Introduction

Cigarette smoking by employees results in increased expenses for
employers. Smokers use the health care system up to 50 percent
more than nonsmokers (Fielding 1984); this means higher health
insurance costs for companies. Studies have reported higher rates of
work-related accidents, disability reimbursement payments, and
absenteeism among employees who smoke than among those who do
not (Terry 1971). Although it is difficult to assess exact dollar
amounts because of the variety of circumstances and assumptions
involved (Warner 1983), estimates of excess annual costs to employ-
ers per smoking employee generally run from $200 to $500 (Lute and
Schweitzer 1978; Kristein 1982). Costs attributable to smoking
among employees in the high risk occupations discussed in this
Report are likely to be considerably higher than these overall
estimates.

These data, as well as consideration for the welfare of their
employees, have led a number of businesses to establish workplace
antismoking programs. Because of the magnitude of the health
effects of smoking and the benefits of cessation, smoking cessation
programs are likely to yield a higher return on investment than
worksite health promotion programs targeting other risk factors
such as obesity and lack of exercise (Fielding 1984). Surveys reveal
that 11 to 15 percent of American businesses provide smoking
reduction programs and many more are considering such programs
(Dartnell Inst. 1977; NICSH 1980). In response to the recommenda-
tions of a panel of experts concerning priorities for health promotion
activities, the Health Insurance Association of America has estab-
lished a smoking reduction program that is available to its members
(Fielding 1984). From one-third to one-half of the large organizations
have designated no-smoking areas (Dartnell Inst. 1977; NICSH 1980).

A great variety of worksite smoking-modification approaches have
been devised, including monetary incentives and contests for not
smoking, distribution of self-help materials, physician messages and
health education lectures on the adverse effects of smoking, and
stop-smoking clinics (Chesney and Feuerstein 1979; Danaher 1980;
Klesges and Glasgow 1985; Orleans and Shipley 1982). Stop-smoking
sessions have been led by coworkers, volunteers from health
organizations, commercial cessation consultants, and health profes-
sionals. Ongoing multiple risk factor intervention programs, either
for the entire workforce or for individuals at especially high risk of
developing cardiovascular disease, have been offered. The purpose of
this chapter is to critically review the literature on such programs.
First, however, it is helpful to consider both the potential advantages
and the possible disadvantages of worksite smoking modification
programs versus the more traditional, clinic-based programs.

477


The potential advantages of worksite-based smoking modification
programs can be considered from the perspective of employees,
employers, and public health researchers. For employees, the
primary potential advantages appear to be increased convenience
(particularly if the program is held during work hours), reduced
expenditure if the company pays all or part of the program fee, and
the opportunity to participate with friends and coworkers rather
than a group of strangers. For the employer, potential benefits
include increased worker productivity, better employee morale, and
better employee and public relations from health promotion efforts.
The potential monetary savings from reduced absenteeism and
medical costs are also appealing.

For public health researchers, worksite programs offer the advan-
tages of a much larger number (and possibly different types) of
smokers involved in efforts to quit than would otherwise be the case,
greater ease in obtaining long-term followup data, and the opportu-
nity to provide sustained or ongoing programs rather than one-time
offerings. In worksite programs, treatment is conducted in the
environment in which participants spend a large portion of their
day, which should facilitate generalization of treatment effects and
potentially lead to the establishment of nonsmoking norms. Possibly
the greatest potential resource available in worksite programs from
all three perspectives is the additional incentive and motivational
components that can be brought to bear through both monetary and
social support manipulations.

It is important to realize, however, that these potential benefits do
not occur automatically (Klesges and Glasgow 1985), and that they
may be offset by possible disadvantages of worksite smoking
modification programs. From an employee perspective, participation
may interfere with work activities or be outwardly condoned, but not
supported, by a supervisor. Meetings may be held at inconvenient
times or in inconvenient locations. If promotional activities are not
handled appropriately, workers may feel coerced to participate.
From an employer's perspective, there are the direct costs of the
program, such as advertising, counselor time, and materials, as well
as indirect costs, such as time off work for employees to participate.
Sponsoring an antismoking program can also create employee
relations problems. Nonsmoking employees may resent the time off
work available to smokers and may demand that their own
participation in health promotion programs be subsidized. The
critical issue here may be company norms, whether time off is
consistent with previous company practice regarding other programs
for employee benefit. In organizations in which workers are exposed
to hazardous substances such as asbestos, unions may view smoking
cessation programs as attempts by management to absolve them-

478


selves of responsibility for occupationally related disabilities (Ellis
1980).
There are also problems from the perspective of public health
researchers in conducting programs in the workplace. Most of these
potential disadvantages result from a reduced degree of control over
variables that can influence outcome. For example, company pro-
gram planners (organizational steering committee) might decide to
conduct additional stop-smoking activities (e.g., changes in company
smoking policies, added incentives for not smoking, participation in
other health promotion activities, a contest with a rival business)
that are not part of the study design. Finally, some participants may
take part solely as a way of getting out of work rather than from a
desire to change their smoking behavior.

Criteria for Evaluating Worksite Programs

The criteria for evaluating program effects are considered under
three general headings: changes in participants' smoking behavior,
effects on smoking and health-related variables for all employees in
t.he organization, and "secondary" effects of a program on nonhealth
variables of concern to employers. Most reports on worksite-based
programs assess only one or two of these areas.

Changes in Participants' Smoking Behavior

The same considerations that apply to the measurement of adult
smoking behavior in clinic settings apply also to worksite smoking
modification programs. Specification of reported smoking data is
particularly important. Following a program, there is often a
bimodal distribution of smoking rate, with a number of individuals
successfully quitting and many nonquitters smoking at close to their
baseline rate. Presentation of reductions in the "average" number of
cigarettes smoked can therefore be misleading. It is important to
separate data about subjects who are abstinent from data about
those who are still smoking, albeit at a reduced rate, when reporting
either reductions in smoking behavior or biochemical indices of
smoking exposure.

It is critical, of course, to have information about the long-term (6
to 12 months minimum) effects of smoking modification programs
(Lichtenstein and Brown 1982; McFall 1978). Interest in research in
the "dynamics of cessation and relapse" is much more recent (US
DHHS 1983, p. 246; Ockene et al. 1982). It is helpful to know, for
example, whether a 30 percent long-term abstinence rate resulted
from the same 30 percent of participants remaining abstinent
throughout the followup period or from IO percent new quitters, IO
percent previous relapsers, and.10 percent who remained abstinent
throughout the assessment periods.

