Model
All significant predictors (Latinoethnicity, whiteeth-
nicity, baseline smoking, smoking intention, addiction
concern, morality of drug use, health as a value, per-
ceived stress, and program success expectancies), as
well as prevention study condition , an attrition
composite variable, and method of collection, were en-
tered in a simultaneous multivariable analysis. The
significant predictors of quit status in this model were
level of baseline smoking (F [df ? 1,557] ? 30.19,
P ? 0.001) and smoking intention (F [df ? 1,557] ?
5.90, P ? .015). Methodof data collection (F [df ?1,557]
?3.65, P ?0.056) andperceivedstress(F [df?1,557] ?
3.46, P ? 0.063) weremarginally significant predictors.
Note. The partial r square is the percentage of variance accounted
for by a variable, controlling for nuisance factors (prevention study
condition, school, and method of collection).
²P ? 0.1.
*P ? 0.05.
The literature review indicates that demographic
characteristics may help to distinguish quitters from
SUSSMAN ET AL.
nonquitters (e.g., during adolescence, whites may be
less likely to quit than members of other ethnicities),
but available studies are inconsistent on this point.
Behavioral predictors provide less equivocal data;
clearly heavier smokers as well as those who have
smoked for a relatively longer duration are relatively
less likely toquit. Although not much work on theasso-
ciations of other drug use and smoking cessation has
been completed, it appears likely that entrenchment in
a drug-using milieu is associated with lower rates of
smoking cessation . In addition, friends' use has
been found to be associated with quit rates: youth who
perceive that their friends smoke may be less likely to
quit. The literature review indicates that social vari-
ables such as overestimating theprevalenceof smoking
among one's peers and endorsing favorable smoking-
related images are not consistently related to quitting.
Perhaps, reports of use by one's primary peer clique is
a less ambiguous measure of perceived social pressure
to smoke or quit .
Finally, there is some support for the predictive effi-
cacy of individual-difference variables. Relatively high
emotional distress, few coping resources, or high risk-
taking may be related to lower rates of quitting. The
most consistent support exists for variables such as
intention to smoke and relatively strong attitudes for
or against cigarette smoking.
In summary, thereview suggests that (a) heavier use
and intention to continue using, (b) perceived friends'
use, (c) favorable attitudes toward continued use, and
(d) a preference for risk-taking all predict continued
cigarettesmoking. Cessation programming should con-
front prosmoking attitudes, perhaps within cliques of
smokers , should encourage quitting right away
rather than doing so in middle adulthood, and should
consider means tailored to help heavier smokers quit.
but not in the multivariable model may be of interest
to those charged with developing cessation programs
due to possible indirect effects on quitting.
The finding that method of data collection was of
borderline significance deserves further discussion.
Classroom self-report was associated with higher rates
of quitting than was telephone report. This difference
in reporting couldbea collection artifact; perhaps those
surveyed in class, believing themselves to be more
closely monitored, were more likely to try to meet the
imagined preferences of the data collection team. On
the other hand, those surveyed confidentially versus
anonymously in classroom at baseline did not differ
in reports of smoking, probably ruling out a collection
artifact. As an alternative interpretation of these re-
sults, all those interviewed by telephone were out of
high school and may have been more stressed and thus
more likely to continue smoking. Consistent with this
interpretation, by removing the method of collection
variable from the multivariable model, the stress vari-
(P ? 0.046).
The fact that studies consistently find level of use or
intention to predict quitting suggests that measures
of tobacco dependence need more exploration among
adolescents. The few dependence measures currently
collected include self-reports of inability toquit  and
reports of the presence of withdrawal symptoms
[13,19,22]. Specific withdrawal symptoms frequently
reported among adolescents include craving, irritabil-
ity, insomnia, hunger, and difficulty concentrating [e.g.,
13,21]. More than half of adolescent smokers who try
toquit report withdrawal symptoms [7,13,17,22]; num-
ber of cigarettes usually smoked and self-reported
depth of inhalation have been found to predict with-
drawal symptoms among adolescents who smoke .
