- Email: [email protected]
Social support in smoking cessation: In search of effective interventions
nm,~_xqoa THERAPY17, 607-619 (1986)
Social Support in Smoking Cessation: In Search of Effective Interventions EDWARD LICHTENSTEIN
University of Oregon and Oregon Research Institute RUSSELL E. GLASGOW
Oregon Research Institute DAVID B. ABRAMS
Brown University~The Miriam Hospital
The results of five recent smoking cessation studies from three separate research programs are summarized. Each study compared a basic cognitive-behavioral cessation program to the same program plus a component designed to enhance social support. Four of the studies found process or correlational data linking social support to outcomes. Nevertheless, there were no significant between-groups smoking-outcome differences in any of the five studies. Several explanations for these findings are considered and it is concluded that social support deserves further consideration in interventions for smoking reduction. Suggestions for future research are offered.
T h e r e is s u b s t a n t i a l e v i d e n c e suggesting t h a t p e r s o n s w i t h r e l a t i v e l y h i g h l e v e l s o f social s u p p o r t r e p o r t f e w e r p s y c h o l o g i c a l a n d p h y s i c a l s y m p t o m s . O n e w a y t h a t s o c i a l s u p p o r t m a y i n f l u e n c e p h y s i c a l h e a l t h is t h r o u g h c h a n g e s in p e r f o r m a n c e o f h e a l t h - r e l a t e d b e h a v i o r s s u c h as d e c r e a s e d c i g a r e t t e a n d a l c o h o l use o r i m p r o v e d d i e t a n d e x e r c i s e r e g i m e n s ( C o h e n & S y m e , 1985; K r a n t z , G r u n b e r g , & B a u m , 1985). B o t h t h e o r e t i c a l p r o p o s i t i o n s a b o u t t h e f a c i l i t a t i v e r o l e o f s u p p o r t in b e h a v i o r c h a n g e (e.g.,
Preparation of this report was supported by the National Heart, Lung, and Blood Institute grants, grant numbers HL29547, HL33739, and HL32318, awarded to Edward Lichtenstein, Russell E. Glasgow, and David B. Abrams respectively. Each project is a collaborative endeavor and we appreciate the assistance and social support of our colleagues. Requests for reprints should be sent to Edward Lichtenstein, Department of Psychology, University of Oregon, Eugene, Oregon 97403.
607 0005-7894/86/0607-061951.00/0 Copyright1986by Associationfor Advancementof BehaviorTherapy All rightsof reproductionin any formreserved.
608
LICHTENSTEIN, GLASGOW, AND ABRAMS
Janis, 1983) and empirical observations relating social support variables to performance of health-related behaviors (Levy, 1983) are congruent with this notion. The importance of social support is also suggested by its incorporation in various intervention programs for addictive behaviors. Programs such as Alcoholics Anonymous, Weight Watchers, and The Five Day Plan for Smoking Cessation all embody social support components such as buddy systems and group support as central features of their approach. Given that alcoholism, obesity, and smoking have been slow to yield to standard behavioral (or other) treatment strategies and given the intuitive, theoretical, and empirical plausibility of social support, it is not surprising that investigators have recently attempted to develop social support intervention strategies (Colletti & Brownell, 1982). These programs have usually attempted to combine or integrate a social support intervention with standard cognitive-behavioral strategies. This approach is most fully developed in the area of obesity where the promising work of Brownell and colleagues (Brownell, Heckerman, Westlake, Hayes, & Monti, 1978) stimulated a number of studies. The results of more recent social support interventions for obesity, however, have been equivocal (e.g., Weisz & Bucher, 1980). Given this theoretical and empirical background, it has been suggested that it would be fruitful to apply social support interventions to smoking cessation (Lichtenstein, 1982). Social support from the social environment could be a source of motivation to expend the effort to try and quit smoking and a source of reinforcement for successful accomplishments. Social support might also provide a buffer against the stress of quitting (e.g., withdrawal) or other stressful events that might precipitate a relapse (Cohen & Wills, 1985). Our three research groups have been independently pursuing this line of investigation over the past several years. We have utilized operationally distinct but conceptually similar social support treatment strategies in the context of smoking modification programs in both community-based clinics and at the worksite. While sharing our experiences and results informally, we were struck by the similarity in our findings. We believed that summarizing the intervention strategies we developed and the results we have obtained in a single paper would be useful to both workers in smoking cessation and to students of social support. To telegraph our basic message, our three programs have found that measures of social support are positively correlated with better treatment outcomes but that attempts to both increase social support and to enhance treatment effectiveness have not been successful. The fact that each of our programs has replicated its treatment outcome findings, together with the convergence of the findings across the three projects, seems to us to be noteworthy. We will describe and summarize the major findings from each of the projects and then offer some speculations about the implications of these data for future research.