479


Objective verification of changes in smoking behavior has become
the standard for defining smoking behavior. Recent reviews have
been conducted of several biochemical measures of smoking status,
including carbon monoxide, saliva thiocyanate, and cotinine (Freder-
iksen and Martin 1979; Leupker et al. 1981; Benowitz 1983; Bliss and
O'Connell 1984). Simply having an informant, usually a spouse or
coworker, "confirm" a participant's smoking status may not be
sufficient corroboration. Such people are not in a position to
continuously observe a participant's smoking behavior throughout
the day and may be persuaded to falsify their report on the
participant's smoking behavior.

Worksitewide Program Effects

The impact of a worksite program may include effects on workers
other than those enrolled in the program and effects other than
smoking cessation. The localized nature of a worksite program and
the repetitive interactions of workers in the program with those who
did not participate may produce changes in the attitudes and
behaviors of the active workforce that promote smoking cessation
and improve employee morale and productivity. For these reasons,
one criterion for evaluating worksite programs should be the
fraction of the workforce whose smoking behavior is altered in
addition to the fraction of the participants who quit. All of these
effects are important in evaluating the reported success rate of a
program because a very high cessation rate for a program may have
little overall impact if only small numbers of employees are willing
to participate (Kanzler et al. 1976). Whenever possible, program
costs should be reported in addition to data on the effects on smoking
patterns of nonparticipating smokers. In the same vein, ongoing
worksite programs conducted over a number of years should attempt
to document the effects of a smoking modification program on
variables such as absenteeism, medical care expenses, and health
services utilization.

General Effects

Variables such as employee morale and productivity, commitment
to the organization, turnover, and employee-employer relations are
important potential secondary effects of a worksite program. Be-
cause these issues do not directly concern the topic of smoking and
health and have been infrequently assessed, they are not considered
in this review. It should be noted, however, that Brownell (1985)
makes a convincing case that if the field of worksite health
promotion is to prosper, concerted attention needs to be directed
toward demonstrating the effects of worksite programs on these
organization management issues. He argues that managers may be
more interested in such results than in changes in health status.

480


General Review of Worksite Programs

A large number of worksite smoking control programs have been
conducted. Unfortunately, only a small percentage of these programs
have been evaluated. The characteristics and results of experimental
investigations of occupational smoking control programs that have
presented more than anecdotal data are outlined in Tables 1 through
3. Many of these studies have consisted of pretest-posttest or post-
test-only evaluations without control conditions and have not
reported objective measures to validate self-reports of smoking
status. The sample size, type of worksite setting, and reported results
of such uncontrolled studies are listed in Table 1. Because of the
absence of comparison conditions, the lack of verification of smoking
status, and the general sparsity of information about program
procedures and treatment effectiveness in these reports, there are a
host of alternative explanations of their results. Therefore, they are
only briefly summarized.

Uncontrolled Studies

Although programs have been conducted in a variety of worksite
settings (Table l), the majority have been either conducted in
companies of small to moderate size with white-collar employees or
offered only to supervisory personnel. The number of participants is
generally small. Self-reported abstinence rates for these uncon-
trolled studies ranged from 25 to 90 percent (median, 60 percent) at
posttreatment and from 6.5 to 91 percent (median, 33 percent) at 6-
month or l-year followup. These figures, while encouraging, must be
interpreted with caution because it is often unclear whether the
reported rates have excluded subjects who dropped out of treatment
or followup, and because, in several studies, subjects received sizable
monetary rewards based upon reports of abstinence that were not
corroborated by objective measures of smoking.

Not known is the impact of the programs listed in Table 1 on
overall rates of smoking in the worksites in which they were
conducted (see Bishop and Fisher 1984). The majority of investiga-
tions do not report rates of participation in their programs, but the
studies that have reported (other than in very small companies as
noted below) have been discouraging. For example, Kanzler and
colleagues (1976) found that despite an intensive promotional
campaign, only 4 percent of smokers in their workplace began the
cessation program. Grove and colleagues (1979) found that of 409
smokers in their worksite, only 101 attended the first meeting, and
only 33 (8 percent of the smokers in the workforce) completed
treatment. Of these 33 subjects, only 9 were abstinent at 6-month
followup. Stachnik and Stoffelmayr (1981), noting these generally
low participation rates, stated: "The question of how one can


TABLE I.--Uncontrolled studies without objective measures
      of smoking status

Stud>

Number of WbJ`Xk
type of worksite

Cessatmn rate cpercent I

          Followup
Posttreatment    INO. months)

Andrew (19831        965 hospital employees

Bauer 119781

81 Bell Laboratorm
employees

Not reported


   90

26 (201

30 161

Bishop and Fisher
119841

lW6 employees in
each of six companies

254x

6X33
1121

Dawley et al 119841

15 VA hospztal
employees and 2 patlents

88

50 16,

Elhs (19WI

Asbestos company
employees

Not reported      30 (48)

Grove et al 119i91      33 Blue Cross employees          33         27 161

Heckler 119801

16 Thomas Llpton, Inc.
employees

Not reported

50 (1)

Kanzler et al (19761

9 psychiatric institute
employees and 21
commumty members

67

40 1121

33 engine manufacturing
company employees

Not repmtea

55 1121

Rosen and
Lichtenstein ,197:)

12 ambulance company
employees

58

33 112,
lat worki

Shepard 119801

26 electromcs mfg.
company employees

Not reported

35 148)
fat work1

Sorman i19791

Not reported

31 112,

Stachmk and
Stoffelmayr ,19831

Employees in three
compames bank,
manufacturer, and
health services

Nat reported

a&91
(61

increase participation in smoking cessation programs should receive
the same attention that the more standard question of which
cessation technique is most effective has received in the past" (p. 49).
The exceptions to these low participation rates are seen in studies in
the companies with fewer than 100 employees that have employed
incentive procedures (e.g., Rosen and Lichtenstein 1977; Sorman
1979; Shepard 1980; Stachnik and Stoffelmayr 1983).