In one study, various other psychoactive substance de-
pendence criteria from the Diagnostic and Statistical
Manual of Mental DisordersÐ Third Edition Revised
also were assessed, such as spending a great deal of
time obtaining the substance, giving up important ac-
tivities because of the substance, or continued use de-
spite knowledge of adverse consequences . Much
more work is needed toascertain the relations between
measures of addiction and smoking cessation among
Project TND Data
The list of significant predictors found in the first-
stage analysis of Project TND data suggests that (a)
degree of involvement in smoking and (b) being part of
an environment that supports smoking or other drug
use or is perceived as stressful is inversely related to
quitting. Generally consistent with the literature re-
view, heavier smoking, intention to continue smoking,
and greater perceived stress emerged as the main pre-
dictors of not quitting. Other variables, including not
being Latino, being white, reporting greater addiction
concern, friends'smoking, greater attitudinal tolerance
for drug use, less belief in health as a value, and lower
expectations of school treatment program success in
combating drug use failed to add to the prediction of
quit rates beyond that supplied by smoking behavior
and intention tosmoke and perceived stress. Still, pre-
dictors that were significant in the first-stage models
Suggestions for Cessation Programming
Continuation high school students in California indi-
cate an interest in lessons that can motivate them not
tousedrugs or justify why it is in their interest tolearn
new lifeskills . Most continuation high school youth
are well aware that other persons perceive them as
even more deviant in behavior than they actually are;
perhaps educators can motivate them to stop smoking
SELF-INITIATED QUITTING AMONG ADOLESCENT SMOKERS
Eriksen MP. Trends in cigarette smoking among US adolescents,
1974 through 1991. Am J Public Health 1995;85:34±40.
3. Breslau N, Peterson EL. Smoking cessation in young adults: age
at initiation of cigarettesmoking and other suspected influences.
Am J Public Health 1996;86:214±20.
4. Taioli E, Wynder E. Effect of the age at which smoking begins
on frequencyof smoking in adulthood. N
5. U.S. Department of Health and Human Services (U.S. DHHS).
Reducing the health consequences of smoking: 25 years of prog-
ress. A report of theSurgeon General. Washington: PublicHealth
Service, 1989. [DHHS Publication No. [CDC] 89-8411]
6. Dwyer J H, Rieger-NdakorerwaGE,Seamer NK, FuchsR,Lippert
P. Low-level cigarette smoking and longitudinal change in serum
cholesterol among adolescents. J AMA 1988;259:2857±62.
7. U.S. DHHS. Preventing tobacco use among young people. A re-
port of the Surgeon General. Atlanta: U.S. DHHS, PHS, CDC,
NCDPHP, OSH, 1994. [U.S. Govt. Printing Office No. 017-001-
8. Pallonen UE. Smoking cessation in adolescence. Unpublished
manuscript, 1987. [Based on a paper presented at the Western
Area Conference on Psychosocial Research, American Cancer
Society, California Division, Los Angeles, 1987]
9. Ary DV, Biglan A. Longitudinal changes in adolescent cigarette
smoking behavior: onset and cessation. J Behav Med 1988;
10. Hansen WB.Behavioral predictorsofabstinence:early indicators
of a dependence on tobacco among adolescents. Int J Addict
11. Mills D, Ewy BM, Dizon J . Smoking cessation in high school.
Health Educ 1978;9:5±9.
12. Sussman S, Dent CW, Nezami E, Stacy AW, Burton D, Flay BR.
Reasons for quitting and smoking temptation among adolescent
smokers: gender differences.
13. Dozois DN, Farrow J A, Miser A. Smoking patterns and cessation
motivations during adolescence. Int J Addict 1995;30:1485±98.
14. Darbyshire J O. Teen smoking cessation: a background paper.
Quebec: Contract with Health and Welfare Canada, 1989.
15. Pallonen UE, Rossi J S, Smith NF, Prochaska J O, Almeida PM.
Applying the stages of change and processes of change concepts
toadolescent smoking cessation. Poster presentation at the Soci-
ety of Behavioral Medicine, Fourteenth Annual Scientific
Sessions, San Francisco, 1993 Mar.
16. U.S. DHHS. The health consequences of smoking; cancer. A
report of the Surgeon General. Washington: Public Health Ser-
vice, 1982. [DHHS Publication No. [PHS] 82-50179]
17. Instituteof Medicine(IOM). Lynch BS, BonnieRJ , editors. Grow-
ing up tobacco free: preventing nicotine addiction in youth and
adolescents. Washington: Natl Academy Press, 1994.
18. St. PierreRW, ShuteRE, J aycox S. Youth helping youth: a behav-
ioral approach to the self-control of smoking. Health Educ
19. Flay BR. Youth tobacco use: risks, patterns, and control. In: Or-
leans CT, Slade J , editors. Nicotine addiction: principles and
management. New York: Oxford Univ. Press, 1993.
20. Brownson RC, DiLorenzo TM, Van Tuinen M, Finger WW. Pat-
terns of cigarette and smokeless tobaccouse among children and
adolescents. Prev Med 1990; 19:170±80.
21. Barker D. Reasons for tobacco use and symptoms of nicotine
withdrawal among adolescent and young adult tobacco usersÐ
United States, 1993. MMWR 1994;43:745±50.