SOCIAL SUPPORT IN SMOKING
609
The Oregon Program Lichtenstein and his colleagues have focused on socially supportive behaviors from a spouse or living partner of the person trying to quit smoking. The Partner Interaction Questionnaire (PIQ) consists of 61 behaviors related to smoking cessation, both positive (e.g., my partner complimented my not smoking) and negative (e.g., my partner commented on my lack of willpower). Subjects first judge the frequency of occurrence of each item and then evaluate how helpful each item was in their effort to stop smoking. A summary measure of partner support, experienced helpfulness, is derived by summing the cross products of the frequency and helpfulness scores. Mermelstein, Lichtenstein, and McIntyre (1983) showed that experienced helpfulness scores obtained at the end of treatment predicted smoking status up to six months posttreatment. These correlational data, plus some of the promising early findings in the application of spouse support to the treatment of obesity (Brownell et al., 1978), led us to derive a partner support program for smokers (McIntyre-Kingsolver, Lichtenstein, & Mermelstein, 1986). Smokers (N = 64) with spouses who were willing to cooperate were randomly assigned to either a multicomponent cognitive-behavioral smoking program (six 2-hour sessions plus intake) or the same program with spouses attending all sessions and receiving training and encouragement. The basic program (Brown, Lichtenstein, McIntyre, & Harrington-Kostur, 1984) includes nicotine fading, self-management training, and relapse prevention. Follow-up data were obtained l, 2, 3, 6, and 12 months posttreatment and self-reported smoking status was corroborated by informants and carbon monoxide measures (at posttreatment and six month follow-up only). PIQ data were obtained at the end of treatment to assess the impact of training. During treatment sessions couples were given feedback about helpful and unhelpful spouse behaviors related to smoking cessation. Common areas of helpful or unhelpful behaviors were pointed out and group members were encouraged to contribute examples from their own experience. Each couple was encouraged to problem solve regarding what kinds of spouse behaviors would be most helpful for the smoker in his/her cessation efforts. Guided group discussions and suggested homework exercises were used to encourage couple problem solving. Spouses were encouraged to reward their partners and to participate in program activities as suggested by both the program and their partner. There was a marginally significant difference in cessation rates favoring the spouse support condition at the end of treatment (72.7% versus 48.4%; p < . 10). Although follow-up abstinence rates for the partner support group were consistently in the expected direction, the differences were neither statistically significant nor clinically meaningful (see Table 1). Posttreatment PIQ scores tended to be higher for the subjects receiving spouse support training, but the difference did not reach significance (spouse training M = 22.3, SD = 18.2; control M = 15.3, SD = 12.6; F = 3.05,
610
LICHTENSTEIN, GLASGOW, AND ABRAMS TABLE 1 DESIGN AND OLr'FCOME OF SOCIAL SUPPORT INTERVENTIONS Results: cessation rates
Study
Sample characteristics
Design
Post-Rx
Follow-up
MclntyreKingsolver et al. (1986)
64 Ss with cooperat i r e spouses, 57% female, M a g e = 38; M cigs/day = 25.6
Basic p r o g r a m (n = 31) Basic p r o g r a m + spouse support (n = 33)
48.4%
32.3% (1 yr)
72.7
36.3
Lichtenstein, et al. (1985)
64 Ss 50% female, M age = 39; M cigs/ d a y = 25.8
Basic p r o g r a m (n = 21) Basic p r o g r a m + spouse s u p p o r t (n = 15) Single S s (n = 28)
57.1
23.8 (6 mo)
66.7
33.3
53.6
35.7
24 medical clinic & telephone Co. e m ployees, 83% female, M age = 34; M cigs/day = 24
Basic p r o g r a m (n = 12) Basic p r o g r a m + coworker support (n = 12)
17.0
27.0 (6 m o )
17.0
17.0
Glasgow et al. (in press)
29 hospital svcs & savings & loan e m ployees, 69% female, M age = 34; M cigs/clay = 25
Basic p r o g r a m Basic p r o g r a m + sig. o t h e r support
54.0 36.0
25.0 (6 m o ) 23.0
A b r a m s et al. (1985)
90 m f g & health insurance employees, 63% female, M age = 41.3; M cigs/ d a y = 28.1
Basic p r o g r a m + health educ.
63.0
23.3 (6 m o )
47.0
23.3
43.0
20.0
Malott et al. (1984)
(n =
30)
Basic p r o g r a m + cognitive trng (n =
30)
Basic p r o g r a m + s u p p o r t trng (n = 30)
df =
1, p < . 10). These results suggested that our training procedures did not have a strong impact on spouse helpfulness, at least as measured by the PIQ. Across all subjects, however, experienced helpfulness was significantly related to both cessation concurrently at the end o f treatment, M abstainers = 22.4, M smokers = 13.5; t(62) = 2.2, p < .05, and prospectively to maintenance o f cessation up to 3 - m o n t h follow-up. Subjects who initially quit but relapsed by three m o n t h s had significantly lower PIQ scores than subjects who remained abstinent, t(38) = 2.33, p < .05. These findings partially replicate the correlational results o f Mermelstein et al. (1983) and confirm the association between experienced helpfulness from a partner and the o u t c o m e o f smoking treatment.
SOCIAL SUPPORT IN SMOKING
611
We conceived several reasons for the weak results of this study. The spouse training components of the program may not have been intensive or salient enough. Both members of several couples in this first study were trying to quit smoking and therefore served in a dual role as both helper and helpee. Subanalyses indicated that such couples tended (but not significantly so) to do less well when only one member of the pair was trying to quit smoking (4 of 14 versus 9 of 19 abstinent at one year). Finally, we felt that the inclusion of a third group of subjects, either without spouses or with unwilling partners, would be informative. Accordingly, a second study, also with 64 subjects, was conducted which remedied these deficiencies (Lichtenstein, Mermelstein, Kamarck, & Baer, in preparation). A manual for the helping spouse was developed and more program time was devoted to spouse training. Only 3 of the 33 couple subjects recruited contained two smokers trying to quit. And finally, a group of subjects without spouses was also included. Because they could not be randomly assigned as were subjects with available spouses, this third group was a quasi-control or comparison group. The basic program and follow-up procedures were the same as in the first study. The results were also quite similar (see Table 1). Again, subjects in the spouse training condition tended to have higher abstinence rates at all assessment points but these differences were small and nowhere approached significance. Interestingly, subjects without spouses tended to have the highest abstinence rates of all, but again the differences were not significant. Posttreatment PIQ scores yielded a similar pattern of results; mean differences were clearly in favor of the spouse training subjects, but again did not reach statistical significance. Unlike our previous two studies, PIQ scores were not related to cessation or maintenance.
