A review of computer and Internet-based interventions for smoking behavior Scott T. Walters a, T , Jo Anne Wright b , Ross Shegog c a University of Texas School of Public Health, 5323 Harry Hines Blvd, V8.112, Dallas, TX 75390-9128, United States b University of Michigan, United States c Center for Health Promotion and Prevention Research, University of Texas School of Public Health, Texas Medical Center, Houston, United States Abstract This article reviews studies of computer and Internet-based interventions for smoking behavior, published between 1995 and August 2004. Following electronic and manual searches of the literature, 19 studies were identified that used automated systems for smoking prevention or cessation, and measured outcomes related to smoking behavior. Studies varied widely in methodology, intervention delivery, participant characteristics, follow-up period, and measurement of cessation. Of eligible studies, nine (47%) reported statistically significant or improved outcomes at the longest follow-up, relative to a comparison group. Few patterns emerged in terms of subject, design or intervention characteristics that led to positive outcomes. The bfirst generationQ format, where participants were mailed computer-generated feedback reports, was the modal intervention format and the one most consistently associated with improved outcomes. Future studies will need to identify whether certain patients are more likely to benefit from such interventions, and which pharmacological and behavioral adjuncts can best promote cessation. D 2005 Elsevier Ltd. All rights reserved. Keywords: Smoking; Cessation; Computer; Internet; Intervention 1. Introduction The recent Institute of Medicine (2001) report paints a picture of an outdated healthcare system in need of innovative and cost-saving methods for improving health outcomes. In this process, smoking 0306-4603/$ - see front matter D 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.addbeh.2005.05.002 T Corresponding author. Tel.: +1 214 648 1519; fax: +1 214 648 1081. E-mail address: [email protected] (S.T. Walters). Addictive Behaviors 31 (2006) 264 – 277
14
Embed
A review of computer and Internet-based interventions for smoking behavior
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Addictive Behaviors 31 (2006) 264–277
A review of computer and Internet-based interventions for
smoking behavior
Scott T. Waltersa,T, Jo Anne Wrightb, Ross Shegogc
aUniversity of Texas School of Public Health, 5323 Harry Hines Blvd, V8.112, Dallas, TX 75390-9128, United StatesbUniversity of Michigan, United States
cCenter for Health Promotion and Prevention Research, University of Texas School of Public Health,
Texas Medical Center, Houston, United States
Abstract
This article reviews studies of computer and Internet-based interventions for smoking behavior, published
between 1995 and August 2004. Following electronic and manual searches of the literature, 19 studies were
identified that used automated systems for smoking prevention or cessation, and measured outcomes related to
S.T. Walters et al. / Addictive Behaviors 31 (2006) 264–277266
promise. At the time of the Strecher (1999) review, however, computer interventions for smoking were
nearly all limited to using the computer to generate printed feedback, which was then mailed to the
recipient. However, in recent years, a few intervention trials have begun to test more sophisticated
methods. Studies now describe multiple iterations of feedback, specific advice or a tailored plan for
quitting, computer generated e-mail reminders, or other multimedia experiences. Such aspects are
particularly apparent in computer programs targeted at youth, some of which incorporate Flash
technology, interactive responses, chat rooms, or video streaming. As discussed below, these newer
interventions are noteworthy both because of their sophisticated presentation (i.e., they look and feel
different), as well as their ability to customize the intervention ipsitively based on the user’s responses to
the program (i.e., they ask questions and respond to the user). By soliciting information and allowing the
program to respond with visual or audible responses, a computer can better mimic the transactional
qualities of human communication (Cassell, Jackson, & Cheuvront, 1998). Such interventions are also
thought to be more persuasive than static text. That is, if the receiver feels there is a bgive and takeQ inthe transaction, they will be more likely to attend to the message, comprehend the argument, and
consider the position (O’Keefe, 1990). The possibilities for tailoring smoking interventions using
interactive computer programs are vast, but the development costs can be high (Science Panel on
Interactive Communication and Health, 1999). These costs may be justified if the program impacts
populations as effectively as other programs or reaches populations which have typically been resistant
to other kinds of interventions. For these reasons, the present study was undertaken to update the
Strecher (1999) review. We were interested in smoking prevention or intervention programs that used the
computer or Internet to calculate or deliver the intervention. Because our intent was to examine the range
of applications under which the automated systems are being used, we chose to examine both adult
cessation and adolescent prevention studies. Although conceptually different, we hoped that one area
might inform the efforts of the other.