482


Controlled Studies

Studies that have included control or comparison conditions are
presented in Tables 2 and 3'.  To emphasize the importance of
worksite and participant characteristics, these characteristics as
well as data on the public health issues of recruitment strategies
employed and on the participation and attrition rates experienced
are listed in Table 2. The type of intervention and experimental
design employed, short- and long-term cessation rates, and type of
biochemical validation of smoking status obtained, if any, are
described in Table 3. In this section, a general discussion of the
status of the worksite smoking modification literature with emphasis
on the characteristics of the most successful programs is followed by
a more detailed review and discussion of several important subtopics
within the occupational smoking modification field-the role of
social support, physician assistance, incentive approaches, employ-
ees at particularly high risk for the development of cardiovascular or
respiratory disease, multiple risk factor reduction programs, and
organizational characteristics that affect program success.

The varied programs conducted have ranged in intensity from a
brief physician message (e.g., Li et al. 1984) to ongoing programs
involving multiple components over a 4- to 5year period (e.g., Rose et
al. 1980). Recent programs have offered participants a variety of
behavior change options. In particular, 7 of the 14 studies outlined in
Tables 2 and 3 allowed subjects to select as goals either smoking
reduction or abstinence.

The most encouraging finding is that the long-term success rates of
the programs reviewed are relatively high. Although initial cessa-
tion rates do not appear to differ from those typically produced by
community-based smoking clinics, the longer term followup data are
more positive if viewed as a percentage of posttreatment cessation
outcome. Abstinence rates at 6 to 24 months after a program are
approximately 60 to 65 percent of those observed at posttest, in
contrast to the 20 to 30 percent figures classically cited for clinic
programs (Hunt and Bespalec 1974; McFall 1978). In fact, the lowest
maintenance rate in the studies summarized in Tables 1 and 3 was
26 percent of the posttest rate, and some studies report followup
results equal to or better than posttest (e.g., Malott et al. 1984; Meyer
and Henderson 1974; Schlegel et al. 1983). On the other hand, much
higher long-term abstinence rates, 50 percent or better of all
subjects, have recently been reported from a number of treatment
programs (US DHHS 19821, and results from the 22-center Multiple

483


lb
$  TABLE 2.-Worksite, subject, and procedural characteristics of controlled outcome studies


                              Participation rate        Characteristics of      Attrition rate
   Study         Size and type of worksite        (percent)           participants        @aTent)       Recruitment slrategies

Abrams et al.
(1985)

WO-employee medical
manufacturing company and
1.6Wemployee insurance
carrier

Gleegow et al.    6OGemployee telephone
we4         company

Glasgow et al. (in  VA hospital. health care
Pm)        services company, and
          savings and loan

Klesgee et al.
w35)

Four banks and one savings
and loan, 11%180 workers
each

Kornitzer.
oramaix et al.
U980)

30 Belgian factories

Li et al.
(1984)

Naval shipyard

Not reported
(estimated 6)

Not reported     25 female, 11 male
(estimated i8)     employees

Not reported

20 female, 9 male employees

88 with competition;
  53 without
  CP <ox4

84 agreed to screening

87

54 clerical end bluecollar
employees

82 female, 25 male
employeee

19,390 male employees, aged
do-59 years; high risk: upper
20 percent of risk
distribution

871 male shipyard workers

42

Not
repted


  7

9

Not
reported

17

Paycheck stuffers, p&ers. newsletter
articles

Employee organization sponsorship,
newsletter notice8, posters

Brochures, posters, newsletter notws
memoe

Brochures. announcements by bank
presidents, time off work for
participation; prize to bank with higheet
participation

Not reported

Participation asked at required screening


TABLE 2.-Continued

Study

Size and type of worksite

Participation rate
  (percent)

Characteristics of
participants

Attrition rate
(percent)       Recruitment strategies

Malott et al.
(198-u

Meyer and
Henderson
11974)

Neppa (1984)

Rand et al.
Il984I

Fhe et al.
11980)

Schlegel et al
1198.7)

Scott et *I.
(19831

Medical clinic and telephone      Not reported
company                  (estimated 7)

Varian Corporation; 240
employees, volunteers for
risk factor screening (13
percent of workforce)


Johnson & Johneon
Corporation

Large city hospital

Not reported

Not reported

Not applicable

24 large British industrial
fWW+

86 agreed to screening

28 Canadian military bases

Not reported

Large VA hospital

100

20 female, 4 male
employees, primarily clerical
and numee

36 employees identified at
screening aa high risk for
cardiovascular disease

36 white-collar employees: 20
women. 16 men

18 female employees

18,210 male employees. 40-
59 years old; high risk:
upper 12-15 percent of
distribution

243 armed forces perwnnel
65 percent male1

26 numm (22 women. 4
men)

0

0

67

Not
reported

612

Not
reported

0 of those
continued
at VA

Newsletter notices, brochures distributed
by supervisors, recruitment in
lunchrooms

Invitation to health screening

Posters, desk drops, company newsletter

Advertisements. word of mouth

Invitation to health screening exam

Posters. new8 releases

Individually approached


E  TABLE 3.-Design and outcome of controlled worksite smoking modification studies

Cessation rate (percent)

Study

Program intensity              Experimental
and componenta                 design

PC&
treatment

Followup
(No. months)

B&hem&
veriticatlon

Abram8 et al.
(1985)

Basic foursession nicotinefading
cewation program; four+eseion
mttintenance treatment

Basic program plus
health education (n= 181;
stress management (n = 18);
or social support (n = 18)

38'           331 (3)            co
33'            27'
6'             6*

Glasgow et al
(1984)

Seven weekly small group meetings
on brand changing and number
reduction; goal choices, abstinence or
controlled smoking

Gradual reduction (n = 12);
abrupt reduction (n= 13);
gradual plus feedback (n=ll)

Not
reported

33' (6)            a.3
0'
01

Glasgow et al.
(in pree.3)

(See Glasgow et al. 1984) Social
support with two meetings.
inatallmenta of manual, and phone
Cdls

Basic treatment progrsm (n=13)
vs. basic treatment plus
significant other social support
(n = 16)

54           25 (6)            co
36             23            SCN

Kleegea et al.
(1985)

(See Glasgow et. al. 1984)
Competition, with monetary prizes,
weekly feedback charta