The North Dakota Program Glasgow, Klesges, and colleagues at North Dakota State University have investigated both co-worker and significant other social support in the context of their worksite smoking control program. A six-session, multicomponent, cognitive-behavioral program offering subjects the options of abstinence or controlled smoking has been developed and adapted for occupational settings (Glasgow, Klesges, Godding, Vasey, & O'Neill, 1984). The program involves weekly group meetings of 4-8 employees focused on making sequential reductions in nicotine content of brand smoked and number of cigarettes per day prior to a decision to attempt cessation (the preferred goal) or to make reductions in the percent of the cigarette smoked midway through the program. The last two sessions are devoted to solidifying behavior changes and to relapse prevention. As one of the major advantages of conducting smoking modification programs in occupational settings is purported to be the existence of peer and environmental support for behavior change (Chesney & Feuerstein, 1979; Stachnik & Stoffelmayr, 1981), we have evaluated the incremental effects of adding social support programming to the basic treatment. A
612
LICHTENSTEIN, GLASGOW, AND ABRAMS
38-item modification of the PIQ, revised to be applicable to controlled smoking and scored to produce both positive and negative interaction subscales, has been used to assess degree o f perceived social support. Six month follow-up data are obtained and carbon monoxide (CO) assessments are used to verify self-reports of smoking status. An initial study by Malott, Glasgow, O'Neill, and Klesges (1984) evaluated the effects of adding co-worker support procedures to the basic program. These procedures involved subject selection o f partners, use of a buddy system, a 17-page partner-support manual, and individualization o f support strategies. There were no differences between conditions at posttest or follow-up on either percent o f subjects abstinent (see Table 1) or on CO reductions among nonabstinent subjects. There were also no between groups differences on the modified PIQ measure. Negative, but not positive, support scores from the PIQ were related to treatment outcome. One unexpected finding to emerge from this study was that a treatment credibility questionnaire completed after the first treatment session revealed that the basic treatment plus social support condition was perceived as significantly less credible than the basic treatment program alone, F(1, 21) = 16.1, p < .001. There were at least three plausible explanations for the lack of betweengroups differences in this study. The basic treatment condition, which involved weekly small group meetings of co-workers who frequently interacted with each other and worked in relatively cohesive organizations, may have itself generated high levels of co-worker social support. Second, the additional partner support assignments increased the complexity of the program and may have interfered with subjects' efforts to change their own behavior. Finally, extrapolating from the correlational findings regarding the importance of negative social interactions related to smoking behavior, we may have placed too much emphasis on increasing positive co-worker interactions and not enough on decreasing negative smokingrelated interchanges. To address these issues, we conducted a second study (Glasgow, Klesges, & O'Neill, in press), this time employing significant others as partners. It was reasoned that the use of subject-identified significant others (usually spouses) would address the first two concerns because spouse support would not be expected to be at maximal levels and subjects did not have the dual role of both changing their own behavior and of being a "partner." To address the third concern we placed greater emphasis on decreasing well-intentioned but detrimental social interactions (e.g., nagging about smoking in the house). Twenty-nine subjects were randomly assigned to our basic treatment condition or to basic treatment plus social support. Significant other support procedures involved two group meetings of partners, semi-weekly mailings of sections of a revised partner support manual, and phone calls from therapists on alternating weeks to discuss progress. As in the first study, partners were provided with a list o f support behaviors that their " s m oker " identified as most helpful. In addition, individualized suggestions, drawn from the modified PIQ, were provided
SOCIAL SUPPORT IN SMOKING
613
as to which specific partner behaviors would be perceived as unhelpful. The importance of partners seeking feedback from subjects was also emphasized. Results were quite similar to those in the first study (see Table 1). Again, there were no differences between conditions at either posttest or followup on abstinence rates or on CO reductions among nonabstinent subjects. In fact, subjects in the basic treatment condition experienced nonsignificantly higher abstinence rates and greater CO reductions than subjects in the social support condition. There were again no between groups differences on the modified PIQ, but analyses of individual PIQ items targeted for intervention suggested that partners did engage in desired support behaviors. Despite the lack of between-groups differences, this investigation also revealed consistent correlational findings suggesting a relationship between social support, as measured by the modified PIQ, and treatment outcome. Specifically, the presence of "'positive" social support was not related to treatment outcome, but at posttest the absence of negative smoking related interactions (e.g., "expressed doubt about your ability to quit") was significantly correlated with concurrent self-report and biochemical measures of more successful treatment outcome (rs = .25 to .49 in Study 1 and .32 to .63 in Study 2).
The Brown Program Abrams, Pinto, Monti, and colleagues (1985) investigated three relapse prevention approaches in randomized trials across three worksite settings. Within each worksite, subjects (ns = 27 to 36) were randomly assigned to (a) cognitive behavioral management, (b) social support/social skills training, and (c) health education and nonspecific support (total N = 90). Groups of between 9 and 12 smokers met at the worksite with a malefemale co-therapist team for 11/2hours weekly for 8 weeks. The first four weeks consisted primarily of a standard nicotine fading program for smoking cessation using the Foxx and Brown (1979) procedure. Subjects set a quit day at Week 5. From Week 3 onwards the relapse prevention training was gradually introduced, with the last 4 sessions (Weeks 5-8) exclusively devoted to one of the three conditions described below. Cognitive behavioral management. This condition focused on intrapersonal coping only. All high-risk-for-relapse situations ultimately were identified as leading to intrapersonal reactions (stress, negative moods, withdrawal, urges to smoke, fatigue) and subjects were trained to use intrapersonal coping skills. The program used relaxation training, cognitive restructuring (thought-stopping and substitution of positive selftalk), and training in coping with "slips" (Abstinence Violation Effects). Within each session a skills acquisition model was used including instructions, modeling, practice, feedback, and homework. Social skills/support treatment. This relapse prevention program focused exclusively on social skills and social support network interventions (buddy systems at the worksite and at home) to prevent relapse. All high
614
LICHTENSTEIN, GLASGOW, AND ABRAMS