3. Methods
3.1. Search strategy
Medline, CINAHL, and PsycInfo databases were used to locate English-language studies published
between 1995 and August 2004. The bibliographies of retrieved articles were scanned for additional
references. Key search terms included (computer or Internet or web) and (behavior change or
intervention or treatment or therapy) and (smoking or tobacco). The search terms were intentionally
broad to ensure, as much as possible, that all relevant articles would be captured.
3.2. Inclusion criteria
Review criteria included English-language peer-reviewed journal articles published since 1995 that
described intervention trials for smoking prevention or cessation that:
1. Used computers (web-based, server-based, or stand-alone programs) as a significant part of the
intervention.
2. Included at least one comparison or control condition.
3. Included at least one intervention condition that was achieved without significant human contact. This
eliminated interventions that involved non-automated e-mail, chat rooms, discussion boards, or direct
personal contact as the primary delivery mode.
4. Reported at least one outcome directly related to smoking behavior.
Abstracts of journal articles were examined independently by two reviewers using the above criteria
to determine relevance. In cases where multiple outcomes were reported from a single intervention trial,
we selected the most recent (or longest) follow up available.
3.3. Data synthesis
A qualitative analysis was undertaken due to the significant heterogeneity between studies in terms of
subject characteristics, intervention delivery, and outcome measure. Comparisons between studies were
made on the basis of continuous (i.e., no smoking from baseline to follow-up) or point abstinence (e.g., no
smoking for a period of time prior to follow-up) rates, which wewere able to derive frommost studies. This
provided the most useful basis for comparison as the studies differed markedly on most other attributes.
4. Results
Literature searches yielded 199 unique references that met our search criteria. After reviewing articles
for relevance, 19 were retained for this review. Four studies were focused on adolescent smoking
prevention, while 15 targeted adult smokers. Tables 1 and 2 summarize the resulting studies. The number
of study participants ranged from 65 to 8352. Participants ranged from 11 to 65 years old, and the
percent of female participants varied from 40.5% to 100%. Follow-up periods ranged 1 to 24 months.
Studies included both treatment seeking, as well as non-treatment seeking participants.
4.1. Adolescent studies
Interventions targeting adolescents have the goals of delaying onset of smoking among those who
have never tried cigarettes or have only experimented with use, or encouraging cessation among regular
users. Of the four studies that targeted adolescents, two reported a significant reduction in smoking
initiation and prevalence as the result of computer-tailored material sent to the home of the student.
Ausems, Mesters, van Breukelen, and de Vries (2002) randomized elementary school students to
receive: (1) a seven-session in-school program focused on the social factors that influence people to
smoke, education about the effects of smoking on the body, and training in refusal skills; (2) three mailed
letters, tailored to students based on beliefs, efficacy, and intent to smoke; (3) both in-school and mailed
conditions, or; (4) control. At a 6-month follow-up, the mailed condition reduced smoking initiation
(10.4% vs. 18.1%) and continuation (13.1% vs. 23.5%) relative to control. The effects of the in-school
program did not significantly differ from control, nor was the combined approach superior to the mailed-
only condition. Using a similar intervention design with vocational school students, Ausems, Mesters,
van Breukelen, & De Vries (2004) found that those in the in-school condition who had ever tried a
cigarette were less likely to continue smoking at 12 months relative to control (29.4% vs. 42.2%,
respectively). The mailed condition prevented smoking initiation at 18 months, relative to control
Table 1
Effects of computer interventions on smoking cessation among adolescents
Study Participants Follow-up
(Months)
Intervention conditions Smoking Summary of findings
Initiat Contin
Ausems
et al.
(2002)
3734 students
ages 11–12
6 1. Seven-session teacher-led
program
14.9ab 21.6ab Mailed intervention reduced
smoking initiation and
prevalence at 5 months.
No effect of the in-school
program over control, or
combined program over
mailed intervention.
2. Three tailored letters
mailed to students’ homes
10.4a 13.1a
3. Seven session teacher-led
program+3 tailored letters
mailed to students’ homes
15.2ab 14.2ab
4. No treatment control 18.1b 23.5b
Ausems
et al.
(2004)
36 vocational
schools (students
ages 12–16 years)
6/12 1. Three-session teacher-led
program
28.0a 29.4a In-school intervention reduced
smoking at 12 months, as
compared to control. Mailed
intervention reduced smoking
initiation at 18 months, as
compared to control (in-school
condition not assessed at 18
months). No additional effect
of the combined intervention.