Quasi-experimental;
basic treatment (n=16) vs.
basic treatment plus
competition (n=91)

31           14 (6)            co
22             18             SCN


TABLE 3.-Continued

Study

Program intensity               Experimental
and components                 design

   Ceesation rate (percent)

post-        Followup
treatment       (No. months)

Biochemical
verification

Kornitzer.
Dram& et al
( 1980)

Multiple risk factor program, written
advice and antismoking posters; high
risk subjects, semiannual physician
counseling and stop-smoking booklet

Treatment (n=7,398) vs.
screening only (n=8,824)

Not
reported

High risk
  19' (241
   12'
Random sample
   12.5
   12.6

No

Li et al.
.(1984)



Malott et al
(1984)

Oneeeeaion physician advice and
stopsmoking pamphlet



(See Glasgow et al. 1984) Coworker
support: partner support manual,
buddy system, individualized support
behaviors

5 to &minute behavioral
counseling (n =215) vs.
warning to quit (n=361)

Basic treatment program (n=12)
vs. basic treatment plus coworker
social support (n=12)

Not
reported



17
17

8.4' (3, 11)


  3.6 1


27 (6)
  17

Meyer and
Henderson
( 1974)

Multiple risk factor program, 9 to 12
meetings; 2- to 3.5hour behavior
modification group meetings with
*po=+=

Behavior modification (n=12) vs.
individual counseling (n-10) vs.
physician advice alone (n = 14)

40           20 (3)
25             25
0             33

co





co

No

Nepp (1984)

Nine written self-help modules;
minimal therapti contact

Quasiexperimental: minimal
contact (n-36) compared with
earlier group cessation program

22           14 (6)            co


TABLE 3.-Continued

Study


Rand et al.
(1984)

    Program intensity              Experimental
     and components                 design


Monetary incentives for low daily Co     Withinilubjecta design (n= 18):
levels, 1 week of reducing Co levels       baaeline-cutdown-abstinence
and 2 weeks of abstinence           l3ds

  Cessation rate (percent)

post-        Followup
treatment       (No. months)



  61         28 (3 wks)

Biochemical
verification



  co

Rose et al
(1980)

Schlegel et al.
(1983)

Multiple risk factor program: poeters
and stopemoking booklets; high risk
subjects, four company physician
consultations

Gmonth program; 16&page workbook;
abstinence or reduced smoking goal
choirs.?; base personnel were
therapists

Treatment (n=9,734) "8.
screening only (n=8,476)

Full treatment (17 sessions)
vs. minimal contact (4
-ions) ~8. self-help; crowed
with nicotine gum/no nicotine
gum

High risk
12' (5 yrs)
  0'
Others
  7'
  0'

45-68'        25-382 (12)
28-315          17-295
Cl4 3           7-10 8

No

No

Scott et al.
(1983)

Brief daily sessions with brand
fading, treatment manual, and Co
feedback, 3 months; abstinence or
reduced smoking goal choice

Treatment (n=16) vs.
no treatment (n=lO)

56'          25 (9)            co
0'             0

NOTE: The aample sizes reflect the number of subjects receiving each treatment condition. which in cane titancen differs fmm the total number of subjecta initiated into the study (see Table 2).
Except for the Kleegen and colleagues (1986) and the Li and c&zaguea Wf34) studies, in experiments using between-eubjecta designa there was random assignment to treatment conditions.
NOTE: CO=carhon monoxide; SCN=saliva tbiocyanate.
1.2 At each assessment point, conditiona that were aigniticantly different (p < 0.05) are identi&d hy different superscripts.
`Results of this factorial study are complex and diflicult to summarize with a notation system; see text for clarification (numbers for treatment conditions not reported).


Risk Factor Intervention Trial (MRFIT) study showed that long-term
abstinence rates actually rose from l-year to 6-year followup under
both intervention and control conditions (US DHHS 1983). These
results again indicate the need to separate point prevalence of
nonsmoking rates from continuous abstinence rates. Unfortunately,
data are seldom presented in terms of survival curves (e.g., Curry et
al., in press) or in a manner that permits assessment of the
consistency in smoking status over time.

These high followup rates suggest that the worksite may offer
more than a convenient location for cessation interventions, and
that interactions or changes in attitudes or behaviors in the worksite
may be important determinants of the successful maintenance of
abstinence. Although there are several potential explanations for
these relatively good maintenance data, the most obvious is the
ongoing contact that coworkers have with each other during the
followup period. Consistent with this hypothesis, the clinic-based
smoking cessation program in the MRFIT study, which produced one
of the most impressive maintenance rates of any study (Hughes et al.
1981; Ockene et al. 19821, involved ongoing contact over several
years. Even if coworkers are not highly supportive of each other,
they may come to provide no-smoking cues for other participants.

It is important to note that people involved in many of the
worksite studies described here are both self-selected and self-
motivated, as there are few usual care or no-treatment conditions.
Some of the high maintenance rates may be reflective of new
quitters who would stop in any program, or even without one.

This comparison of clinic-based programs and worksite programs
assumes that participants in each setting are similar. Although the
demographic and smoking history characteristics of subjects in these
reports do not appear to differ systematically from other studies in
the cessation literature, there may still be marked differences
between the groups. For example, since worksite programs generally
attract a higher percentage of smokers than do community-based
clinics, some of these subjects may be more recalcitrant smokers
(Bishop and Fisher 1984). The convenience of the worksite setting
may also attract more smokers. On the other hand, it may be that
only individuals who have repeatedly failed to quit on their own will
expend the time, effort, and money often involved in participating in
community clinics (Schachter 1982). To date, the primary determi-
nant of participant characteristics in the controlled outcome studies
(Table 2) seems to be the type of worksite in which a program is
conducted. Research is needed on the hypotheses concerning possible
differences between participants in worksite programs and smokers
attending community clinics.

Several differences between studies reporting high success rates
and those with less favorable results are clear. One of the more

157-964 C - 81 - 17                               489


striking differences to emerge is that the results of the better
controlled studies summarized in Table 3 (median posttest cessation
rate, 28 to 31 percent) are generally lower than those of the
uncontrolled studies outlined in Table 1 (median posttest cessation
rate, 60 percent). The most obvious explanation for this finding is
that most of the controlled studies included objective biochemical
indices of treatment outcome and subjects in these studies may have
more accurately reported their smoking status.

