TABLE 33.4ontinued

Snuff use

Chewing tobacco
    use

Any smokeless
tobacco use

  South Carolina                 0.7           5.3

Georgia                     1.4           7.3

  Florida                      1.1            1.9

East South Central Division           2.7            9.4

  Kentucky                   3.2           11.2

  Tennessee                   1.7           9.3

  Alabama                    I .7           6.6

Mississippi                   5.7           11.4

West South Central Division          4.0           5.5

  Arkansas                    6.0           9.5

  Louisiana                    2.5           5.8

  Oklahoma                   4.8           6.7

  Texas                      4.0           4.6

West Region                    1.4           3.3

Mountain Division               2.3            5.4

  Montana                    5.5           8.3

  IdhO                     2.3           6.7

  Wyoming                   .3.4           13.0

  Colorado                   1.2           6.4

  New Mexico                 5.3           5.2

  Arizona                    2.0           3.8

  Utah                      0.9           3.0

  Nevada                     1.5           2.8

Pacific Division                 I .o           2.6

Washington                  I .a           6.1-

  Oregon                     2.7           5.4

  California                   0.7           1.7

  Alaska                     2.5           6.3

  Hawaii                     0.2           0.4

     -
6.1

8.7

2.9

11.6

13.6

10.3

8.3

16.5

9.1

14.7

8.0

11.0

8.2

4.5

1.5

13.7

8.7

15.8

7.5

10.2

5.4

3.7

4.3

3.4

7.1

7.6

2.3

8.8

0.7

SOURCE: CPS 1985 (Marcus et al., in press.)

From 1964-86, there was an go-percent decline in prevalence of both cigar and pipe
smoking among men (Figure 7). The prevalence of cigar smoking declined from 29.7
to 6.2 percent; the prevalence of pipe smoking declined from 18.7 to 3.8 percent.
Reasons cited to explain the drop in cigar sales include the effects of the antismoking
campaign (several airlines have completely banned cigar and pipe smoking on all flights
for many years, but only one airline has done so for cigarette smoking), declining image
of cigar smoking, failure to attract new smokers, insufficient free-sample distribution,
mediocre advertising and promotional activities, and declining quality of the product
(Lazarus 1979).

326


TABLE 34.-Cigar/pipe smoking status (%) by major so&demographic
      variables, United States, 1986

      Cigar/pipe smoking status           Sample
Current user   Former user    Never user     Total      size

Total

Sex

Male

Female

Age

17-19

20-24

25-44

45-64

265

Race

White

Black

Other

Region

Midwest

Northeast

Southeast

West

Marital status

Married/cohabiting

Widowed

Divorced/separated

Never married

Unknown

Education

211 years
12 years
13-15 years
2 16 years

4.3

22.2

73.5

8.7       41.8        49.6

0.3        4.5        95.2

1.5        13.6        85.0

2.0        16.6        81.4

4.4       22.2        73.3

5.9       26.5        67.6

3.9       22.6        13.5

4.4       23.4        72.2

3.7        13.9        82.4

3.5        19.4        77.1

4.8       22.6        72.6

4.6        19.6        75.7

3.8       23.2        73.0

4.1        22.7        73.2

4.8       25.3        69.9

I .8        8.9        89.2

5.6       20.1        74.3

2.8        17.7        79.4

12.4       27.5        60.1

4.9       22.8        72.3

3.6       20.0        76.5

3.9       22.5        73.6

5.3       26.0        68.7

100



100

100



100

100

100

100

100



100

100

100



100

I00

100

100



loo

100

100

100

100



100

100

100

100

13,031

6.317

6,654

560

1,086

5,802

3,616

1,967

11,563

1,096

372

3,236

2,968

4,301

2,526

8,364

1,011

1,446

2,179

31

2,431

4,872

3.1 I8

2,610

327


TABLE 34.-Continued

                    Cigar/pipe smoking status          Sample
                 Current user   Former user    Never user     Total      size

Household income (dollars per year)

<lO.OOo              3.1        16.8        80.1        100     1,220

10,00&19,999           4.0       21.2        74.9       100     2.204

2o,ooo-29,999           4.3       23.1        72.6        100     2,853

30,00&39,999           5.0       24.2        70.8       100     1.735

240,000               5.5        28.1        66.4        100     2,947

Unknown              3.3        17.1         79.6        100     2,072
Poverty levela

Above               4.7        23.9        71.4        100     9,913

Below               3.0        18.6        78.3       100     1,046

unknown             3.3        17.1        79.6       100     2,072

LPoveny level is based on the definition provided by the U.S. Bureau of the Census.
SOURCE: AUTS 1986 (US DHHS, in press, a).

8ol

70



60 i

cl Ciirette
O ???
?? Pipe

Year

FIGURE 7.-Trends in prevalence of using cigarettes, cigars, and pipes, adult
      men, United States, 1964-86
SOURCE: AUTSs (US DHHS 1988).

328


PART II. CHANGES IN KNOWLEDGE ABOUT THE DETERMINANTS
             OF SMOKING BEHAVIOR

Introduction: Historical and Conceptual Overview

This Section reviews the past 25 years' growth in scientific knowledge of the deter-
minants of smoking. Broad conceptual shifts in understanding smoking are first
reviewed by comparing current knowledge, as reflected in the 1988 Surgeon General's
Report as well as in more recent investigations. with that reflected in two previous Sur-
geon General's Reports during the past 25 years: the 1st Report, issued in 1964, and
the 15th Anniversary Report, issued in 1979.

1964 Surgeon General's Report

The first Surgeon General's Report devoted a chapter to the psychosocial aspects of
smoking and another to the issue of smoking as drug addiction or drug habituation.
These topics continue to receive contemporary attention. A third chapter in the 1964
Report discussed morphological characteristics of smokers as important determinants
of smoking (e.g., physique, somatotype, and weight). With the exception of body
weight, there has been a decline in the attention paid to these variables. The relation-
ship between body weight and smoking cessation, especially among women, has
received much recent attention (US DHHS 1988).

The 1964 Report's Chapter on Psychosocial Aspects of Smoking related smoking to
a variety of demographic factors including socioeconomic status (smoking being more
prevalent among "lower or working classes" but less prevalent among extremely poor,
e.g., unemployed groups) and gender (smoking being more prevalent among men).
With regard to gender, the Report anticipated contemporary concerns about smoking
by women (US DHHS 1980b), noting that "The proportion of women smokers has in-
creased faster than that of men smokers in recent years" (US PHS 1964, p. 363).

The 1964 Report's chapter on psychosocial aspects also linked smoking to such broad
personality factors as extraversion and orality. While some research continues to show
relationships with extraversion (e.g., Eysenck 1980; Mangan and Golding 1984), most
contemporary research focuses on more specific psychological, biological, and social
variables and their interactions. The 1964 Report noted that smoking might function
to reduce tension but reported little research related to this possibility. In contrast, the
1988 Report on nicotine addiction reviews considerable laboratory and field research
on the relationship between smoking and stress and concludes that stress increases
cigarette consumption among smokers and is related to initiation of smoking among
adolescents and relapse among abstainers (e.g., US DHHS 1988).

The 1964 Report devoted much attention to the role of nicotine in smoking behavior,
an issue that continues to be of central interest, as reflected in the 1988 Report. Both
reports concluded that nicotine is a critical and substantial determinant of smoking. The
focus in 1964, however, centered on whether smoking fit the World Health
Organization's (WHO) definition of addiction, which emphasized the importance of
physical dependence (WHO 1957). The Report concluded that there was no proof of

329


physical dependence and that smoking was a habit, as was use of cocaine, am-
phetamines, and other drugs. More recent perspectives (e.g., Pomerleau and Pomer-
leau 1984), culminating in the 1988 Report, have integrated psychosocial and phar-
macologic processes into a single model of addiction or dependence. The 1988 Report
demonstrated that there have been substantial data amassed since 1964 that confirm
that by the criteria defining addiction, nicotine should be categorized as addicting.