22. McNeill AD, West RJ , J arvis M, J ackson P, Bryant A. Cigarette
as a way of rejecting the stereotypic view that these
youth are ªsocial undesirables.º A second approach to
smoking cessation proceeds from the fact that many
smokers entertain upwardly mobile life goals and are
well aware that ill health can interfere with achieving
these goals. If asked, adolescent smokers may admit
that good health is an important prerequisite for
achieving life goals and that they should, therefore,
quit smoking. Getting people to rebel against negative
stereotypes and stressing the importance of good
health to those who want to improve their lot in life,
together with other motivational concepts [e.g., 60],
may be taught through interactive dialogue with the
health educator, through group games (e.g., a talk
show format), or through motivational interviewing
techniques. Through use of such techniques, smokers
might be encouraged to quit immediately rather than
wait until well into adulthood.
As these youth try to quit, they may need help in
following through after their initial decision. To help
heavier users, pharmacological adjuncts  as well as
instruction in coping with cessation should be consid-
ered. Thorough instruction in withdrawal symptoms,
including those likely tobe experienced after the acute
withdrawal phase, is needed. Also, during quit periods
youth should be able to inform school staff or their
family physician of their efforts to quit and should be
able toobtain adult support. New quitters may become
irritablefor a few weeks and havedifficulty concentrat-
ing;ifadultsboth at homeandat school couldbereason-
ably tolerant during this period, these youth may try
harder tomaintain cessation. If, as theadolescent liter-
ature indicates, repeated quit attempts are associated
with a greater likelihood of successful cessation among
adolescents, cessation information should be provided
repeatedly in school health education curricula, and
physicians shouldroutinely andrepeatedly advisetheir
adolescent patients to quit smoking.
Finally, counteraction of social influences still needs
further work in the cessation context. For example,
what happenstofriendshipsamongadolescent smokers
when one of the friends quits? Can quitters maintain
acceptanceoftobacco-usingfriends?Or will thisbediffi-
cult because smoking represents rebellion and quitting
reflects a symbolic return to conventional society?
These sociological questions need to be addressed to
better understand how to increase adolescent smoking
cessation [e.g., 62,63].
R E F E R E NCE S
1. J ohnston LD, O'Malley PM, Bachman J G. National survey re-
sults on drug use from the Monitoring the Future Study, 1975±
1993. Vols. 1 and 2. Rockville (MD): U.S. DHHS, 1996. [NIH
Publication No. 94-3810]
2. Nelson DE, Giovino GA, Shopland DR, Mowery PD, Mills SL,
A28 Download full-text
SUSSMAN ET AL.
withdrawal symptoms in adolescent smokers. Psychopharmacol-
23. Green DE. Teenagesmoking. Immediateandlong-term patterns.
Washington: Natl Inst of Education, 1979.
24. Cohen S et al. (16 coauthors). Debunking myths about self-quit-
ting: Evidence from 10 prospective studies of persons who at-
tempt to quit smoking by themselves. Am Psychol 1989;
25. Hahn G, Charlin VL, Sussman S, Craig S, Dent CW, Stacy A,
Hansen WB, Burton D, Flay BR. Adolescents' first and last use
situations of smokeless tobacco and cigarettes: similarities and
differences. Addict Behav 1990;15:439±48.
26. Laoye J A, Creswell W, Stone EB. A cohort of 12,205 secondary
school smokers. J School Health 1972;42:47±52.
27. Pirie PL, Murray DM, Luepker RV. Gender differences in ciga-
rette smoking and quitting in a cohort of young adults. Am J
Public Health 1991;81:324±7.
28. Lopes C, Sussman S, Galaif E, Crippens DL. The impact of a
videotape onsmoking cessation
women. Am J Health Promot 1995;9:257±60.
29. Alexander HM, Callcott R, Dobson AJ , Harden GR, Lloyd DM,
O'Connell DL, Leeder SR. Cigarette smoking and drug use in
schoolchildren: IV. Factors associated with changes in smoking
behaviour. Int J Epidemiol 1983;12:59±65.
30. Hurt RD, Eberman KM, Slade J , Karan L. Treating nicotine
addiction in patients with other addictive disorders. In: Orleans
CT, Slade J , editors. Nicotine addiction: principles and manage-
ment. New York: Oxford Univ. Press, 1993.
31. Lotecka L, MacWhinney M. Enhancing decision behavior in high
school ªsmokers.º Int J Addict 1973;18:479±90.
32. Sussman S, Dent CW, Burton D, Stacy AW, Flay BR. Developing
school-based tobacco use prevention and cessation programs.
Thousand Oaks (CA): Sage, 1995.
33. StoneSL, Kristeller J L. Attitudes of adolescents towardsmoking
cessation. Am J Prev Med 1992;8:221±5.