The intensity of smoking modification programming also appears
to be related to treatment outcome. Programs involving only a brief
session or two or relying primarily on self-help materials generally
produced the lowest cessation rates-from 4 to 14 percent long-term
abstinence (e.g., Li et al. 1984; Nepps 1984). In contrast, the best
cessation rate reported came from an intensive multicomponent
program involving 20 group meetings over a 7-month period
(Stachnik and Stoffelmayr 1983). The only worksite study to directly
compare different levels of program intensity found that a greater
number of sessions was associated with higher cessation rates
(Schlegel et al. 1983).

Another fairly consistent finding is that programs conducted in
larger worksites generally seem to produce poorer outcomes. Of the
abstinence-based programs, studies taking place in worksites with
100 or fewer employees (e.g., Miller 1981; Shepard 1980; Stachnik
and Stoffelmayr 1983) seem to attain the highest abstinence rates,
and the larger scale controlled trials (e.g., Kornitzer, Dramaix et al.
1980; Li et al. 1984; Rose et al. 1980) to result in much lower
abstinence rates. Bishop and Fisher (1984), who have conducted
programs in a variety of different sized worksites, concluded that
larger worksites also typically produce lower participation rates
than do smaller companies. However, a greater number of employees
may still be served.

Programs addressing multiple risk factors (e.g., Rose et al. 1980) in
worksite settings generally yielded poorer cessation rates (and
tended to be conducted in larger worksites) than did smoking-
modification-only programs. Multiple risk factor programs may
achieve greater overall reductions in morbidity and mortality
because of their effects on other risk factors, however. Finally,
programs providing incentives for smoking abstinence (e.g., Shepard
1980; Stachnik and Stoffelmayr 1983) were among those with the
more impressive outcomes, although not objectively verified. These
findings are discussed in more detail in the section on incentives.

Remaining Issues

Few worksite studies have investigated participant characteristics
associated with treatment outcome. The most consistent finding to
emerge is that smokers of lower numbers of cigarettes have greater

490


success at quitting than do heavier smokers (e.g., Kornitzer, Dramaix
et al. 1980; Li et al. 1984; Rand et al. 1984). This finding is consistent
with results of clinic-based programs (e.g., Ockene et al. 1982) and
suggests that special attention needs to be devoted to ways to
successfully treat heavy smokers.

Relatively little is known about the long-term effects of the non-
abstinence-based smoking reduction programs under review (Tables
2 and 3). Allowing subjects to choose the treatment goal of
abstinence or of reduced smoking may attract more participants,
provide initial success experiences that can be built upon in later
cessation efforts, and benefit those subjects not able to achieve or
maintain abstinence. On the other hand, such programs may
dissuade subjects from pursuing a goal of complete cessation and
allow participants to rationalize that smoking is not hurting them
because they have made changes in smoking rate, topography, or
cigarette brand. It is beyond the scope of this chapter to review the
complex literature on potential compensation effects resulting from
changes in smoking behavior (see McMorrow and Fox 1983; Moss
and Prue 1982). It should be noted, however, that subjects selecting
smoking reduction goals have generally shown reliable, although not
always large, reductions in carbon monoxide levels (e.g., Glasgow et
al. 1984).

There has been improvement in the research methodology em-
ployed in worksite smoking modification studies. The majority of
studies conducted in the recent past have included both comparison
conditions and biochemical measures of smoking exposure. There are
still, however, a number of methodological deficiencies in current
worksite studies. Comparison conditions do not usually include no-
intervention or usual care groups; differences among interventions
are most commonly studied, and randomization is not always used.
Little is known about participation rates (only one-third of the
studies listed in Table 2 provide information on the percentage of
smokers participating). Data are seldom presented on effects of a
program on all smokers in an organization, and no data have been
published on characteristics of employees who participate in work-
site programs versus those who do not. Finally, there are so few data
on the health benefits or cost effectiveness of specific programs that
this information was not included in Table 3. Only the large-scale
multiple risk factor reduction trials (Kornitzer, De Backer et al.
1980; Rose et al. 1980) have presented such data.

Special Issues Relevant to Worksite Programs
Social Support

One of the most frequently cited reasons for conducting smoking
modification programs in occupational settings is the potential for


invoking peer and environmental support for nonsmoking (Chesney
and Feuerstein 1979; Stachnik and Stoffelmayr 1981). It has been
argued that peer support has important, long-lasting effects on the
outcome of stop-smoking efforts (Janis 1983), and there have been
several calls for increased study of the role of social support in
smoking modification (Klesges and Glasgow 1985; Lichtenstein 1982;
Stachnik and Stoffelmayr 1981). There are also correlational find-
ings that suggest the importance of social support to successful
smoking cessation (e.g., Coppotelli and Orleans, in press; Mermel-
stein et al. 1983).

Given this background, it is surprising that a large-scale correla-
tional study of occupational settings by Caplan and colleagues (1975)
found that the degree of perceived support from coworkers was
inversely related to smoking status. Among workers with low levels
of job stress, ex-smokers reported lower levels of support than
current smokers. There was no relationship between social support
and smoking status for people with high levels of job stress.
However, Caplan and colleagues' measure of social support was not
specific to smoking cessation, and may have been more an index of
the employee's responsibility for supervising or otherwise interact-
ing with other worksite personnel. More recent studies by Malott
and colleagues (1984) and by Glasgow and colleagues (in press) have
found a complex relationship between social support and outcome of
worksite smoking modification programs. Using a measure that
produced a score for both supportive and nonsupportive (negative)
social interactions, these two studies found that the presence of
smoking-related negative social interactions was inversely related to
treatment success. The presence of positive social support, which is
more frequently the target of social support interventions, was not
related to outcome.

Social support procedures such as use of a buddy system and
inclusion of nonsmoking coworkers or family members in treatment
sessions have been part of a variety of worksite programs (e.g., Bauer
1978; Sorman 1979; Stachnik and Stoffelmayr 1983). Unfortunately,
it is impossible to evaluate the contribution of social support in these
studies because of the multitude of other intervention strategies also
employed. In a review of studies on the effects of worksite incentive
programs for smoking cessation, Shepard and Pearlman (in press)
concluded that incentive programs that included spouses produced
better outcomes than those that did not.