Although the 1964 Report did conclude that " . . . there is no single cause or explruta-
tion of smoking . . ." (US PHS 1964, p. 376), its discussion of research reflected an ex-
pectation that one or a very few key causes of smoking might be found. Along these
lines, the Report emphasized the extent to which evidence demonstrated a cause to be
sufficient. For example, in discussing evidence that smoking as a sign of masculinity
may motivate many men to smoke, it labeled as "troublesome" the fact that " . . . some,
but not so many others choose this particular means [that is, smoking] of giving
evidence of their masculinity" (US PHS 1964, p. 373). Since the 1964 Report, models
of causal inference in the behavioral sciences have changed to emphasize multiple
causes interacting to bring about complex behavior patterns, and not one cause in itself
that is necessary or sufficient.

1979 Surgeon General's Report

The 1979 Report gave much attention to prevention and to the determinants of smok-
ing and smoking cessation, devoting 9 of 23 chapters to these topics. Thus, there was
recognition of different stages of smoking behavior and of determinants varying as the
stages change. Since the 1979 Report, researchers have continued to elaborate on mul-
tiple stages in the development and cessation of smoking.

The 1979 Report also recognized that multiple factors interact to encourage and sup-
port smoking. The Chapter "Behavioral Factors in the Establishment, Maintenance and
Cessation of Smoking" posited smoking as ". . . a behavior-a highly complex act. . .
based on various biochemical and physiological processes . . ." (US DHEW 1979a, pp.
16-25). It included research on drug and nondrug factors and called smoking "the
prototypical substance-abuse dependency." The Chapter "Smoking in Children and
Adolescents: Psychosocial Determinants and Prevention Strategies" explicitly viewed
the initiation of smoking as determined by an array of factors. Likewise, the Chapter
"Psychosocial Influences on Cigarette Smoking" linked multiple factors to main-
tenance and cessation of smoking, including personality characteristics, multiple drug
use, coexisting chronic disease, price "elasticity" of consumer demand for cigarettes,
and differences among cultures in their attitudes toward smoking as personal gratifica-
tion. The importance of identifying multiple, interacting factors had been enunciated
by Schwartz and Dubitzky in 1968 in their research on smoker profiles and the influence
of multiple variables on smoking cessation, maintenance of cessation, and relapse
(Schwartz and Dubitzky 1968).

The 1979 Report's recognition of an array of determinants was reflected in a recom-
mendation for future research: "There are multiple psychosocial influences on cigarette
smoking. Multivariate research is needed . . ." (US DHEW 1979a. pp. 18-25). Multi-
ple regression analyses and causal modeling have now become much more common in

330


smoking research (e.g., McAlister, Krosnick, Milbum 1984; Mosbach and Leventhal
1988).
The 1979 Report also was noteworthy in focusing attention on systematic cessation
efforts, taking both pharmacologic and psychosocial factors into account. The exten-
sive treatment of cessation research in a separate chapter was a first for the Surgeon
General's Report and set a precedent for reviewing the intervention literature in sub-
sequent reports.

Current Views

Current explanations assume that smoking is determined by multiple causes, no one
of which is sufficient. The interplay of psychosocial and pharmacologic forces con-
tinues to occupy investigators of nicotine addiction as it does investigators of other drug
addictions. While the 1964 Report tended to see such factors as mutually exclusive,
the 1988 Report (US DHHS 1988) viewed these various pharmacologic, biochemical,
and psychosocial processes, such as conditioning, as interacting in the determination
of nicotine addiction. In fact, conditioned drug-taking behavior is now thought to be
central to the concept of addiction; physical dependence is neither necessary nor suffl-
cient (US DHHS 1988). The biological power of nicotine may make the learned be-
haviors that form smoking patterns stronger and more resistant to change. At the same
time, the plentitude of dally circumstances, activities, and emotions to which smoking
is conditioned ties this behavior to numerous rituals of daily life and contributes to the
difficulty of breaking this addiction (Fisher, Bishop et al. 1988a; Pomerleau and Pomer-
leau 1987; Russell, Peto, Pate1 1974; US DHHS 1988). This interplay between be-
havior and the pharmacologic effects of nicotine is mirrored in research on smoking
cessation, in which nicotine-containing chewing gum and behavioral interventions have
been shown to enhance one another (e.g., Hall et al. 1985; Killen, Maccoby, Taylor
1984; Schneider et al. 1983). In reviewing the evidence for defining smoking as an ad-
diction, the 1988 Report made the important point that the interplay between social, be-
havioral, and pharmacologic factors that define tobacco addiction is similar to that seen
with other drug addictions.

The continuum of smoking behavior can be viewed as occurring in different stages.
The 1964 Report identified two stages (or processes): "Taking Up" and "Discontinua-
tion." Current work identifies three major stages-development, maintenance of
regular smoking, and cessation. Several investigators have offered descriptions of
various smaller stages within smoking development (e.g., Leventhal and Cleary 1980;
Flay et al. 1983). These include, for example, preparation, initiation, experimentation,
and transition to regular smoking (Flay et al. 1983). Similarly, the process of cessa-
tion has been specified in smaller stages (e.g., Marlatt 1985; Prochaska and DiClemente
1983; Rosen and Shipley 1983). These include, for example, precontemplation (not
yet considering quitting), contemplation, action, and maintenance or relapse (Prochas-
ka and DiClemente 1983).

Evolution of theoretical models of stages in smoking over the past 25 years is depicted
in Figure 8, indicating the stages described around three periods of time, the 196Os,

331


EXPERIMENTATION

REGULAR SMOKING

CONSOLIDATED ADDICTICN

C

DlSCONTlNUATlCfd I--

RELAPSE

RECYCLING

     1964 . . . . . . . . . . . . . . . . ..." . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..-................. 1979 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..." . . . . . .._.............-" . . . . . . . . . . . . . . 1909
FIGURE K-Evolving theoretical concepts of the natural history of smoking,
       1964-89

1970s and 1980s. In 1964, only two broad stages were noted, while in 1989, as many
as nine can be observed.

Stages are not explanations of attitudes or behaviors. For example, precontempla-
tion is a description of the attitudes toward smoking and likely responses to antismok-
ing messages of the individual uninterested in stopping. It is not an explanation or a
cause of that lack of interest. Neither the sequence of stages nor the boundaries among
them are rigid. For example, a young experimenter may stop smoking without ever
making the transition to regular smoking. A smoker in the regular smoking stage is, at
the same time, a precontemplator or contemplator in the cessation stage. The regular
smoking stage is abandoned when the smoker moves into action and stops smoking.
Although the boundaries among stages and their sequence may be blurred, the concept
serves as a framework for understanding the determinants of smoking behavior. Dif-
ferent determinants are operative to different degrees during each stage.

The three broad stages of smoking and their multiple interacting determinants provide
the organization for the remainder of this Chapter. Within the stage framework, his-
torical trends in determinants are discussed primarily within three general domains.
The three domains do not constitute a model; they are a useful way to organize the deter-
minants of smoking. The first domain is composed of pharmacologic processes and
conditioning, the basic factors that interact to support smoking. The combining of these
into one domain reflects present awareness that pharmacologic processes and con-
ditioning interact to produce addiction (US DHHS 1988). The second domain includes
cognition and decisionmaking. The stages of smoking reflect appraisals of oneself, of
social experiences, and of information, such as that presented in campaigns to deter

332


smoking. The ways in which individuals process such information and make choices
about smoking have been the foci of substantial research. The third domain includes
personal characteristics (e.g., personality and demographic factors) and social context,
which includes the important influences of the social, cultural, and economic environ-
ment. Personal characteristics themselves are affected by these environmental influen-
ces and mediate their effect rather than independently determine smoking.

Table 35 presents some of the determinants, within each of the domains, that have a
strong effect on the indicated stage of smoking. As such, the table provides an outline
of the discussion that follows.