risk situations were identified as either originating in the social environm e n t or as requiring social support for coping. Participants were first asked to identify their " p r o x i m a l social networks," that is, the five individuals that they saw and confided in the most frequently, and to classify each person as potentiaily supportive, neutral, or unsupportive. Selection o f at least one supportive individual who had already successfully quit smoking was r e c o m m e n d e d . Participants were instructed how to deal effectively with persons who might act as smoking triggers and how to initiate an assertive request for someone else not to smoke. Conversational skills, the seeking out o f support when stressed, and handling praise or criticism from friends and relatives was practiced. Finally, subjects were trained to encourage neutral individuals to provide more active nonsmoking support (e.g., arranging exercise with friends). Once again, an intensive behavioral skills acquisition program was used, including instructions, modeling, feedback, role-playing, and homework. Health education and nonspecific support. Subjects in this condition received no behavioral skills training for relapse prevention. Instead, they were given information about the health consequences o f smoking and especially facts about "withdrawal" effects. A series o f nonspecific group discussions were held to aid the participants' efforts to quit smoking and to equate for contact time with the other two conditions. Participants sometimes discussed high-risk-for-relapse situations, but therapists provided no coping skills training per se. Overall results (see Table 1) revealed no statistically significant differences across the three treatment conditions or the three different worksite replications at posttest or follow-up (Abrams et al., 1985). Abstinence at follow-up was verified by CO in all cases. Thus, neither the social support nor the cognitive behavioral conditions were superior to the discussioncontrol condition at follow-up. Despite the negative overall results o f the social support program, correlational data based on pre- and posttreatment measures point to the importance o f social support variables in treatment. Based on a recent item-and-factor analysis o f the Partner Interaction Questionnaire (PIQ), a revised version o f the scales (consisting o f a 10-item " s u p p o r t i v e " subscale and a 10-item " u n s u p p o r t i v e " subscale) was administered pre- and posttreatment (Kamarck & Lichtenstein, personal c o m m u n i c a t i o n , Nov e m b e r 14, 1984). At pretreatment the supportive subscale o f the revised PIQ was significantly related to 6 - m o n t h follow-up smoking rate, r = - . 2 1 , p < .05). The ratio o f smoking friends to total friends in the social network o f the participants was also positively correlated with smoking rate at 6 - m o n t h follow-up, r = .33, p < .01). Analyses o f posttreatment measures o f social support revealed two subject characteristics correlated with 6 - m o n t h follow-up: a generalized social support for nonsmoking scale (Abrams & Habif, 1981; r = - . 4 6 , p < .01), and the PIQ unsupportive subscale, r = .24, p < .05. The PIQ results replicate previous research by Mermelstein et al. (1983) and the N o r t h Dakota program reported above. It was o f clinical interest to note that m a n y consumers
SOCIAL SUPPORT IN SMOKING
615
did not find aspects o f the social support training program to be useful; for example, they resisted suggestions such as the need to assertively request that other smokers not smoke in their presence.
DISCUSSION The results across the three different research programs are remarkably consistent. In five separate studies across the three projects the addition o f a social support component to a standard behaviorally-based cessation program in no case resulted in significant improvements in treatment outcome. Only in one of the two Oregon studies and only at the end of treatment was there an encouraging trend in favor o f the social support condition. The Oregon data, taken alone, could be viewed as encouraging evidence for partner support but they were limited by insufficient power to detect a small but real effect. The North Dakota and Brown studies also have limited statistical power but the patterns o f means and the consistency of the results across the three programs make it unlikely that we are committing a Type II error with respect to social support as conceptualized in these studies. Although there was some variation in support enhancement procedures, we recognize that only a limited range o f theoretical and operational social support options have been tested. It must also be acknowledged that the largely negative outcome data in these studies could be owing to the irrelevance o f support, as conceptualized in these studies, for smoking cessation o r to the inadequacy of our social support training. The basic hypothesis that increased support would facilitate cessation and maintenance may not have been tested adequately owing to our failure to substantially increase social support in the experimental groups. The PIQ data from the Oregon and North Dakota studies are consistent with this notion. The failure to increase social support may be related, as we discuss below, to the stability or "'traitlike" nature o f social skills or to dyadic support exchanges. All three programs, however, found correlational results consistent with the hypothesis that experienced or perceived social support is associated with successful cessation and/or maintenance. In several cases, degree of social support was prospectively related to later smoking status, suggesting that these results are not artifactual (e.g., unsuccessful subjects attributing their failure to lack of support). Thus, the importance of social support in smoking cessation is consistently confirmed in these three sets o f studies. These correlational data also suggest that it would be premature to abandon searching for therapeutic uses o f social support. The Oregon and North Dakota programs engaged key members o f the smokers' natural support system (spouses or co-workers) and attempted to train them in how to be more supportive or, especially in the North Dakota program, less likely to sabatoge. The data indicate that it may be difficult to change the behavior o f spouses or significant others within the constraints o f a smoking cessation program. Neither program showed significant treatment effects and there were at best only marginal effects on social support as assessed by the PIQ, a valid measure o f experienced
616
LICHTENSTEIN, GLASGOW, AND ABRAMS
helpfulness. The "costs" of involving significant others or spouses in cessation programs, at least in the ways attempted in these studies, clearly seem to outweigh the treatment gains. Social support exchanges between significant others may be stable or "trait-like" in nature and thus difficult to modify within time-limited smoking cessation programs. The structured treatment program (e.g., attendance at treatment sessions) may produce temporary increases in spouse supportive behaviors but the dyad then reverts to its customary mode of operation. It is possible that, in dyads which are naturally supportive of one another, smoking cessation is more likely to occur irrespective of the nature of treatment. The correlational data from Oregon and North Dakota are consistent with these conjectures: partner support training did not produce significant between group differences on the PIQ whereas PIQ scores were-across conditions--related to outcome. This is indirect evidence, however, because PIQ scores were only obtained after training. A valid pretreatment measure of partner support could be useful in demonstrating training effects and possibly in selecting couples more likely to accomplish behavior changes. By contrast, the Brown program focused primarily on the smokers themselves and attempted to train them in how to deal adaptively with both smoking and nonsmoking persons in their social networks. The Brown approach would have the advantage of being less costly or invasive because the active cooperation of a spouse or significant other was not required (cf. Abrams, Elder, Carlton, Lasater, & Artz, 1986). However, this approach was also unsuccessful. Further, there