The few worksite smoking studies that have attempted to experi-
mentally manipulate the level of social support have produced
discouraging results. Abrams and colleagues (1985) compared a
social support/social skills training program including a buddy
system with health education and cognitive-behavioral stress man-
agement procedures as ways to improve the long-term effectiveness

492


of a nicotine-fading cessation program. By the end of the program,
subjects in the social support/skills condition had relapsed signifi-
cantly more than subjects in the other conditions, and these
differences persisted at followup. In addition, consumer satisfaction
ratings revealed that subjects liked the social support/skills program
less well than other options. Abrams and colleagues concluded on the
basis of these findings that factors such as social support, theoretical-
ly assumed to enhance treatment, may actually reduce the effective-
ness of a treatment program in some instances.

Malott and colleagues (1984) evaluated the effects of adding a
coworker support component to a multicomponent treatment pro-
gram offering subjects the options of abstinence or controlled
smoking. They found that the addition of coworker support did not
improve treatment outcome on any dependent variable and that
subjects found the condition with social support to be less credible
than the basic treatment program. A replication and extension of the
Malott group's (1984) study by Glasgow and colleagues (in press)
involved family or significant-other social support and included a
partner-support manual, two group meetings for supportive others,
individualization of support procedures, and semiweekly phone calls
to partners. The results of this study were consistent with those of
Malott and colleagues (1984): no incremental effects of the social
support program were obtained from any dependent variable.

Thus, in research conducted to date, the inclusion of existing social
support procedures has not been found to enhance outcome in
worksite smoking modification programs. This is not to say that
social support is not important to treatment success, but that the
issue is more complex than was initially believed. It may prove
difficult to alter existing levels of social support, and novel ways of
enhancing coworker and family support for smoking modification
need to be developed.

Physician Advice

Because as many as 70 percent of adults in our country visit their
physician at least once in a given year (US DHEW 1979), there has
been growing interest in finding ways in which physicians can
convince patients to give up smoking (Ewart et al. 1983; Russell et al.
1979).

Some of the best data come from recent European clinical trials
(Rose et al. 1980, 1982; Kornitzer, Dramaix et al. 1980). The Belgian
Heart Disease Prevention project (Kornitzer, Dramaix et al. 1980)
found that significantly more individuals at high risk for developing
heart disease stopped smoking in an intervention condition empha-
sizing semiannual physician messages than in a screening-only
control condition (see Table 3). When comparing a representative
sample of all intervention subjects (many of whom did not receive

493


the physician messages) with a similar sample of control subjects,
however, almost identical and fairly low cessation rates were
observed in both conditions.

The most intensive worksite physician-intervention program was
evaluated in a similar study conducted by Rose and colleagues (1980).
In this study, the 12 to 15 percent of subjects in the intervention
condition who were at greatest risk for cardiovascular disease were
provided four physician consultations, each approximately 15 min-
utes in length. Results were similar to the Belgian study in that a
greater percentage of the high risk subjects stopped smoking in the
intervention group than in the control condition (see Table 3). But
considering all participants (many of whom did not receive the
intensive physician messages), there were no significant differences
between conditions.

A series of four physician visits was also utilized in a carefully
controlled study by Rose and Hamilton (1978) of British civil
servants at high risk of cardiorespiratory disease. Although not
actually conducted in occupational settings (and therefore not
included in Tables 2 and 3), subjects for the study were recruited
through their worksite. This study produced the highest cessation
rates of any physician-advice study (self-reported abstinence rates of
51 and 36 percent at 1 and 3 years, respectively). Over a third of
these subjects were still smoking pipes and cigars, however, and a
relatively high percentage of "normal care" subjects also stopped
smoking (10 and 14 percent at 1 and 3 years, respectively). Although
there were differences in favor of the intervention condition in rate
of decline in airway obstruction and rates of phlegm production,
there were no differences between conditions in absenteeism over a
l-year period or in overall mortality over a 7- to lo-year period.
As these studies indicate, there are both advantages and disadvan-
tages in using physician stop-smoking messages in worksite settings.
One distinct advantage is that if stop-smoking advice is incorporated
into regularly scheduled physician visits, a relatively large number
of workers can be advised and treated quickly and cost effectively
(Lichtenstein and Danaher 1978). Another advantage of the physi-
cian model is that it is relatively unobtrusive in comparison with
management-sponsored programs (Danaher 1980). Physician advice
can also be used to augment other interventions rather than to
replace them, by assisting workers in deciding to seek help and by
promoting participation in intervention programs, therefore facili-
tating better use of these programs.

Although self-reported cessation rates resulting from physician
advice are low in an absolute sense, research is underway to attempt
to increase the impact of stop-smoking messages. Li and colleagues
(1984), in a study conducted in a navy shipyard clinic, found that
implementation of a 3- to Sminute session of behavioral counseling

494


by staff physicians significantly increased cessation rates over those
resulting from a simple warning to quit smoking (see Table 3). In
addition, the compliance of health care providers with treatment
protocols also affects outcome. For example, Li and colleagues (1984)
reported great difficulty in getting clinic physicians to consistently
deliver a brief 3- to 5minute message to patients, yet Ewart and
colleagues (1983) found that providing physicians with regular
performance feedback appears to improve the quality and quantity
of stop-smoking messages. Future research should identify proce-
dures to improve both the implementation and the out.come of
physician stop-smoking advice. Perhaps a stop-smoking message in
conjunction with other interventions may increase success rates.

Basic research on the effects of threatening communications such
as those describing health risks of smoking indicates that such
messages have their greatest impact if individuals know not only
what to do (e.g., stop smoking) but how to do it, and believe
themselves capable of acting (Leventhal 1970). The recent approval
of nicotine chewing gum by the U.S. Food and Drug Administration
and the availability of high quality self-help stop-smoking manuals
(e.g., Davis et al. 1984) now present an opportunity for physicians to
deliver a health warning accompanied by concrete recommendations
for what to do and how to do it. Recent data suggest that nicotine
chewing gum may assist heavier or more addicted smokers in
quitting (Fagerstrom 1978, 1984; Raw et al. 1980), and gum prescrip-
tions can be written at the same time that a stop-smoking message is
given. Only one study reviewed in this chapter has investigated the
use of nicotine gum (Schlegel et al. 1983). That study, which did not
involve physician advice, found that the gum enhanced treatment
outcome in self-help conditions, but in the context of an intensive 17-
session treatment program, subjects receiving the gum actually had
lower cessation rates than subjects not receiving gum. Particularly
in companies that employ their own medical staff or in which
employees are at risk because of occupational hazards, programs
combining physician stop-smoking advice with other intervention
options should be evaluated.