34. Chassin L, Presson CC, Sherman SJ . Cognitive and social influ-
ence factors in adolescent smoking cessation. Addict Behav
35. Hansen WB, Collins LM, J ohnson CA, Graham J W. Self-initiated
smoking cessation among high school students. Addict Behav
36. Skinner WF, Massey J L, Krohn MD, Lauer RM. Social influences
and constraints in the initiation and cessation of adolescent to-
bacco use. J Behav Med 1985;8:353±76.
37. Sussman S, Dent CW, Stacy AW, Burton D, Flay BR. Identifica-
tion of which high risk youth smoke cigarettes regularly. Health
38. Chassin L, Presson CC, Sherman SJ , Edwards DA. Four path-
ways to young-adultsmoking
psychological antecedents in a midwestern community sample.
Health Psychol 1991;10:409±18.
39. Pederson LL, Lefcoe NM. Change in smoking status among a
cohort of late adolescents: prediction and explanation of initia-
tion, maintenance and cessation. Int J
40. Spielberger CD,J acobsGD, CraneRS,Russell SF.Ontherelation
between family smoking habits and the smoking behavior of
college students. Rev Appl Psychol 1983;37:143±52.
41. Burton D, Wohl G. The joy of quitting: how tohelp young people
stop smoking. New York: Collier, 1979.
42. Stein J A, NewcombMD, Bentler PM.Initiation andmaintenance
of tobacco smoking: changing personality correlates in adoles-
cence and young adulthood. J Appl Social Psychol 1996;26:
43. Stewart L, Livson N. Smoking and rebelliousness: a longitudinal
study from childhood to maturity. J Consult Psychol 1966;30:
44. Sussman S. Development of a drug abuse prevention curriculum
for high risk youth. J Psychoactive Drugs 1996;28:169±82.
45. Sussman S, Dent CW, Simon TR, Stacy AW, Galaif ER, Moss MA,
Craig S, J ohnson CA. Effectiveness of social influence substance
abuse prevention curricula in comprehensive and continuation
high schools. Drugs Soc 1995;8:65±81.
46. Dent CW, Sussman S, Stacy AW. Theimpact of a written parental
consent policy on estimates from a school-based drug use survey.
Eval Rev 1997;21:698±712.
47. Hollingshead AB, Redlich FC. Social class and mental illness.
New York: Wiley, 1958.
48. Marin G, Sabogal F, Marin BVO, Otero-Sabogal R, Perez-Stable
EJ . Development of a short acculturation scale for Hispanics.
Hispanic J Behav Sci 1987;9:183±205.
49. Sussman S, Dent CW. The correlates of addiction concern among
adolescents at high risk for drug abuse. J Subst Abuse 1996;
50. Gambrill ED, Richey CA.Anassertioninventory for useinassess-
ment and research. Behav Ther 1975;6:550±61.
51. J ensen G, Brownfield D. Gender, lifestyles, and victimization:
beyond routine activity. Violence Victims 1986;1:85±99.
52. Shields IW, Whitehall GC. Neutralization and delinquency
among teenagers. Criminal J ustice Behav 1994;21:223±35.
53. Simon TR, Stacy AW, Sussman S, Dent CW. Sensation seeking
and drug use among high risk Latino and Anglo adolescents.
Pers Individ Differ 1994;17:1083±90.
54. Cohen S, Kamarck T, Mermelstein R. A global measure of per-
ceived stress. J Health Social Behav 1983;24:385±96.
55. Radloff LS. The CES-D scale: a self-report depression scale for
research in the general population. Appl Psychol Measure 1977;
56. Lau RR, Hartman KA, Ware J E. Health as a value: methodologi-
cal and theoretical considerations. Health Psychol 1986;5:25±43.
57. Heckman J . Sample bias as specification error. Econometrica
58. SAS/STAT guide for personal computers, version 6 edition. Cary
(NC): SAS Inst., Inc., 1987. [computer software]
59. Sussman S, Dent CW, Stacy AW, Craig S. One-year outcomes of
Project Towards No Drug Abuse. Prev Med 1998;27:632±42.
60. Aubrey LL.Motivational enhancement therapy (MET) for adoles-
cent substance abusers. Paper presentation for the TobaccoCes-
sation Project for Youth, AMA±CDC, Chicago, 1996 Feb.
61. Schubiner H. Tobacco cessation and youth: brief office interven-
tionfor adolescents.Paper presentationfor theTobaccoCessation
Project for Youth, AMA±CDC, Chicago, 1996 Feb.
62. Eckert P. Beyond the statistics of adolescent smoking. Am J
Public Health 1983;73:439±41.
63. Sussman S, Dent CW, Stacy AW, Burciaga C, Raynor A, Turner
GE, Charlin V, Craig S, Hansen WB, Burton D, Flay BR. Peer
group association and adolescent tobacco use. J Abnorm Psy-
status: adolescent social±