Development of Smoking

Pharmacologic Processes and Conditioning

Historically, little attention was paid to the role of pharmacologic effects of nicotine
and conditioning in the initial development of smoking behavior. For example, among
teenagers, psychosocial determinants have been assumed to play a dominant role (Table
35), as for other dependence-producing substances. Once a smoker starts to inhale,
however, it is possible that the pharmacologic properties of nicotine contribute to
continued smoking (Kozlowski 1988). A few studies have investigated the potential
role of individual-specific psychophysiological responses to nicotine and the
development of smoking (Kozlowski and Harford 1976; Silverstein et al. 1982).
Reactions to initial cigarettes and the interpretation of these reactions may predispose
individuals to continuing or not continuing smoking. Hirschman, Leventhal, and Glynn
(1984). for example, found that the initial early physical reaction was predictive of con-
tinued smoking. Dizziness was related to a rapid progression to a second cigarette,
while coughing and a sore throat were related to discontinuation.

It is not clear how long it takes for the transition from experimental to regular smok-
ing, and there is likely to be much variation (e.g., Hirschman, Leventhal, Glynn 1984).
However, results from several recent studies suggest that teenagers become more ad-
dicted to smoking than was previously believed. Survey data (Green 1979; Johnson
1986) indicate that teenagers make frequent and often unsuccessful quit attempts. Other
studies confirm that teenagers have difficulty stopping and report reasons for the dif-
ficulty-social pressure, urges, withdrawal symptoms-similar to those seen with
adults (Biglan and Lichtenstein 1984; Hansen et al. 1985; Weissman et al. 1987). Be-
cause smoking among children and adolescents is generally confined to relatively few
situations, the level of nicotine dependence is limited in this group. Nevertheless, the
reports of withdrawal symptoms and relapses among teenage smokers attest to the
strength of nicotine dependence even among those still in the early stages of smoking.

More work is needed in this area to facilitate our understanding of the development
of smoking addiction. Research on adolescent initiation has not applied the same bio-
behavioral concepts and measurement tools (e.g., plasma nicotine or cotinine levels) as
have been applied to adult smoking, Sensitive human subjects issues related to work-

333


TABLE 35.-Determinants of smoking within each domain by stage

Domain

Onset/development

  Stage

Regular use

Cessation

Pharmacologic processes
and conditioning

Initial psychopharmacologic effects en-
courage transition from experimental to
regular use

Numerous conditioned associations among
smoking, environmental events, and phar-
macologic effects of nicotine

Cognition and decision-
making

Poor awareness of long- and short-term
health consequences and addictive na-
ture of smoking

Positive characteristics arc attributed to
smokers and smoking

Personal characteristics
and social context

Inclination toward problem behaviors

Extraversion

Peer and family norms and values sup-
port smoking

Youth-oriented advertising

Health consequences are minimized or
depersonalized

Posmve characteristics are attributed to
smokers and smoking

Stress/negative affect are reduced by
nicotine

Social acceptability and peer and family    Skills for coping with stimuli associated
norms support continued smoking       with smoking

Cigarette marketing encourages and
legitimizes smoking

Withdrawal symptoms and conditioned
and reinforcing effects of nicotine
encourage relapse

Increased awareness of smoking-related
symptoms or illness

Perceived benefits of cessation

Belief in one's ability to stop

Social norms and suppott for stopping and
maintained abstinence

Economic. educational, and personal
resources to minimize stress and maintain
cessation


ing with minors must be resolved; these have slowed understanding of how depend-
ence develops.

Cognition and Decisionmaking

Knowledge of the health effects of smoking is likely to influence initiation for some
teenagers. Teenagers reported that one-third of their earliest refusals of cigarettes were
based on fear of the effects of smoking on health, attractiveness, or athletic performance
(Friedman, Lichtenstein, Biglan 1985). In early adulthood, British medical students'
rating of smoking as a "major" or "not major" health risk was associated with their
smoking status as reflected by surveys in 1972 and 198 1 (Elkind 1982). Heavy smokers
among college women evaluated health outcomes of smoking less negatively than did
nonsmokers (Loken 1982). The latter two cross-sectional studies, however, may pos-
sibly reflect the effect of behavior on cognition rather than the effect of cognition on
behavior.

Cognitive appraisals of the attractiveness or desirability of smoking or of smokers
are associated with current smoking or intentions to smoke (Barton et al. 1982;
McAlister, Krosnick, Milbum 1984), as are beliefs or attributions of the functional role
of smoking (Murray and Perry 1984). Tenth graders inclined to smoke indicated greater
congruity between the value they place on interest in the opposite sex and the extent to
which they ascribe such interest to smokers (Barton et al. 1982). Intentions to smoke
were also associated with congruity between the personal value of a characteristic and
its attribution to smokers. Murray and Perry's analyses (1984) of the functional mean-
ing of substance use by youth elucidated a variety of attributions correlating with young
people's substance use. The report that smoking was useful for relieving boredom was
most highly correlated with smoking. Data from England (Charlton 1984) demonstrate
that children who smoke compared with nonsmoking children are more likely to agree
that "Smoking keeps your weight down." This attribution was especially prominent
among older girls.

School health education programs to discourage smoking have traditionally assumed
that knowledge of the health consequences of smoking would deter adolescents from
smoking (Chapter 6). This assumption has received limited support in the prevention
literature (Thompson 1978). Despite school health education programs, children, espe-
cially those who smoke, continue to harbor several misconceptions about smoking.
These misconceptions include overestimating the prevalence of both peer and adult
smoking, underestimating the negative attitudes of their peers, and minimizing the ad-
dictive nature of smoking (Leventhal, Glynn, Fleming 1987). The overestimating of
prevalence may represent the combined influence of social context and cognitive fac-
tors in determining smoking.

Contemporary smoking prevention programs ("psychosocial prevention curricula")
emphasize knowledge of short-term consequences of smoking likely to be more per-
tinent to adolescents who have limited future orientations (Glasgow et al. 1981), and
knowledge about the variety of social influences (parental, peer, and media) that affect
the development of smoking (Flay 1985; Evans et al. 1978; Chapter 6). Decisionmak-

335


ing skills (Botvin and Wills 1985) and analysis of cigarette marketing strategies (Evans
et al. 1978) also are now taught to help youth make more informed choices.

Personal Characteristics and Social Context

Personal Characteristics

The 1964 Surgeon General's Report described as "one of the best designed studies"
(US PHS 1964, p. 365) an investigation in which heavy smokers were found to be more
extraverted than were medium smokers, who were in turn more extraverted than were
light smokers (Eysenck et al. 1960). The 1964 Report also cited two other papers with
similar findings (McArthur, Waldron, Dickinson 1958; Schubert 1960). More recent
work by Cherry and Kieman (1976, 1978) found that neuroticism and extraversion
measured at age 16 were positively related to smoking status at age 25, suggesting a
causal relationship. Their combined effects showed substantial ability to predict sub-
sequent cigarette use. Eysenck (1980) has argued that the association between smok-
ing and the personality dimensions of extraversion and neuroticism implies a constitu-
tional predisposition for smoking analogous to that seen with other drug addictions (US
DHHS 1988). Work on extraversion and smoking does seem to reflect a consistent
relationship between them (US DHEW 1979a; Ashton and Stepney 1982).

Studies have linked initiation of smoking with rule breaking in school, general delin-
quency, age at first intercourse, inadequate contraceptive use, low levels of child com-
pliance within the family, low levels of responsibility, nonconventionality, impulsivity,
rebelliousness, and previous use of alcohol and other substances (Brook et al. 1983;
Chassin et al. 1984; Jessor and Jessor 1977; Mittelmark et al. 1987; Russell 197 1; Zabin
1984). Academic success, as measured by grade point average, is strongly linked to
the rate of smoking (Johnson 1986). High school dropouts (Pirie, Murray, Luepker
1988) and high school seniors not planning to go to college (Johnston, O'Malley, Bach-
man 1987) are much more likely to smoke than are those planning higher education,
and this difference has increased over the past 10 years (Table 20). Similar factors are
observed with other drug addictions (US DHHS 1988). Jessor (1987) views this
covariation as reflecting a problem behavior syndrome. Biglan and Lichtenstein (1984)
questioned this interpretation, arguing against the inference of underlying personality
factors to explain the acknowledged covariation among smoking and other problem be-
haviors.