2. Three tailored letters
mailed to students’ homes
25.0a 37.0ab
3. Three-session teacher-led
program+3 tailored letters
mailed to students homes
29.4a 45.0ab
4. No treatment control 40.9a 42.2b
Aveyard
et al.
(2001)
8352 students
ages 13–14
12/24 1. Three class sessions+3
interactive computer sessions
17.7a 83.2ay No benefit of intervention in
terms of initiation or cessation
at 1 or 2 years.2. No treatment control 16.5a 85.0ay
Pallonen
et al.
(1998)
135 adolescent
smokers
6 1. Tailored computer program
focused on smoking cessation
59.4az 30% of participants made at
least one 24-h quit attempt at
6 months, with no differences
between groups.
2. Standard baction orientedQcomputer program
59.4az
Initiat=Percent of baseline never-smokers who initiated smoking during longest follow-up period; Contin=Percent of baseline
ever-smokers who reported smoking at longest follow-up period.
Letters indicate significant ( p b .05) contrasts between intervention conditions: a=a, a=ab, b=ab, apbpc.y Continuation rates: bRegular weekly smokingQ at baseline, bregular weekly smokingQ at follow-up.z No 24 h quit attempts during the last 2 months.
S.T. Walters et al. / Addictive Behaviors 31 (2006) 264–277268
(27.2% vs. 47.9%, respectively), but the in-school condition was not reassessed at 18 months. There was
no additional effect for the combined condition at any follow-up point.
Conflicting results were reported by Aveyard et al. (2001) who failed to find differences between
groups that received a classroom lecture or computer intervention. At 1- and 2-year follow-ups, the
groups showed similar rates of smoking and stage-of-change movement. Pallonen et al. (1998) compared
a computer intervention with material consistent with the student’s stage of change, to a computer
intervention that emphasized only bactionQ strategies (e.g., suggestions for changing behavior). At a 6-
month follow-up, the authors found no difference between the two groups in terms of cessation.
4.2. Adult studies
All adult interventions targeted cessation in regular smokers. Of the 15 studies targeting adult
smokers, seven reported significantly improved cessation among treatment subjects as compared to
The modal intervention format involved computer-generated feedback and advice sent through the
mail. Prochaska and colleagues tested the efficacy of mailed material in two trials. In the first trial
(Prochaska, Velicer, Fava, Rossi, & Tsoh, 2001), smokers were randomized to a mailed intervention or
control. The mailed intervention included information about the pros and cons of quitting, a normative
comparison, and advice commensurate with the smoker’s stage of change. At 24 months, the mailed
information produced 26% point abstinence, as compared to 20% in the control condition. In the second
study (Prochaska, Velicer, Fava, Ruggiero et al., 2001), a series of brief provider phone contacts were
tested as an adjunct to the mailed report. At 24 months, those who received the report again showed
greater point abstinence (21%) as compared to control (15%), with no additional benefit of phone contact.
Using a similar design, Curry, McBride, Grothaus, Loutie, and Wagner (1995) recruited smokers from
a large health maintenance organization (HMO) who were not requesting help with cessation. Using an
incremental design, the authors randomized participants to receive: (1) a self-help manual; (2) a self-help
manual and computer-generated feedback; (3) a self-help manual, computer-generated feedback, and
three telephone calls to reinforce use of the self-help materials; or (4) control. The feedback included
information on stage of change, self-identified reasons for wanting to quit, and cessation advice. The
counseling condition focused on encouraging smokers to use the self-help materials. Telephone
counseling (but not mailed feedback) significantly increased cessation at 3 months (11%, 4%, 5%, and
6% point prevalence for counseling, feedback, booklet, and control conditions, respectively). However,
differences were nonsignificant at 12 and 21 months.
Etter and Perneger (2001) randomized participants to a mailed intervention or control. An average of
1.5 times over a 6-month period, intervention participants received a letter tailored to stage-of-change,
level of dependence, self-efficacy, and other characteristics. Seven months after beginning the program,
intervention participants were more likely to be abstinent than control participants (5.8% vs. 2.2% point
prevalence). In a similar study, Dijkstra, De Vries, and Roijackers (1998) found that tailored information
that emphasized both the benefits of smoking cessation, as well as advice on cessation skills, was more
effective than either presented in isolation.
Three studies compared computer-tailored information to similar, but untailored, information.