was both empirical and anecdotal evidence from the North Dakota and Brown programs to suggest that subjects found these support programs not to their liking, an important bit of consumer information. This dislike was reflected in the differences in credibility ratings obtained in one of the North Dakota studies and in anecdotal observations of the Brown subjects' resistance to following social support program suggestions. These observations suggest another preintervention assessment issue. A smoker's social network features and the perceived availibility of support may moderate his/her interest in, and need for, additional social support. Assessment of such individual differences could permit better matching of support interventions to smokers' needs. There may be other important individual differences in responsivity to or need for social support. Social support may primarily make a difference only for persons experiencing high levels of stress--the buffering hypothesis (Cohen & Wills, 1985). The analyses reported here do not consider interactions between social support and stress. The Oregon program has searched for a buffering effect and has found some evidence of it, but only at the end of treatment and only with a support measure other than the PIQ. More attention to buffering possibilities would be useful. There is also the possibility of sex differences; women may be more responsive to social support when quitting smoking. Data from recent studies by Pechacek et al. (1986) and Fisher and Bishop (1986) suggest
SOCIAL SUPPORT IN SMOKING
617
that, in smoking cessation programs, women are more responsive to social support manipulations then men. The Coppotelli and Orleans (1985) study of quitting among wives is also consistent with this notion although quitting husbands were not included for comparison. The present studies did not have sufficient numbers to investigate this possibility but it should be considered in future research. There are several additional issues in the application of social support to smoking cessation that merit consideration. More detailed analysis of the specific function that social support plays in smoking cessation would be useful. The functions (e.g., tangible assistance, emotional support) described by Cohen and Wills (1985) may be useful in this effort. Recent findings by Coppotelli and Orleans (1985) exemplify this approach. They found, for example, that, when husbands were perceived to minimize stress by avoiding interpersonal conflict and by taking over some of the quitter's usual chores and to be empathic and tolerant of moodiness, wives were more likely to achieve short-term cessation. The instrument used in that study appears to be useful for wife-perceived husband support; other instruments may be needed for other dyads or larger social networks. The timing of support--that is, when in the process of cessation and maintenance social support is most useful--is another issue to consider. Data from both Oregon (Mermelstein et al., 1983; Mermelstein, Cohen, Lichtenstein, Baer, & Kamarck, 1986) and Brown suggest that experienced support from significant others is important in short-term maintenance although the ratio of smokers to nonsmokers in one's social network is more important for long-term maintenance. The role of the smokers in one's environment suggests that extra-treatment factors should be considered in future research, possibly in much the same way as Moos and his colleagues (e.g., Cronkite & Moos, 1980; Moos & Finney, 1983) have approached the treatment of problem drinkers. In this expanded framework, posttreatment support--for example, family or marital functioning and life stressors--is seen to play a major role in maintenance or relapse (Moos & Finney, 1983). Smoking in the social network is difficult to modify and presents a constant source of smoking cues in social situations. One way to address the issue of reducing the number of smoking cues in the environment is to advocate stronger no-smoking policies and other environmental manipulations at the worksite and in public or civic settings. This suggestion argues for going beyond the traditional behavioral approaches to smoking cessation and for looking at the separate and interactive effects of organizational, community, and public policy interventions with behavioral treatment approaches (Abrams et al., 1986). Furthermore, social support treatments may interact synergistically with "systems level" interventions such as competitions between worksites or changes in a smoking policy. Attempts should be made to capitalize on such synergistic interaction. Finally, it may be that the small group treatment format employed by our three programs, or any counselor-guided program (Janis, 1983), pro-
618
LICHTENSTEIN, GLASGOW, AND ABRAMS
vides sufficient social support so as to obscure the contribution of additional components. If this conjecture is valid, social support should be more important for minimal interaction programs or in the context of unaided quitting. We realize this paper raises questions more than it provides answers. We are still persuaded that the social environment is a key factor in smoking cessation and maintenance, and we think that our experience will alert researchers and practitioners to the complexity of the issues so that they will not repeat our disappointing intervention findings. A more comprehensive survey of the role of social support in smoking cessation can be found in Cohen, Lichtenstein, Mermelstein, Mclntyre-Kingsolver, Baer, and Kamarck (in press). We hope to encourage other investigators to take different and more powerful approaches to social support that will take into account both individual (or dyadic) differences and larger social networks and systems.
REFERENCES Abrams, D., Elder, J., Carlton, R., Lasater, T., & Artz, L. (1986). Health and industry: A behavioral medium perspective. In M. Cataldo & T. Coates (Eds.), Behavioral medicine in industry. New York: Wiley. Abrams, D., & Habif, V. (1981). The social support for non-smoking scale. Unpublished manuscript, The Memorial Hospital/Brown University, Providence, RI. Abrams, D. B., Pinto, R. P., Monti, P. M., Jacobus, S., Brown, R., & Elder, J. P. (1985, March). Health education vs. cognitive stress management vs. social support training for relapse prevention in worksite smoking cessation. Paper presented at the annual convention of the Society for Behavioral Medicine, New Orleans, LA. Brown, R., Lichtenstein, E., McIntyre, K., & Harrington-Kostur, J. (1984). Effects of nicotine fading and relapse prevention in smoking cessation. Journal of Consulting and Clinical Psychology, 52, 307-308. Brownell, K., Heckerman, C., Westlake, R., Hayes, S., & Monti, P. (1978). The effect of couples training and partner cooperativeness in the behavioral treatment of obesity. Behaviour Research and Therapy, 16, 323-333. Chesney, M. A., & Feuerstein, M. (1979). Behavioral medicine in the occupational setting. In J. McNamara (Ed.), Behavioral approaches to medicine: Application and analysis, pp. 267-290. New York: Plenum. Cohen, S., Lichtenstein, E., Mermelstein, R. J., McIntyre-Kingsolver, K. O., Baer, J. S., & Kamarck, T . W . (in press). Social support and smoking cessation and maintenance. In B. H. Gottleib (Ed.), Creating support groups: Formats, processes, and effects. New York: Sage. Cohen, S., & Syme, S. L. (1985). Social support and health. Orlando, FL: Academic. Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering hypothesis. Psychological Bulletin, 98, 310-357. Cotletti, G.,&Brownell, K. (1982). The physical and emotional benefits ofsocial support: Application to obesity, smoking, and alcoholism. In M. Hersen, R. Eisler, & P. Miller (Eds.), Progress in Behavior Modification: Vol. 13 (pp. 110-178). New York: Academic. Coppotelli, H. C., & Orleans, C. T. (1985). Partner support and other determinants of smoking cessation maintenance among women. Journal of Consulting and Clinical Psychology, 53, 455-460.