Peer and Family Influences

The influences of peers and parents were considerations in the 1964 Report and
remain a major contemporary issue (e.g., Krosnick and Judd 1982). Understanding of
the effect of peers has increased since the 1964 Report noted little available evidence
of their influence on the onset of smoking. It acknowledged that imitation ". . . may
play a role in inducing some, and perhaps many children to take up smoking" (US PHS
1964, p. 372). Studies noted that children of smoking parents were more likely to smoke


than children of nonsmoking parents (NIH 1975; Wohlford 1970); and smoking
teenagers were more likely to have friends who smoked than were nonsmoking
teenagers (Gordon and McAlister 1985; Levitt and Edwards 1970). The chapter on
children and adolescents in the 1979 Report (US DHEW 1979a) reviewed the influence
of social learning theory on models of the initiation of new behavior.  More recent
studies have supported the importance of peer models (e.g., Antonuccio and Lich-
tenstein 1980; Kniskem et al. 1983). The 1988 Report discussed similar factors in the
determination of other drug dependence.

The impact of peer smoking on adolescent smoking has been identified in a number
of studies (e.g., Chassin et al. 1984; Hundleby and Mercer 1987; McAlister, Krosnick,
Milbum 1984; Mittelmark et al. 1987), including their impact on initial smoking
episodes (Friedman, Lichtenstein, Biglan 1985) and continuation of smoking among
those who already have experimented with cigarettes (Biglan and Lichtenstein 1984).
These influences seem to rest on the importance of modeling of smoking, as well as on
the setting of norms among subgroups of adolescents. The importance of bidirection-
al influences in smoking and smoking cessation among young people has been noted
by Chassin, F'resson, and Sherman (1984). In some cases, a young person's member-
ship in a particular peer group may expose him or her to the example to smoke or to
quit; however, in other cases, a young person may actively seek membership in a peer
group that represents or is consistent with his or her established intentions about smok-
ing.

More recent research has both reaffirmed the importance of parent and peer influen-
ces and attempted to explore the points at which they exert their influence during the
process from onset-the initial smoking episode-to regular use (e.g., Friedman et al.
1985; Hirschman, Leventhal, Glynn 1984). The literature has tended to underscore the
role of parental example and influence for initiation of smoking by young children and
adolescents, and the primacy of peer influences among older youth. In application, this
emphasis has often translated into an almost exclusive intervention focus on the social
influences of peers for older adolescents (see Chapter 6).  Some of the intervention
programs include peer leaders chosen by their classmates (Murray et al. 1987). Kros-
nick and Judd (1982) found no evidence for decreases in parental influences on smok-
ing during adolescence, although they did find that peer influence increases during this
period. These studies often include important methodological advances wherein
interviews and self-monitoring are used to augment questionnaire data.

A growing body of literature implicates family climate or family interaction patterns
in smoking. Family characteristics such as indifference, low levels of trust, parental
restrictiveness, and low levels of parental involvement are associated with smoking as
well as with marijuana and alcohol use (Hundleby and Mercer 1987). Other research
has demonstrated that low levels of adolescent involvement in family decisionmaking
predict subsequent experimentation with cigarettes among adolescents (Mittelmark et
al. 1987). A variety of characteristics in fathers, including harsh criticism, impulsivity,
stereotyped male interests, poor ego integration, and lower levels of interpersonal re-
latedness has also been demonstrated to be associated with a greater likelihood of sons'
smoking (Brook et al. 1983). A decreased likelihood of sons' smoking was associated
with paternal affection, emotional support, attentiveness, participation in meaningful

337


conversations, and higher expectations for the sons. It appears that adolescent smok-
ing is more likely in restrictive, punitive, and unempathetic families in which children
are uninvolved in decisionmaking. On the other hand, families who provide multiple
avenues for identity formation and expression of feelings may obviate the utility of
smoking or other problem behaviors as a mode of identity expression (Jessor 1987).

Personal characteristics and attitudes may mediate peer influence on smoking as well
as other drug dependencies (US DHHS 1988). Research indicates greater impact of
peer smoking among adolescents scoring low on a measure of obedience to parental
authority and high on a measure ofrebelliousness (McAlister, Krosnick, Milbum 1984).
The interactions among social influences, personality, and smoking were highlighted
in a study in which seventh and eighth graders described the informal reference or af-
filiation groups they observed among their schoolmates and identified the group with
which they felt the closest affiliation (Mosbach and Leventhal 1988). Two of the four
groups that emerged, "hot-shots" (78 percent female, popular leaders in academic and
extracurricular activities) and "dirts" (63 percent male, characterized by problem be-
haviors such as drinking, poor academic performance, and cutting classes), were iden-
tified as primary reference groups by only 14.7 percent of respondents but accounted
for 55.6 percent of the smokers. In discriminant function analyses, a "macho" dimen-
sion was highly associated with one high smoking prevalence group, the "dirts," but
not with the "hot-shots." In contrast, academic and social leadership was associated
with the "hot-shots" but not with the "dirts." As were the "dirts," the "jocks" were also
63 percent male and high on the macho dimension but low on use of both hard liquor
and cigarettes. Adolescent smoking, then, is closely related to individual identification
with groups, but these groups differ markedly in their association with other problem
behaviors and psychosocial characteristics. Depending on group affiliation, different
personality and attitudinal characteristics may be related to smoking.

Social class differences in the onset of smoking continue to be observed as noted in
Part I of this Chapter. Racial differences in onset and prevalence and historical shifts
in these differences are also well demonstrated in the first part of this Chapter. Sussman
and colleagues (1987) in their study of psychosocial predictors of cigarette smoking
onset by approximately 1,000 white, black, Hispanic, and Asian adolescents in Southern
California demonstrated that different variables predict onset in these different groups.
A good predictor for whites but not for other ethnic groups was adult and peer models
of smoking behavior, while for blacks, risk-taking preference was a good predictor.
These findings possibly reflect unique cultural and social contexts and suggest that
tailoring socially relevant treatment components to adolescent subgroups may be
beneficial (Sussman et al. 1987).

Cigarette Marketing

Beyond the family and peer group, an important social context determinant of the
onset of smoking is the marketing of cigarettes. There have been longstanding con-
cerns about the impact of cigarette advertising on both children and adults as evidenced
by the ban on radio and television advertisements, effective in 1971. Yet, "cigarette

338


advertisements continue to appear in publications with large teenage readerships"
(Davis 1987, p. 730).

Marketing campaigns seem designed to appeal to specific personality characteristics
of groups of potential buyers. In this respect, they exemplify interactions between per-
sonal characteristics and the environment. The Marlboro brand was the leading choice
of a group of white adolescent male (48 percent) and female (38 percent) smokers sur-
veyed in Louisiana in 1981 (Hunter et al. 1986). In a sample of 306 high school stu-
dents in Georgia, Marlboro was the preferred brand of 76 percent of smokers who iden-
tified a single preferred brand (Goldstein et al. 1987). Similar findings were reported
by Glantz (1985). These figures contrast with the overall domestic market share of
Marlboro, which was 24 percent in 1987 (Titer 1988). Given the associations of rebel-
liousness and behavioral problems with adolescent smoking, as reviewed above, there
may be a relationship between the noted disparity of overall brand preference and the
emphasis on the tough independence of the "Marlboro Man." In fact, this pattern may
be a reflection of extensive market segmentation, in which specific brands are marketed
for specific gender or ethnic groups, often with campaign messages and symbols aimed
at those groups (Davis 1987). Teenage girls, relative to boys, are more likely to believe
that smoking controls weight (Charlton 1984) and are good targets for advertisements
that emphasize the desirability of being slender (Gritz 1986).