Borland, Balmford, Segan, Livingston, and Owen (2003) randomized smokers calling a quitline service
to: (1) mailed untailored self-help materials, (2) computer-generated tailored advice, or (3) computer-
generated tailored advice and brief telephone counseling. At a 3 month follow-up, more individuals in
the tailored advice and counseling condition were abstinent (21% point prevalence) than in the other
conditions (12%), but by 12 months, the effect was nonsignificant when adjusted for multiple
comparisons. Lennox et al. (2001) compared computer-tailored information, non-tailored information, or
control. Interestingly, the authors found that among heavier smokers, the non-tailored letter (but not the
tailored letter) produced greater cessation rates as compared to control. At 6 months, the non-tailored
Notes to Table 2:
Cessation=Rates of cessation at longest follow-up for 24 h point abstinence and 30 day point abstinence. Letters indicate
significant ( p b .05) contrasts between intervention conditions within outcome category: a=a, a=ab, b=ab, apbpc.* Continuous abstinence.y 7 day point abstinence.z 7 day quit attempt.§ Values averaged across number of contacts.f Sustained abstinence of 10 weeks.
S.T. Walters et al. / Addictive Behaviors 31 (2006) 264–277272
group reported cessation rates of 4.4%, as compared to 3.5% and 2.6% in the tailored and control groups,
respectively.
Three studies tested multiple iterations of feedback. Dijkstra, De Vries, Roijackers, and van
Breukelen (1998) found that multiple tailored letters were more effective than either non-tailored
letters or a single tailored letter. In another study, Dijkstra, De Vries, and Roijackers (1999) found that
multiple tailored letters produced greater stage transition and intent to quit, but not actual quit
attempts. Finally, Velicer, Prochaska, Fava, LaForge, and Rossi (1999) compared interactive vs. non-
interactive computer reports, crossed with number of mailed contacts (i.e., one, two, three, or six). At
an 18-month follow-up, there was an effect of the stage-matched materials (average 30-day cessation
rates of 18.4% and 13.1% for tailored and non-tailored, respectively), but no dose-response effect in
terms of number of mailings.
Two studies tested the efficacy of computer-generated feedback as an adjunct to nicotine replacement
therapy. One study (Shiffman, Paty, Rohay, Di Marino, & Gitchell, 2000) evaluated the effectiveness of
feedback materials as an adjunct to nicotine polycrilex gum. In this study, participants who had
purchased nicotine gum were randomized to receive computer-tailored materials with or without a phone
call, or to receive only the standard information packaged with the gum. At 12 weeks, participants who
had received the tailored information reported significantly higher 30-day abstinence rates (27%) as
compared to control (18%), with no additional effect of the phone call (27%). The intervention in this
study appeared to be effective across gender, level of dependence, and self-efficacy. In contrast,
Shiffman, Paty, Rohay, Di Marino, and Gitchell (2001) did not find an effect when similar tailored
information was compared to standard materials among smokers receiving the nicotine patch. At 12
weeks, abstinence rates did not differ between the two groups. However, among participants who
reported using the materials, cessation rates were generally higher.
Three adult studies tested direct contact with interactive computer programs. Lawrence, Aveyard,
Evans, and Cheng (2003) evaluated a cessation program for expectant mothers. Pregnant smokers
presenting at a general practice clinic were randomized to receive: (1) brief cessation advice delivered by
a midwife, (2) bEnhancedQ cessation advice from a midwife trained in the Transtheoretical Model, plus a
series of self-help manuals, or (3) bEnhancedQ cessation advice from a midwife trained in the
Transtheoretical Model, a series of self-help manuals, and a 20-min computer program that provided
feedback on stage-of-change and other aspects of smoking. At 30-weeks gestation and 10 days postnatal,
there were no significant differences between groups in terms of cessation rates. Lenert, Munoz, Perez,
and Bansod (2004) tested an automated email system that sent timed educational messages to smokers.
The authors used two consecutive waves of participants visiting a website. The first wave received a
single-point-in-time educational message via a website, and the second wave received an enhanced
website intervention that also sent follow-up emails. Although the two groups were not randomly
assigned, the authors suggest that they did not differ on factors associated with 30-day quit rates. At a
30-day follow-up, the enhanced intervention appeared to increase the rate at which participants set a quit
date (97%) as compared to the standard website (91%). Participants in the enhanced group also reported
greater rates of 24-h quit attempts (83% vs. 54%) and 30-day quit rates (13.6% vs. 7.5%). Finally,
O’Neill, Gillispie, and Slobin (2000) compared a computer intervention targeted to college students to a
non-smoking focused control intervention. The smoking intervention consisted of four computer
sessions that progressed along the lines of the stages of change. Early sessions emphasized
consciousness raising and self-evaluation, whereas later sessions targeted commitment and planning.