SOCIAL SUPPORT IN SMOKING
619
Cronkite, R., & Moos, R. (1980). The determinants of posttreatment functioning of alcoholic patients: A conceptual framework. Journal of Consulting and Clinical Psychology, 48, 305-316. Fisher, E. B., Jr., & Bishop, D. (1986, March). Gender, social support, and smoking cessation. Paper presented at the meeting of the Society of Behavioral Medicine, San Francisco, CA. Foxx, R. M., & Brown, R. A. (1979). Nicotine fading and self-monitoring for cigarette abstinence or controlled smoking. Journal of Applied Behavior Analysis, 12, 111-125. Glasgow, R. E., Klesges, R. C., Godding, P. R., Vasey, M. W., & O'Neill, H. K. (1984). Evaluation of a worksite smoking control program. Journal of Consulting and Clinical Psychology, 52, 137-138. Glasgow, R. E., Klesges, R. C., & O'Neill, H. K. (in press). Programming social support for worksite smoking modification: An extension and replication. Addictive Behaviors. Janis, I. L. (1983). The role of social support in adherence to stressful decisions. American Psychologist, 38, 143-160. Krantz, D. S., Grunberg, N. E., & Baum, A. (1985). Health psychology. Annual Review of Psychology, 36, 349-383. Levy, R. L. (1983). Social support and compliance: A selective review and critique of treatment integrity and outcome measurement. Social Science and Medicine, 17, 13291338. Lichtenstein, E. (1982). The smoking problem: A behavioral perspective. Journal of Consulting and Clinical Psychology, 50, 804-819. Lichtenstein, E., Mermelstein, R. J., Kamarck, T. W., & Baer, J.S. (1986). Partner support and smoking cessation: Replication and extension. Manuscript in preparation. Malott, J. M., Glasgow, R. E., O'Neill, H. K., & Klesges, R. C. (1984). Coworker social support in a worksite smoking control program. Journal of Applied Behavior Analysis, 17, 485-495. McIntyre-Kingsolver, IC, Lichtenstein, E., & Mermelstein, R. J. (1986). Spouse training in a multicomponent smoking cessation program. Behavior Therapy, 17, 67-74. Mermelstein, R., Cohen, S., Lichtenstein, E., Baer, J. S., & Kamarck, T. (1986). Social support and smoking cessation and maintenance. Journal of Consulting and Clinical
Psychology. Mermelstein, R., Lichtenstein, E., & McIntyre, K. (1983). Partner support and relapse in smoking cessation programs. Journal of Consulting and Clinical Psychology, 51, 465466. Moos, R., & Finney, J. (1983). The expanding scope of alcoholism treatment evaluation. American Psychologist, 38, 1036-1044. Pechacek, T. F., Sillers, C., Glasgow, R. E., Murray, D., Blake, S. M., Mittelmark, M. B., Loken, B., & Luepker, R.V. (1986, August). Effect of social support on gender differences in smoking cessation. Paper to be presented at the annual meeting of the American Psychological Association as part of a symposium titled "Gender differences in smoking cessation" (B. Grim, Chair), Washington, D.C. Stachnik, T. J., & Stoffelmayr, B. E. (1981). Is there a future for smoking cessation programs? Journal of Community Health, 7, 47-56. Weisz, G., & Bucher, B. (1980). Involving husbands in treatment of obesity: Effects of weight loss, depression, and marital 'satisfaction. Behavior Therapy, 11, 643-650. RECEIVED:November 4, 1985 FINAL ACCEPTANCE:March 24, 1986
Social Support in Smoking Cessation: In Search of Effective Interventions EDWARD LICHTENSTEIN
University of Oregon and Oregon Research Institute RUSSELL E. GLASGOW
Oregon Research Institute DAVID B. ABRAMS
Brown University~The Miriam Hospital
The results of five recent smoking cessation studies from three separate research programs are summarized. Each study compared a basic cognitive-behavioral cessation program to the same program plus a component designed to enhance social support. Four of the studies found process or correlational data linking social support to outcomes. Nevertheless, there were no significant between-groups smoking-outcome differences in any of the five studies. Several explanations for these findings are considered and it is concluded that social support deserves further consideration in interventions for smoking reduction. Suggestions for future research are offered.
T h e r e is s u b s t a n t i a l e v i d e n c e suggesting t h a t p e r s o n s w i t h r e l a t i v e l y h i g h l e v e l s o f social s u p p o r t r e p o r t f e w e r p s y c h o l o g i c a l a n d p h y s i c a l s y m p t o m s . O n e w a y t h a t s o c i a l s u p p o r t m a y i n f l u e n c e p h y s i c a l h e a l t h is t h r o u g h c h a n g e s in p e r f o r m a n c e o f h e a l t h - r e l a t e d b e h a v i o r s s u c h as d e c r e a s e d c i g a r e t t e a n d a l c o h o l use o r i m p r o v e d d i e t a n d e x e r c i s e r e g i m e n s ( C o h e n & S y m e , 1985; K r a n t z , G r u n b e r g , & B a u m , 1985). B o t h t h e o r e t i c a l p r o p o s i t i o n s a b o u t t h e f a c i l i t a t i v e r o l e o f s u p p o r t in b e h a v i o r c h a n g e (e.g.,
Preparation of this report was supported by the National Heart, Lung, and Blood Institute grants, grant numbers HL29547, HL33739, and HL32318, awarded to Edward Lichtenstein, Russell E. Glasgow, and David B. Abrams respectively. Each project is a collaborative endeavor and we appreciate the assistance and social support of our colleagues. Requests for reprints should be sent to Edward Lichtenstein, Department of Psychology, University of Oregon, Eugene, Oregon 97403.