Some market segmentation appears more subtle, guided by smoker characteristics
not as apparent as race and gender. McCarthy and Gritz (1987) surveyed students in
grades 6, 9, and 12 regarding their attitudes about cigarette advertisements. Among
their findings was the closer relationship, for those youth more likely to be smokers,
between personality self-ratings and personality ratings assigned to models in cigarette
advertisements. Thus, the way adolescents see themselves appears to be related to their
attraction to certain advertisements. This congruity among psychological correlates of
teenage smoking, marketing themes, and teenage preferences is especially striking
when one considers that the tobacco industry denies that campaigns are aimed at
teenagers (Davis 1987).

Summary

The increased understanding of the multiple and interacting determinants of the
development of smoking and of the relation of these determinants to the stages of
development of smoking is a reflection of progress over the last 25 years. The delinea-
tion of stages-from onset to regular use-has been an especially influential develop-
ment (Figure 2). The development of the addictive processes in teenagers has recent-
ly become better appreciated and understood (Biglan and Lichtenstein 1984;
Hirschman, Leventhal, Glynn 1984). While information about the long-term disease
consequences of smoking has an important role in adolescent smoking initiation, aware-
ness of the short-term health consequences and the influence of peers and advertising
are now seen as more critical for adolescent decisionmaking. The effects of peers and
family are both supported. Cigarette marketing appears to target teenagers despite the
cigarette companies' reported policy efforts to restrict such advertising.

339


Regular Smoking

Pharmacologic Processes and Conditioning

Pharmacologic processes and conditioning play complementary and major roles in
maintaining regular smoking. Early theories of smoking tended to view pharmacologic
processes and conditioning as separate explanations of regular smoking (e.g., Hunt
1970; Table 35). They are now viewed as complementary and interacting processes
(US DHHS 1988). The 1988 Surgeon General's Report on nicotine addiction affirmed
the critical role of nicotine and its varied and powerful pharmacologic effects on the
central nervous system (CNS) in the development and maintenance of regular smok-
ing. This acknowledgment and its implications for intervention represent a sig-
nificant shift in perspective over the 25-year history of the Surgeon General's Reports.
Concurrently, increased knowledge of smoking as an addiction has clarified the impor-
tant role of conditioning in addiction. Conditioning and related processes link the
biological effects of nicotine to the many behaviors that make up smoking and to the
many concurrent physical and environmental stimuli that guide it.

Nicotine Addiction

The 1964 Report distinguished between drug addiction and drug habituation (US PHS
1964; Table 36) and concluded that smoking is habituation. As noted in the 1988
Report. the addiction/habituation distinction was dropped in 1964 by the WHO short-
ly after the release of the 1964 Report (US DHHS 1988).

The 1988 Surgeon General's Report on nicotine addiction noted the following three
major conclusions: (1) cigarettes and other forms of tobacco are addicting; (2) nicotine
is the drug in tobacco that causes addiction; (3) the pharmacologic and behavioral
processes that determine tobacco addiction are similar to those that determine addic-
tion to drugs such as heroin and cocaine (US DHHS 1988, p. 9). These conclusions
were based on a thorough review of research on addictive aspects of smoking extend-
ing over nearly a century.

The criteria that guided the 1988 Report's conclusion that smoking is an addiction
are summarized in Table 36. As documented by extensive research cited in the Report,
smoking meets all the criteria. Smoking is continued despite a desire to quit and, in
many cases, despite clear harm to the individual. A central criterion concerns psychoac-
tive effects of a drug on the CNS. Rapid absorption of nicotine into the bloodstream
and consequent delivery to the CNS are features common to all popular forms of tobac-
co use. Recent evidence'confirms that nicotine is absorbed by the brain, which con-
tains receptors specific for this agent (e.g., London et al. 1985; London, Waller,
Wamsley 1985); has euphoric effects and perhaps sedative or other anxiolytic effects
mediated by neurohormonal processes (e.g., Henningfield, Miyasato, Jasinski 1985);
and reinforces behavior, even among animals or human subjects blind to whether they
received saline placeboor nicotine (Henningfield, Chait, Griffiths 1983,1984). As with
other addictive drugs, prolonged ingestion of nicotine leads to tolerance, a tendency to
consume increasing amounts of a drug, presumably to achieve a desired euphoric or

340


TABLE 36.-Comparison of characteristics of addiction, habituation, and dependence in 1964 and 1988 Surgeon General's Reports

Characteristics of drug addiction and habituation in 1964 Surgeon General's Repot?

Characteristics of drug addiction in

1988 Surgeon General's Reportb

Dmg addiction                    Drug habituation

A state of periodic or chronic intoxication produced
by the repeated consumption of a drug (natural or
synthetic).

A condition resulting from the repeated consumption of
a drug.

Its characteristics include:

Its characteristics include:

(1) an overpowering desire or need (compulsion) to
continue taking the drug and to obtain it by any means;

(I) a desire (but not a compulsion) to continue taking the
drug for the sense of improved well-being it engenders;

(2) little or no tendency to increase the dose;

(2) a tendency to increase the dose;

(3) a psychic (psychological) and generally a physical
dependence on the effects of the drug; and

(3) some degree of psychic dependence on the effect of
the drug, but absence of physical dependence and hence
of an abstinence syndrome; and

(4) detrimental effects on the individual and on
society.

(4) detrimental effects, if any, primarily on the
individual.

Primary Criteria

Highly controlled or compulsive pattern of drug use.
Psychoactive or mood-altering effects involved in
pattern of drug taking.

Drug functioning as reinforcer to strengthen behavior
and lead to further drug ingestion.

Additional Criteria

Tolerance (increased doses either tolerated without
discomfort or needed to achieve desired effects).

Physical dependence (withdrawal syndrome upon
termination of drug taking).

Use despite harmful effects.
Pleasant (euphoric) effects.

Stereotypic patterns of drug use.

Relapse following drug abstinence.

Recurrent drug cravings.

    `SOURCE: US PHS ( 1964. p. 35 I ).
&    %OURCE: US DHHS (1988. pp. I94,25C-253).


other effect. Prolonged use also leads to physical dependence, as indexed by various
psychological and physical withdrawal symptoms following cessation of smoking. The
inclusion of tobacco dependence as a disorder in the Diagnostic and Statistical Manual
of Mental Disorders III, the official diagnostic reference for the American Psychiatric
Association (1980), was another major marker in the shift of scientific opinion about
the addictive nature of cigarette smoking.

Central to the 1964 view was the distinction between compulsive use (addiction) and
the less compulsive "desire" (habituation). The difference was noted to rest primarily
on the source of the desire or compulsion. The 1964 Report emphasized "serious per-
sonality defects from underlying psychologic or psychiatric disorders" (US PHS 1964,
p. 351) as a defining factor in compulsive use and therefore in addiction. Evidence
gathered since the early 1960s contradicts the assumptions that underlying pathology
drives the compulsive use seen in addiction. Drugs commonly viewed as addictive,
e.g., heroin, may be abandoned with little apparent effort as with many Vietnam
veterans addicted to heroin who gave it up after their return to the United States (Robins,
Helzer, Davis 1975; US DHHS 1988). On the other hand, the extent to which smok-
ing can be highly compulsive is suggested by its continuance in the face of substantial
awareness of its harm, as by cardiac patients (Baile et al. 1982; Burling et al. 1984;
Ockene et al. 1985; US DHHS 1984). The generality of nicotine's effects argues against
its compulsive use resting on individual psychopathology; the basis for nicotine addic-
tion rests on the interaction of conditioning processes and nicotine action in the brain.

Mechanisms of Nicotine Action

Much research in the 1970s on the behavioral effects of nicotine has been guided by
the nicotine regulation (or titration) model put forth over the years by Jarvik (1977),
Jarvik, Click, and Nakamura (1970), Russell (1976), and Schachter, Silverstein and col-
leagues (1977). According to this model, smokers regulate their smoking to maintain
a certain level of blood nicotine within a range of upper and lower limits (Herman and
Kozlowski 1979; Kozlowski and Herman 1984). This includes the avoidance of
withdrawal symptoms or anticipated withdrawal by maintaining a nicotine level above
a lower limit and avoidance of toxicity by maintaining it below an upper limit.