Incentives

Recently there has been increased interest in the motivational
factors associated with smoking behaviors (Shepard and Pearlman,
in press; Brownell 1985). This section focuses on two recent
approaches to increasing motivation: personal incentives and compe-
tition among participants.

Rosen and Lichtenstein 11977) published the first report on the
effects of an employee incentive program for stopping smoking. Of
the employees of a small ambulance company who smoked, 75
elected to participate in a program that involved a $5 per month

495


bonus for not smoking at work-for all employees regardless of their
initial smoking status. At the end of the year, the owner also
matched the total amount of bonuses received during the program.
No other intervention techniques were used, and no stop-smoking
meetings were held. This study underscores the potential power of
incentives to modify smoking behavior: at posttest, 58 percent of the
pretest smokers reported no longer smoking at work.

Sorman (1979) reported on a program that combined personal
incentives, a stop-smoking program, social support, low-calorie food
alternatives, and an exercise program. Of 202 employees, 55 enrolled
in the program in which each employee who quit smoking for 1 year
received a $200 reward. Thirty-one percent of the participants
reported having successfully stopped smoking for the entire year.
Shepard (1980) presented results from an ongoing incentive program
involving weekly $7 paycheck bonuses to employees not smoking at
work. After approximately 4 years, only 20 percent of the employees
reported smoking in the worksite compared with 67 percent at
pretest.

Stitzer and Bigelow (1982, 1983, in press) have conducted a variety
of studies that cogently demonstrate that contingent reinforcement
for reductions in carbon monoxide (CO) levels of expired breath
samples can produce reliable short-term reductions in CO levels in
hired cigarette smokers. A recent study (Rand et al. 1984) investi-
gated contingent reinforcement for smoking abstinence in 18 hospi-
tal employees. After 1 week of baseline smoking and a week-long
"cutdown test," subjects could earn $12 a day for 2 weeks if they
totally abstained from smoking and if their daily CO readings were
consistent with abstinence (< 11 ppm). Sixty-one percent of the
participants were abstinent throughout the 2-week contingency
period, and 28 percent remained abstinent throughout a 3-week
followup.

Finally, Stachnik and Stoffelmayr (1983) evaluated a comprehen-
sive 7-month-long worksite program involving sizable financial
incentives as well as health information, social support, and public
commitment to nonsmoking in the context of 20 gradually paced
group meetings. The program was conducted in three different
worksites, with from 47 to 70 percent of smokers enrolling in the
program and an astounding 80 to 91 percent of participants
reporting abstinence 6 months after participation in the program.
These results are obviously very impressive, but it is not possible to
evaluate the contribution of incentives versus the other procedures
employed.

Shepard and Pearlman (in press) recently reviewed 15 (mostly
unpublished) programs that used incentives to produce changes in
smoking behavior in the worksite. Some programs provide incentives
for not smoking at the worksite, and others have a goal of total

496


abstinence. Although incentive programs seem to be gaining in
popularity and self-reported cessation rates appear high, there are a
number of problems with evaluations of incentive-based smoking
programs. Most studies are uncontrolled, and with notable excep
tions (e.g., Rand et al. 19841, measures of smoking status during
nonwork hours and biochemical verification of smoking status are
lacking. Clearly, incentive programs deserve further investigation,
because they appear to be effective, relatively inexpensive, and easy
to implement in a variety of different settings.

Another approach to providing incentives for improvements in
health-related behaviors is to arrange competitions among different
worksites or teams within a given worksite. For example, Brownell
and colleagues (in press) report high participation rates, low attrition
rates, and impressive outcome data in a recent worksite obesity
competition. Given these promising results, Klesges and colleagues
(1985) conducted a worksite smoking competition among four banks.
Prizes to benefit both smokers and nonsmokers were awarded to the
bank with (1) the highest participation rate, (21 the largest reduc-
tions in carbon monoxide levels at posttest, and (3) the greatest
abstinence rate at the 6-month followup. All participants in this
study received a gradually paced smoking control program previous-
ly used in worksite settings (Glasgow et al. 1984). Finally, a
"Smoking Barometer" placed in the lobby or lounge of each worksite
provided employees with weekly feedback on how their bank was
doing compared with the other three.

Participation rates in the program were exceptionally high.
Overall, 88 percent of all bank employees who were smokers entered
the program, compared with a 53 percent participation rate at a
comparable savings and loan organization that received the identical
program without competition, a significant between-groups differ-
ence. There were, however, no differences between conditions in
cessation rates among participants in the program. This may have
been because subjects in the competition condition were more
nicotine dependent, as assessed by the Fagerstrom (19781 Tolerance
scale, than subjects in the noncompetition condition (p<O.O21.
Because of the higher participation rate, however, the competition
condition produced a higher long-term cessation rate (15 percent)
throughout the worksite than the comparison condition (7 percent).

One of the major advantages of incentive and competition pro-
grams is that they do not require large amounts of therapist or
participant time. If the success rates of the uncontrolled studies
(Table 1) can be replicated in controlled studies, incentive programs
may prove to be the most cost-effective approach to worksite
smoking modification. On the other hand, at least some people may
require additional guidance and support (Danaher 1980). One
convenient, low-cost method of providing skills training in the

497


context of incentive programs is through provision of self-help stop-
smoking materials (Glasgow and Rosen 1978; Windsor and Bartlett
1984). Although a number of worksite studies have employed written
self-help materials as part of multicomponent interventions (Li et al.
1984; Kornitzer, Dramaix et al. 1980; Nepps 1984; Rose et al. 1980),
self-help manuals have not been used in incentive programs for
smoking cessation and studies to investigate their unique contribu-
tion to treatment outcome have not been conducted.