607 0005-7894/86/0607-061951.00/0 Copyright1986by Associationfor Advancementof BehaviorTherapy All rightsof reproductionin any formreserved.
608
LICHTENSTEIN, GLASGOW, AND ABRAMS
Janis, 1983) and empirical observations relating social support variables to performance of health-related behaviors (Levy, 1983) are congruent with this notion. The importance of social support is also suggested by its incorporation in various intervention programs for addictive behaviors. Programs such as Alcoholics Anonymous, Weight Watchers, and The Five Day Plan for Smoking Cessation all embody social support components such as buddy systems and group support as central features of their approach. Given that alcoholism, obesity, and smoking have been slow to yield to standard behavioral (or other) treatment strategies and given the intuitive, theoretical, and empirical plausibility of social support, it is not surprising that investigators have recently attempted to develop social support intervention strategies (Colletti & Brownell, 1982). These programs have usually attempted to combine or integrate a social support intervention with standard cognitive-behavioral strategies. This approach is most fully developed in the area of obesity where the promising work of Brownell and colleagues (Brownell, Heckerman, Westlake, Hayes, & Monti, 1978) stimulated a number of studies. The results of more recent social support interventions for obesity, however, have been equivocal (e.g., Weisz & Bucher, 1980). Given this theoretical and empirical background, it has been suggested that it would be fruitful to apply social support interventions to smoking cessation (Lichtenstein, 1982). Social support from the social environment could be a source of motivation to expend the effort to try and quit smoking and a source of reinforcement for successful accomplishments. Social support might also provide a buffer against the stress of quitting (e.g., withdrawal) or other stressful events that might precipitate a relapse (Cohen & Wills, 1985). Our three research groups have been independently pursuing this line of investigation over the past several years. We have utilized operationally distinct but conceptually similar social support treatment strategies in the context of smoking modification programs in both community-based clinics and at the worksite. While sharing our experiences and results informally, we were struck by the similarity in our findings. We believed that summarizing the intervention strategies we developed and the results we have obtained in a single paper would be useful to both workers in smoking cessation and to students of social support. To telegraph our basic message, our three programs have found that measures of social support are positively correlated with better treatment outcomes but that attempts to both increase social support and to enhance treatment effectiveness have not been successful. The fact that each of our programs has replicated its treatment outcome findings, together with the convergence of the findings across the three projects, seems to us to be noteworthy. We will describe and summarize the major findings from each of the projects and then offer some speculations about the implications of these data for future research.
SOCIAL SUPPORT IN SMOKING
609
The Oregon Program Lichtenstein and his colleagues have focused on socially supportive behaviors from a spouse or living partner of the person trying to quit smoking. The Partner Interaction Questionnaire (PIQ) consists of 61 behaviors related to smoking cessation, both positive (e.g., my partner complimented my not smoking) and negative (e.g., my partner commented on my lack of willpower). Subjects first judge the frequency of occurrence of each item and then evaluate how helpful each item was in their effort to stop smoking. A summary measure of partner support, experienced helpfulness, is derived by summing the cross products of the frequency and helpfulness scores. Mermelstein, Lichtenstein, and McIntyre (1983) showed that experienced helpfulness scores obtained at the end of treatment predicted smoking status up to six months posttreatment. These correlational data, plus some of the promising early findings in the application of spouse support to the treatment of obesity (Brownell et al., 1978), led us to derive a partner support program for smokers (McIntyre-Kingsolver, Lichtenstein, & Mermelstein, 1986). Smokers (N = 64) with spouses who were willing to cooperate were randomly assigned to either a multicomponent cognitive-behavioral smoking program (six 2-hour sessions plus intake) or the same program with spouses attending all sessions and receiving training and encouragement. The basic program (Brown, Lichtenstein, McIntyre, & Harrington-Kostur, 1984) includes nicotine fading, self-management training, and relapse prevention. Follow-up data were obtained l, 2, 3, 6, and 12 months posttreatment and self-reported smoking status was corroborated by informants and carbon monoxide measures (at posttreatment and six month follow-up only). PIQ data were obtained at the end of treatment to assess the impact of training. During treatment sessions couples were given feedback about helpful and unhelpful spouse behaviors related to smoking cessation. Common areas of helpful or unhelpful behaviors were pointed out and group members were encouraged to contribute examples from their own experience. Each couple was encouraged to problem solve regarding what kinds of spouse behaviors would be most helpful for the smoker in his/her cessation efforts. Guided group discussions and suggested homework exercises were used to encourage couple problem solving. Spouses were encouraged to reward their partners and to participate in program activities as suggested by both the program and their partner. There was a marginally significant difference in cessation rates favoring the spouse support condition at the end of treatment (72.7% versus 48.4%; p < . 10). Although follow-up abstinence rates for the partner support group were consistently in the expected direction, the differences were neither statistically significant nor clinically meaningful (see Table 1). Posttreatment PIQ scores tended to be higher for the subjects receiving spouse support training, but the difference did not reach significance (spouse training M = 22.3, SD = 18.2; control M = 15.3, SD = 12.6; F = 3.05,
610
LICHTENSTEIN, GLASGOW, AND ABRAMS TABLE 1 DESIGN AND OLr'FCOME OF SOCIAL SUPPORT INTERVENTIONS Results: cessation rates
Study
Sample characteristics
Design
Post-Rx
Follow-up
MclntyreKingsolver et al. (1986)
64 Ss with cooperat i r e spouses, 57% female, M a g e = 38; M cigs/day = 25.6
Basic p r o g r a m (n = 31) Basic p r o g r a m + spouse support (n = 33)
48.4%
32.3% (1 yr)
72.7
36.3
Lichtenstein, et al. (1985)
64 Ss 50% female, M age = 39; M cigs/ d a y = 25.8
Basic p r o g r a m (n = 21) Basic p r o g r a m + spouse s u p p o r t (n = 15) Single S s (n = 28)
57.1
23.8 (6 mo)
66.7
33.3
53.6
35.7
24 medical clinic & telephone Co. e m ployees, 83% female, M age = 34; M cigs/day = 24
Basic p r o g r a m (n = 12) Basic p r o g r a m + coworker support (n = 12)
17.0
27.0 (6 m o )
17.0
17.0
Glasgow et al. (in press)
29 hospital svcs & savings & loan e m ployees, 69% female, M age = 34; M cigs/clay = 25
Basic p r o g r a m Basic p r o g r a m + sig. o t h e r support
54.0 36.0
25.0 (6 m o ) 23.0
A b r a m s et al. (1985)
90 m f g & health insurance employees, 63% female, M age = 41.3; M cigs/ d a y = 28.1
Basic p r o g r a m + health educ.