This formulation has been criticized as failing to explain the self-perceived positive
effects or benefits of smoking that may promote use (Pomerleau and Pomerleau 19841
Leventhal and Cleary 1980). Interestingly, the 1964 Surgeon General's Report devoted
only 1 l/2 pages to such effects. In the last few years, several investigators (e.g., Ock-
ene et al. 1988; Pomerleau and Pomerleau 1984) have proposed that smoking, by vir-
tue of the varied actions of nicotine, provides several positively perceived effects and
is employed by many smokers as a responsive and effective coping strategy. This im-
plies that smokers can be reinforced for continued smoking without maintaining a min-
imum blood nicotine level. The 1988 Report devoted an entire chapter to this topic.

An influential and historically important model of perceived positive effects of smok-
ing stressed the psychological effects of nicotine and other pharmacologic aspects of
smoking (Pomerleau and Pomerleau 1984). This model holds that nicotine increases
the release of a number of neuroregulatory hormones, conferring on smoking the ability

342


to act as stimulant or sedative depending on level of ingestion, background hormone
levels, and the like. Nicotine thus can serve to reduce anxiety or produce euphoria (US
DHHS 1988) and enhance vigilance for certain cognitive tasks (e.g., Warburton et al.
1986). The work of Grunberg (1986; US DHHS 1988) also suggests that nicotine may
aid smokers in maintaining lower body weight. Although objective judgment indicates
that the health effects of smoking are more important than the weight maintenance ef-
fects (Abrams et al. 1987), the latter seem to be of particular importance to some women
(Klesges and Klesges, in press; US DHHS 1988). This growing recognition that
smokers may value several effects of cigarettes can be used not so much to justify the
behavior but rather to direct intervention strategies (e.g., physical activity) that might
help people meet needs previously served by cigarettes. Interventions also are likely
to be seen as more credible to smokers if the coping value of cigarettes is recognized
(Ockene et al. 1988).

Conditioning and Smoking

What most distinguishes recent analyses of the conditioning of smoking from earlier
views (e.g., Hunt 1970) is their emphasis on the conditioning of the biological effects
of nicotine. The occurrence of stimuli previously associated with the effects of nicotine
will tend to evoke responses related to those effects or cues for further consumption
(e.g., Abrams et al., in press; Herman 1974; Niaura et al. 1988; Rickard-Figueroa and
Zeichner 1985). Such conditioned effects may link smoking to aversive states al-
leviated by nicotine. For example, investigations described earlier (e.g., Schachter, Sil-
verstein et al. 1977) suggested that smoking covaries with stress, which is hypothesized
to deplete nicotine. Leventhal and Cleary (1980) suggested that stress as well as other
emotions may be alleviated by nicotine and would then come to serve as cues for smok-
ing. Pomerleau and Pomerleau (1984, 1987) identified neurohumoral effects of
nicotine as the paths of its impact and elaborated on the ways such effects might be con-
ditioned to circumstances surrounding smoking so as to regulate it in the future.

Two influential theories of addiction emphasize the role of relief of withdrawal or
anticipated withdrawal in smoking. As suggested by Wikler's classic work with opioids
(Wikler 1973; Wikler and Pescor 1967), withdrawal symptoms may be conditioned to
the circumstances in which they occur. This would set the stage for stimuli associated
with prior drug taking to elicit withdrawal symptoms and urges. With smoking, greater
withdrawal symptoms have been noted when cessation occurs in natural rather than ar-
tificial environments, presumably because those natural environments contain
numerous cues associated with prior smoking (Hatsukami, Hughes, Pickens 1985).
Within this model, return to smoking after brief or extended abstinence is reinforced
by the reduction in such conditioned withdrawal symptoms.

Opponent-process theory (Solomon and Corbit 1973) suggests that the reduction of
aversive withdrawal symptoms may be the result of the interaction of the immediate
response to a drug, called the "A" state, and the delayed response, the "B" state. The
B state is "opposed" to or opposite the A-hence "opponent process"; if the A is
pleasurable, the B will be aversive. Initially, the A state is stronger. While initial,
pleasurable responses to nicotine may encourage increased smoking, regular smoking

343


leads the aversive B state to become stronger, which in turn may be reduced or avoided
by the A-state consequences of further smoking. After regular smoking has been es-
tablished, the A state serves only to avoid or reduce the aversive B state. That is, regular
smoking is pursued to reduce displeasure rather than to bring about the pleasure that
may have been its initial appeal. It is important to note that there is little evidence on
the validity of the Wikler theory or opponent-process theory as applied to smoking.

In contrast to models emphasizing relief of withdrawal, a recent review (Niaura et al.
1988) proposes an "appetitive" model of responses to cues associated with smoking.
Evidence indicates that cues surrounding smoking are more strongly conditioned to its
positively perceivedeffects than to withdrawal symptoms. That is, cues associated with
intake of nicotine (e.g., holding a cigarette or inhaling) come to elicit conditioned
responses similar to the effects of nicotine (e.g., relaxation, heightened arousal). These
effects are strong reinforcers and encourage continued efforts to obtain or ingest the
drug. These reinforcing effects may be more critical than the reduction of withdrawal
symptoms after periods of abstinence.

Critical to understanding the appetitive model is the idea that negative emotions are
not necessarily withdrawal symptoms. However, negative emotions previously al-
leviated by nicotine may serve as cues for seeking repetition of smoking's reinforcing
effects (Stewart, DeWit, Eikelboom 1984). For example, social anxiety may be the oc-
casion for smoking, which is then reinforced by nicotine's ability to reduce anxiety.
The anxiety, however, is a response to a stressful situation, not a symptom of withdrawal
from cigarettes. Smoking is reinforced by the anxiety reduction, not by reduction of
withdrawal symptoms.

The many ways smoking is conditioned to circumstances around it may explain "the
thorough interweaving of the smoking habit in the fabric of daily life" (Pomerleau and
Pomerleau 1987, p, 119). The sheer repetition of smoking also strengthens such inter-
weaving. It is estimated that the average pack-a-day smoker of 20 years' duration has
inhaled cigarette smoke over 1 million times (Fisher and Rost 1986; Pomerleau and
Pomerleau 1984), each inhalation providing an opportunity for conditioning smoking
to numerous circumstances of daily life. Moreover, with years of smoking, the emo-
tional states and daily circumstances conditioned to it may continue to increase, result-
ing in urges to smoke being conditioned to almost every circumstance encountered and
complicating the task of maintaining abstinence.

Cognition and Decisionmaking

Cognitive and decisionmaking processes play a lesser role in the maintenance of
regular smoking relative to the other factors discussed here. Smokers have long
believed that they derive positive effects from smoking. The "pros" of smoking have
been embodied in the instruments used in decisionmaking studies (Mausner and Platt
197 1; Velicer et al. 1985) and in the Horn and Waingrow (1966) Reasons-for-Smok-
ing Scale.

As documented in Chapter 4 of this Report, public knowledge of the health conse-
quences of smoking has increased steadily over the past 25 years. Eighty-seven per-
cent of current smokers now report that they understand that smoking is harmful to their

344


health (ALA 1985) and two-thirds of high school seniors report "great risk" being as-
sociated with pack-a-day smoking (Johnston, O'Malley, Bachman 1987). Why, then,
do so many persist in regular smoking? One reason may be that they do not appreciate
just how dangerous smoking is. For example, 75 percent of current smokers agreed
that smoking is a cause of lung cancer (ALA 1985), while 94 percent of nonsmokers
and 90 percent of former smokers agreed to this. For emphysema, the parallel figures
were 75 percent of current smokers compared with 91 percent and 90 percent of former
smokers and nonsmokers, respectively (ALA 1985). Surveys indicate a general insen-
sitivity to the relative level of risk associated with smoking. Health professionals rated
nonsmoking as the first priority among things Americans can do to protect their health.
The public rated nonsmoking as 10th. behind such worthy but, for most Americans,
less critical behaviors as consuming adequate vitamins and minerals and drinking water
of acceptable quality (Fisher and Rost 1986). As discussed below, the health belief
model (Rosenstock 1974) requires that smokers believe they are personally vulnerable
to a threat before they will be motivated to attempt change. It has been suggested that
personalized acceptance ("Cigarette smoking is dangerous to my health") always lags
behind general acceptance ("Cigarette smoking is dangerous to health") (Fishbein
1977; Lichtenstein and Bernstein 1980; Shiffman 1987) (See Chapter 4). These con-
siderations suggest that many smokers still find it possible to discount the riskiness of
their behavior.