Overall, the results of incentive- and competition-based programs
are very promising. However, almost all of these studies have been
conducted in small worksites, and systematic replications of these
findings in controlled investigations in larger companies are needed.
Future research should also investigate the types of personal
incentives (e.g., paycheck bonuses versus lotteries for quitters) and
competitions (e.g., within a worksite versus between worksites) that
work best in different organizations. Finally, nonsmokers should be
carefully considered in incentive programs-in their role as support-
ers of quitters and as nonsmokers, wit.h bonuses for all nonsmokers,
old and new-and worksite resources should be provided for all
employees.

High Risk Populations

With the exception of the few large-scale clinical trials listed in
Table 2 (e.g., Kornitzer, Dramaix et al. 1980, Li et al. 1984; Rose et al.
1980; Schlegel et al. 1983), the majority of participants in worksite
smoking programs have been young, from middle or upper socioeco-
nomic levels, and in occupations that do not place them at increased
health risk. Although smoking cessation efforts should continue with
such populations, an argument can be made for focusing efforts on
individuals at particularly high risk for cancer, cardiovascular
disease, or respiratory disease-the major causes of excess morbidity
and mortality due to cigarette smoking. Given limited resources, it
should be more cost effective to direct interventions primarily
toward those most likely to develop disease. Three overlapping
approaches have been used to reach the following types of high risk
smokers in the worksite: blue-collar male workers, workers at risk
because of occupational hazards, and individuals predisposed to
disease for reasons in addition to smoking (e.g., obesity, hyperten-
sion, abnormal lipid levels).

Few worksite programs have been offered by companies employing
primarily male blue-collar workers, even though such groups have
higher than average smoking rates. Although there is little or no
documentation of the reasons for this inconsistency, it may be due to
mistaken beliefs among health professionals that such individuals
would be less likely to participate in or follow through with a
smoking modification program. Ellis (1980) found that blue-collar

498


workers may be at least as interested in quitting as others in the
general population. Resistance on the part of some labor unions may
constitute another reason for fewer programs conducted in blue-
collar worksites.
Initial worksite smoking research with blue-collar workers has
been conducted both at military bases and in Veterans' Administra-
tion (VA) hospitals. Schlegel and colleagues (1983) offered programs
at 28 different military bases across Canada. Although their outcome
results were comparable with other studies, they did report relative-
ly low followthrough (compliance) rates with homework assign-
ments. Also, they must have failed to recruit a high percentage of
smokers, because an average of fewer than 10 smokers per base
participated. Perhaps a different intervention approach is needed for
blue-collar populations-one that does not require employees to
devote much time and effort, and one that tailors materials and
tasks to be appropriate socioculturally. Dawley and colleagues (1980,
1984) conducted research with employees and patients in VA
hospitals and have advocated increased smoking modification efforts
within the VA system. They point out that it is ironic that for many
years the Nation's largest health care provider sold cigarettes at
`cutrate prices, owing to tax-exempt status.
Little systematic research with smokers in jobs that place them at
risk because of occupational hazards has been reported. For example,
asbestos workers are at high risk for respiratory disease, and
asbestos workers who smoke increase their risk synergistically (US
PHS 1977; see the chapter on asbestos in this Report). Ellis (1980)
informally reported on the results of a program for former employ-
ees of an asbestos company that involved incorporating antismoking
advice into regularly scheduled appointments with company physi-
cians, pairing written self-help materials with feedback on physical
status, and offering individual smoking cessation counseling. Over a
4-year period, this relatively low cost intervention was associated
with a 30 percent reduction in the proportion of employees who
reported being smokers. A related publication by Ellis (1979)
provides suggestions for recruiting and treating asbestos workers.
The most extensive no-smoking program involving high risk
occupations has been been conducted by the Johns-Manville asbestos
company. In addition to a smoking ban throughout the worksite and
a company policy of no longer hiring new employees who smoke
(Cooper 1978), Johns-Manville launched an intensive antismoking
campaign at 14 company sites. This program involved an educational
campaign coordinated with SmokEnders cessation clinics and the
institution of the companywide smoking ban. Although systematic,
published reports of this program could not be located, Orleans and
Shipley (1982) reported participation rates of 15 to 20 percent in the
cessation clinics and an approximately 75 percent posttreatment


quit rate among participants (apparently without biochemical vali-
dation). In the only controlled study to date of worksite intervention
in high risk occupations, Li and colleagues (1984) recently studied
the effects of physician stop-smoking advice on asbestos-exposed
naval shipyard workers. They found, somewhat surprisingly, that
subjects who had abnormal pulmonary function tests did not have
higher cessation rates (4 percent prolonged abstinence) than workers
with normal pulmonary function tests (6 percent prolonged absti-
nence) who received the same intervention.

The third approach to reaching high risk participants has been to
conduct comprehensive health screenings to identify individuals at
risk for the development of chronic disease. Rose and colleagues
(1980) assigned people at high risk of developing heart disease who
worked in 24 large industrial companies to intervention or screen-
ing-only control conditions. High risk people in the intervention
condition were more successful at quitting smoking (12 percent
cessation at 5 years) than similar subjects in the control condition (0
percent cessation rate). Similar results were reported by Kornitzer
and Dramaix and colleagues (1980) in a parallel large-scale trial
conducted in Belgium (see Table 3).

In summary, some promising initial studies have been conducted
with high risk individuals. But such studies are few in number and
much more intensive study of ways to best reach high risk
individuals is needed.

Multiple Risk Factor Reduction Programs

A number of organizations have offered smoking cessation pro-
grams as part of employee wellness or lifestyle-modification pro-
grams. Such programs typically focus on achieving modifications in
several risk factors in addition to cigarette smoking, such as obesity,
elevated cholesterol levels, hypertension, and a sedentary lifestyle.
Some programs also include components on stress management and
modifying Type A behavior. Almost all programs include an initial
health screening to identify risk factors, but subsequently there is a
considerable divergence in approaches. Some programs focus solely
on high risk participants (e.g., Meyer and Henderson 1974; Ware and
Block 1982); others invite all employees to participate regardless of
risk status (e.g., Naditch 1984). There is also considerable variation
in how smoking programs are implemented, with some programs
holding separate meetings for smokers and others including informa-
tion on smoking modification as part of their general wellness
program.

The concept of providing smoking modification services as part of
a more general lifestyle program is appealing. Stopping smoking can
be seen as one aspect of adopting a more healthy lifestyle, and other
program components such as increased levels of exercise may

500