At the end of the 6-week study period, 48% of the intervention participants showed forward movement
in their stage of change as compared to 21% in the control group. Although there was an initial trend
towards greater quitting in the intervention group, by seven months quit rates did not differ between the
groups.
5. Discussion
While computer-based smoking prevention and cessation programs show promise in influencing
tobacco-related behaviors, published studies show mixed results in terms of translating the educational
experience to real-world practice. Of the 19 automated, computer-based interventions that we reviewed,
nine (47%) showed evidence of effectiveness at the longest follow-up.
This review should not be considered an exhaustive analysis. The scope of our investigations was
limited to English-language publications catalogued in three specialist databases. Published studies of
this type are also subject to publication bias-the tendency for studies that show beneficial effects to be
published. As we have noted, study components, subject characteristics (demographics and intention to
treat), length of follow-up, and outcome measures were highly variable. This heterogeneity of methods
makes it difficult to determine consistent predictors of efficacy. Insufficient reporting of the
interventions, subject characteristics, and outcome measures further compounded this difficulty. For
instance, in some studies the outcome measure was simply described as babstinenceQ or bpointabstinenceQ with no further information on how these outcomes were operationalized.
Despite these limitations, three observations are consistent with a previous review (Strecher, 1999).
Studies with smokers who were treatment seeking (e.g., who wanted to quit smoking) typically resulted
in higher abstinence rates than studies that used general population samples of smokers. Second, most
studies that used multiple follow-up points found that the effects of the intervention were lessened with
time. Finally, in all cases, the intervention conditions produced rates of abstinence that were at least
equivalent, if not higher, than non-treatment controls. That is, no intervention condition seemed to make
smokers more likely to smoke relative to control.
Aside from these broad observations, there remains a lack of clarity about what types of computer-
based applications are most effective. The paradox is that while the number of smoking cessation
programs is growing (high dissemination), we have little understanding of how, why, and under what
conditions, such interventions might work (low evaluation). For instance, although Edwards, Elliott,
Conway, and Woodruff (2003) identified 87 Internet programs devoted to teen smoking cessation, our
review found no outcome studies of any of these programs in the peer reviewed literature. To estimate
and improve the effectiveness of computer- and Internet-based interventions, it will be important for
future research efforts to emphasize the importance of (1) theoretical foundations to design and
develop computer-based programs, and (2) rigorous evaluation methods to determine their
effectiveness.
The use of computers for generating tailored interventions has evolved with the capabilities of
computer technology. As with Strecher’s (1999) review, the modal intervention appeared to follow
what Brug, Campbell, and van Assema (1999) call bfirst generationQ formats. In this format,
participants have no direct contact with the computer. Typically, participants complete a paper
screening instrument, which is then entered into the computer and used to derive mailed feedback. Of
the 14 first generation interventions in this review, eight (57%) reduced smoking over control. In a
bsecond generationQ format, participants interact directly with the system. This format provides
S.T. Walters et al. / Addictive Behaviors 31 (2006) 264–277274
immediate communication between participant and program, but the participant must have access to,
and know how to operate, a computer. Research using second generation tailoring has demonstrated
improved disease management and behavioral determinants of awareness and intentions in other areas
(Bartholomew et al., 2000; Onema, Brug, & Lechner, 2001; Shegog et al., 2005). However, the
effectiveness of second generation tailoring has not been widely tested in the field of smoking
cessation. Only one intervention, a smoking website that generated personalized email messages,
appeared to use this format (Lenert et al., 2004). This intervention increased cessation over control at
12 months. The recently emerging bthird generationQ of computer-based applications employs both
iterative as well as ipsitive feedback, adjusting at multiple time points to the characteristics of the user.
These programs are structured modularly (rather than as a single linear program), so that they can
better adjust to the user’s needs. For instance, whereas a second generation program might tailor a
series of messages, a third generation intervention might add, delete, or rearrange components in
response to the user. None of the third generation interventions in this review (2 for adolescents, 2 for
adults) had a significant impact on smoking over control.
Because the number of computer-based health education programs has increased significantly over
the past ten years, there is a need for smoking prevention and cessation programs that are