Another possible reason for some smokers' insensitivity to smoking risks is that they
have not always been given the full message, or they have been given mixed messages,
including prosmoking messages (advertising) from the cigarette industry. Factors that
impede public awareness and acceptance of the health hazards of smoking include
cigarette advertising and promotion and cigarette companies' public relations and lob-
bying activities, which are also reviewed in Chapters 6 and 7.
Other issues related to persistence of smoking will be covered in the Section on Quit-
ting and Relapse.

Personal Characteristics and Social Context

Personal Characteristics

The 1964 Surgeon General's Report linked smoking in adulthood and adolescence
to extraversion, or as it defined it, a tendency "to live faster and more intensely" (US
PHS 1964, p. 366), and this relationship has been confirmed in later studies (e.g., Ash-
ton and Stepney 1982). However, reviews indicate that there is no consistent evidence
relating smoking to neuroticism or emotional instability (Smith 1970; US DHEW
1979a). More recent studies have continued to find relationships with smoking and be-
haviors linked to extraversion: coffee and alcohol consumption (Istvan and Mataraz-
zo 1984); circadian phase differences, being an "evening type" as opposed to a "mom-
ing type" (Ishihara et al. 1985); alcohol consumption, driving accidents, divorce,
frequent job changes, low levels of vocational success, and impulsivity (Eysenck 1980).

345


Another personality construct that received a great deal of attention earlier in the
smoking literature was Rotter's (1966) internal versus external locus-of-control dimen-
sion (e.g., Foss 1973; Best and Steffy 1975; Best 1975; Straits and Sechrest 1963). Two
general hypotheses characterized work in this area. The first noted that smokers tended
to have a more external locus of control, that is, perceive that things occur because of
fate, not because of one's own actions, compared with nonsmokers. The second held
that smokers with a greater internal locus of control, that is, a perception that things
happen because of one's own actions, would be more successful in quitting. A review
of this literature revealed inconsistent support for both hypotheses (Baer and Lich-
tenstein 1988b).

The multidimensional health locus of control scale (Wallston, Wallston, DeVellis
1978) was an attempt to anchor the locus of control construct specifically to health be-
havior consistent with the trend away from broad, dispositional traits (Mischel 1973).
Most studies using this scale examined the effect of health locus of control on cessa-
tion attempts. Three investigations reported small but significant prospective relation-
ships between subscales of the Health Locus of Control Scale and maintenance of
abstinence (Kaplan and Cowles 1978; Rosen and Shipley 1983; Shipley 1981).

A popular approach to understanding social or psychological problems has been
through typologies. Tomkin's typology of smoking and affect regulation was very in-
fluential in the 1960s and early 1970s (Ikard and Tomkins 1973; Tomkins 1966.1968).
Tomkins originally proposed a fourfold typology including positive affect, negative af-
fect, habitual, and addictive smoking. This model gave rise to the Reasons-for-Smok-
ing Scale (Horn and Waingrow 1966), which continues to be used widely in public
education and cessation programs despite receiving little empirical support (Shiffman
1988). Validity studies have yielded the most consistent support for the negative af-
fect smoking construct (Ikard and Tomkins 1973; Pomerleau, Adkins, Pertschuk 1978;
Joffe, Lowe, Fisher 1981).

The support demonstrated for negative affect smoking is also consistent with recent
reviews' emphasis on stress reduction as being among those biological effects of
nicotine that maintain regular smoking (e.g., Leventhal and Cleary 1980, Pomerleau
and Pomerleau 1987). Much evidence for such effects comes from the retrospective
reports of relapsers and smokers attempting to stop, which are reviewed later in this
Chapter. However, relatively few data demonstrate that heightened stress leads to
greater smoking. Among them are Ikard and Tomkin's observations (1973) of greater
incidence among race track spectators during horse races-presumed to be times of
stress-than in the periods before and after races, and Silverman's observations of
nicotine-induced reductions in aggression among rats (197 1). A number of other
studies reviewed in the 1988 Surgeon General's Report link smoking and negative af-
fect but. as noted in that review, are not conclusive as to whether reduction of negative
affect makes a substantial contribution to regular smoking. Design problems include
comparisons of smokers smoking with smokers who are deprived, leaving unclear, for
instance, whether smoking reduces negative affect or whether, for regular smokers, not
smoking merely causes an aversive, deprivation state. As concluded in the 1988 Report,
" . . . caution must be exercised in generalizing about smoking and nicotine's effects on
stress and mood . . ." (US DHHS 1988, p. 405).

346


Less direct support for effects of stress on smoking lies in studies of smoking
prevalence among groups who are disadvantaged in our society, including psychiatric
outpatients (Hughes et al. 1986) and male users of soup kitchens (McDade and Keil
1988). Of the 38 subgroups defined by gender and economic, educational, vocational,
or marital status listed in the 1988 Report, divorced or separated men had the highest
prevalence of smoking, 48.2 percent (US DHHS 1988). Other social problems such as
alcoholism and suicide are also more prevalent in this group (Kaplan and Sadock 1985).

Beyond those groups with significant disadvantages such as psychopathology and
very low income, the more general effects of income and education are quite substan-
tial. For instance, preliminary data from the 1987 NHIS indicate a 35percent smok-
ing prevalence among adults with less than a high school education, more than twice
the 16.3 percent prevalence among those with postgraduate college training (see Part
I). Prevalence among both women and men declines with increases in income range.
Among unemployed men, the prevalence is 44.3 percent (US DHHS 1988). Such trends
indicate that the social and economic context affects the relationship of personal charac-
teristics with smoking. Consistent with this, trends presented in Part I of this Chapter
indicate that observed differences of race and sex are attributable to effects of income
and education (see also Novotny, Warner et al. 1988).

Social Context Influences

The arrival at regular use roughly corresponds to the period of transition from adoles-
cence to adulthood. At least until very recently, the social changes that accompany this
passage-entering a university, the military, or the workforce-have been associated
with a marked change in the acceptability of smoking. For high school students, smok-
ing is often prohibited on school property, even if the prohibition is poorly enforced.
In the workforce, community college, and university setting, smoking has been wide-
ly accepted. The military until recently had supported smoking among its men and
women, as reflected in low prices for cigarettes at military exchanges and commissaries
and by the announcement of breaks with "The smoking lamp is lit." The extent to which
smoking is a part of the role of the serviceman was shown in a survey of Navy enlisted
men with a mean age of 22.6 years and a mean of 3.9 years' service. Seventy-two per-
cent were self-reported smokers (Burr 1984). That the military has an effect on creat-
ing rather than attracting smokers is suggested by a comparison of prevalence among
naval recruits, 27.6 percent, and shipboard men, 49.8 percent (Cronan and Conway
1988). The military has recently recognized the enormous costs attendant to the high
prevalence of smokers within its ranks and has begun efforts directed at reducing the
percentage of smokers among its personnel (See Chapters 6 and 7).

Cigarette marketing, discussed above and in Chapter 7, continues to be an important
influence encouraging adult smoking, with several possible direct and indirect influen-
ces on smoking patterns (Warner 1985).

347


Summary

The past 25 years have seen a deepening appreciation of the importance of nicotine
in maintaining regular smoking. In contrast to the 1964 Surgeon General's Report,
cigarette smoking is now defined as an addiction (US DHHS 1988). Earlier emphasis
on the maintenance of blood nicotine levels as a means to avoid withdrawal has been
balanced by the awareness that nicotine's varied effects make smoking an efficient
coping strategy for affect regulation and perhaps weight regulation. Conditioning
models of smoking have become more sophisticated and firmly integrated with the
pharmacologic actions of nicotine to explain addiction. While the public is now better
informed about the health consequences of smoking, many smokers still minimize their
perception of their vulnerability amid extensive marketing of tobacco products. Broad,
dispositional traits or motives are now seen to be of limited value in understanding
smoking. The role of social settings and social influence in encouraging regular smok-
ing is also better understood.

Cessation and Relapse

A large body of literature on determinants of cessation has evolved, driven by the
need to provide empirical and theoretical guidelines for intervention programs. All
three sets of determinants-pharmacologic processes and conditioning, cognition and
decisionmaking, and personality and social context-play an important role in the ces-
sation stage (Table 39). It is with respect to cessation, especially, that the concept of
stages-treating stopping as a process over time-has evolved (Figure 8) and now
guides research and interventions (e.g., Marlatt 1985). The influential and well-articu-
lated cessation stage model of Prochaska and DiClemente (1983) defines four stages
of cessation. Precontemplation is the stage in which the smoker is neither considering
stopping nor actively processing smoking-and-health information. During the con-
templation stage, smokers are thinking about stopping and are processing information
about the effects of smoking and ways to stop. in the action or cessation stage, the
smoker is no longer smoking and has been without cigarettes for less than 6 months.
The maintenance phase involves establishment of long-term abstinence, while relapse
is the resumption of smoking. When relapse occurs, the smoker recycles to any one of
the three previous stages.

Specific cognitive and behavioral processes are employed during the different stages
of cessation (Prochaska and DiClemente 1983). Determinants of each stage are also
different. Thus, factors that affect an initial decision to stop smoking may not predict
success in stopping or sustained maintenance after stopping. Working from a related
but different stage model-initial decision, initial control, maintenance-Rosen and
Shipley (1983) used health locus of control, desire to stop, and self-esteem to predict
self-initiated smoking reduction. Using regression analysis, a different set of predic-
tors was demonstrated at each stage, suggesting the possible need for different inter-
vention techniques at each stage of the smoking reduction process.

An important implication of a stage model is that interventions may need to address
cessation's several stages. The precontemplator's tendency to ignore quitting strategies

348


may need to be met with continued personalized information on smoking and health;
the contemplator may need social support to attempt cessation; and the abstainer may
need help that emphasizes the development of relapse prevention skills. There are as
yet no data available to demonstrate the effect of interventions tailored to specific stages
of cessation. Thus, a model like the Prochaska and DiClemente stage model is best
viewed as a tentative conceptualization, useful for guiding research and interventions.
The next section considers changes in our understanding of the determinants of cessa-
tion in relation to the stages in the cessation process.

Pharmacologic Processes and Conditioning

Pharmacologic processes and conditioning exert a strong influence on the process of
quitting. One indicator of the role of addiction is that heavier, more dependent smokers
in intervention programs are less likely to quit than are lighter, less dependent smokers
(e.g., Hall et al. 1984;Ockene et al, 1982b), especially when smokers with much
variability in baseline smoking are studied, as in the Multiple Risk Factor Intervention
Trial (MRFIT) (Hughes et al. 1981). As is noted in the 1988 Surgeon General's Report,
"Withdrawal symptoms, whether elicited by acute deprivation or by conditioned
stimuli, are hypothesized to be the link between dependence and relapse" (p. 523), al-
though some analyses (e.g., Niaura et al. 1988) place greater emphasis on positive ef-
fects of smoking in motivating relapse. Further evidence of the influence of addiction
comes from intervention studies evaluating nicotine-containing gum. Several studies
have found that nicotine polacrilex gum is more effective when used with nicotine-de-
pendent smokers (as measured by the Fagerstrom (1978) addiction questionnaire) than
with less dependent smokers (Hall et al. 1985; Killen et al. 1984; Schneider et al. 1983).
Nicotine polacrilex gum most likely is effective because it reduces withdrawal
symptoms frequently noticed in the first days and weeks of abstinence (Hughes et al.
1984; West et al. 1984). Recently, more work has focused on nicotine replacement
strategies or other pharmacologic treatment adjuncts reflecting the importance of
biological factors in smoking and cessation (Grabowski and Hall 1985; US DHHS
1986b; US DHHS 1988).

Conditioning mediates the role of the pharmacologic effects of nicotine in cessation.
As noted in the discussion of regular smoking, numerous conditioned environmental
stimuli are likely to evoke urges or cues to smoke. Recent work by Abrams and col-
leagues demonstrates that former smokers manifest psychophysiological reactivity to
smoking cues long after they have quit (Abrams et al., in press; Abrams 1986). Con-
ditioned reactivity to environmental cues, then, may be more decisive in the later stage
of maintenance after withdrawal symptoms have subsided.

Research on relapse triggers reflects current interest in specific, situational vari-
ables. Primary triggers include stress, interpersonal conflict, dysphoria, presence of
other smokers, and alcohol consumption (Marlatt and Gordon 1980; Shiffman 1982).
Although the data are primarily retrospective reports from relapsed or tempted subjects,
there is convincing consistency on the importance of stress and negative affect in deter-
mining maintenance or relapse (Baer and Lichtenstein 1988a; Marlatt and Gordon
1980; Ockene et al. 1982a; Shiffman 1982; US DHHS 1988). The mechanism whereby

349


a lapse becomes a full return to smoking has also recently been analyzed as a series of
stages (Marlatt 1985). These include a high-risk occasion that triggers a smoking lapse
(that is, a brief return to smoking) and a subsequent interpretation of the lapse that may
lead to abandoning the cessation effort and a return to regular smoking. Much recent
attention has been paid to the importance of coping responses in dealing with both
high-risk situations and lapses (e.g., Shiffman 1984; Shiffman and Wills 1985). The
available data suggest that the absence of any coping response is predictive of relapse
but there are few differences that relate to the use of specific coping strategies used
(Shiffman 1984).

Cognition and Decisionmaking

The role of cognitions in smoking cessation is evident in the relapse model noted
above (Marlatt 1985). In this model, a lapse diminishes self-efficacy or self-confidence
and expectations for long-term success. These diminished efficacy expectations then
become the basis for an.individual to abandon the effort and return to regular smoking
(Marlatt 1985). In fact, lapses are highly predictive of subsequent relapse (Brandon,
Tiffany, Baker 1986; Baer et al. 1988).

Researchers have long noted the relationship of knowledge about the health conse-
quences of smoking, beliefs about personal susceptibility, attitudes toward smoking,
and expectations about the benefits of quitting to cessation efforts and their long-term
success or failure. Cognitive-behavioral models of smoking cessation emphasize the
importance of an individual's interpretation of health risks and perceived self-efficacy
for refraining from smoking (Pechacek and Danaher 1979) as well as attributions about
addiction and lapses during the maintenance stage (Marlatt 1985).

Expectancy-Value Models

Expectancy-value models have guided approaches to smoking cessation for many
years (e.g., Kirscht 1983; Mausner and Platt 1971; Sutton 1987). Outcome expecta-
tions refer to expected consequences that would occur if one continued smoking or quit
smoking (Bandura 1977). Their value refers to the personal importance or weight given
to the various possible outcomes and can be extended to perceptions about what sig-
nificant others wish one to do (Fishbein 1982). Expectations include the positive (e.g.,
enjoyment) and negative (e.g., disease) consequences of smoking and the positive (e.g.,
enhanced lung capacity) and negative consequences (e.g., loss of enjoyment,
withdrawal symptoms) of quitting. Expectancy-value models tend to assume that
human behavior is rationally guided by logical or at least internally consistent thought
processes (Henderson, Hall, Linton 1979).

Decisionmaking models represent one variant of the expectancy-value approach and
have been (e.g., Mausner and Platt 197 1) and continue to be (Velicer et al. 1985) ap-
plied to smoking cessation. The more recent applications (Velicer et al. 1985) may
prove more useful because they take into account stage of change (Prochaska and Di-
Clemente 1983). Changes in the relative level of pro and con views of smoking, for
example, appear related to stages of quitting. Smokers not contemplating quitting
report substantially higher levels of pro than con views, while those contemplating quit-

350