63.0
23.3 (6 m o )
47.0
23.3
43.0
20.0
Malott et al. (1984)
(n =
30)
Basic p r o g r a m + cognitive trng (n =
30)
Basic p r o g r a m + s u p p o r t trng (n = 30)
df =
1, p < . 10). These results suggested that our training procedures did not have a strong impact on spouse helpfulness, at least as measured by the PIQ. Across all subjects, however, experienced helpfulness was significantly related to both cessation concurrently at the end o f treatment, M abstainers = 22.4, M smokers = 13.5; t(62) = 2.2, p < .05, and prospectively to maintenance o f cessation up to 3 - m o n t h follow-up. Subjects who initially quit but relapsed by three m o n t h s had significantly lower PIQ scores than subjects who remained abstinent, t(38) = 2.33, p < .05. These findings partially replicate the correlational results o f Mermelstein et al. (1983) and confirm the association between experienced helpfulness from a partner and the o u t c o m e o f smoking treatment.
SOCIAL SUPPORT IN SMOKING
611
We conceived several reasons for the weak results of this study. The spouse training components of the program may not have been intensive or salient enough. Both members of several couples in this first study were trying to quit smoking and therefore served in a dual role as both helper and helpee. Subanalyses indicated that such couples tended (but not significantly so) to do less well when only one member of the pair was trying to quit smoking (4 of 14 versus 9 of 19 abstinent at one year). Finally, we felt that the inclusion of a third group of subjects, either without spouses or with unwilling partners, would be informative. Accordingly, a second study, also with 64 subjects, was conducted which remedied these deficiencies (Lichtenstein, Mermelstein, Kamarck, & Baer, in preparation). A manual for the helping spouse was developed and more program time was devoted to spouse training. Only 3 of the 33 couple subjects recruited contained two smokers trying to quit. And finally, a group of subjects without spouses was also included. Because they could not be randomly assigned as were subjects with available spouses, this third group was a quasi-control or comparison group. The basic program and follow-up procedures were the same as in the first study. The results were also quite similar (see Table 1). Again, subjects in the spouse training condition tended to have higher abstinence rates at all assessment points but these differences were small and nowhere approached significance. Interestingly, subjects without spouses tended to have the highest abstinence rates of all, but again the differences were not significant. Posttreatment PIQ scores yielded a similar pattern of results; mean differences were clearly in favor of the spouse training subjects, but again did not reach statistical significance. Unlike our previous two studies, PIQ scores were not related to cessation or maintenance.
The North Dakota Program Glasgow, Klesges, and colleagues at North Dakota State University have investigated both co-worker and significant other social support in the context of their worksite smoking control program. A six-session, multicomponent, cognitive-behavioral program offering subjects the options of abstinence or controlled smoking has been developed and adapted for occupational settings (Glasgow, Klesges, Godding, Vasey, & O'Neill, 1984). The program involves weekly group meetings of 4-8 employees focused on making sequential reductions in nicotine content of brand smoked and number of cigarettes per day prior to a decision to attempt cessation (the preferred goal) or to make reductions in the percent of the cigarette smoked midway through the program. The last two sessions are devoted to solidifying behavior changes and to relapse prevention. As one of the major advantages of conducting smoking modification programs in occupational settings is purported to be the existence of peer and environmental support for behavior change (Chesney & Feuerstein, 1979; Stachnik & Stoffelmayr, 1981), we have evaluated the incremental effects of adding social support programming to the basic treatment. A
612
LICHTENSTEIN, GLASGOW, AND ABRAMS
38-item modification of the PIQ, revised to be applicable to controlled smoking and scored to produce both positive and negative interaction subscales, has been used to assess degree o f perceived social support. Six month follow-up data are obtained and carbon monoxide (CO) assessments are used to verify self-reports of smoking status. An initial study by Malott, Glasgow, O'Neill, and Klesges (1984) evaluated the effects of adding co-worker support procedures to the basic program. These procedures involved subject selection o f partners, use of a buddy system, a 17-page partner-support manual, and individualization o f support strategies. There were no differences between conditions at posttest or follow-up on either percent o f subjects abstinent (see Table 1) or on CO reductions among nonabstinent subjects. There were also no between groups differences on the modified PIQ measure. Negative, but not positive, support scores from the PIQ were related to treatment outcome. One unexpected finding to emerge from this study was that a treatment credibility questionnaire completed after the first treatment session revealed that the basic treatment plus social support condition was perceived as significantly less credible than the basic treatment program alone, F(1, 21) = 16.1, p < .001. There were at least three plausible explanations for the lack of betweengroups differences in this study. The basic treatment condition, which involved weekly small group meetings of co-workers who frequently interacted with each other and worked in relatively cohesive organizations, may have itself generated high levels of co-worker social support. Second, the additional partner support assignments increased the complexity of the program and may have interfered with subjects' efforts to change their own behavior. Finally, extrapolating from the correlational findings regarding the importance of negative social interactions related to smoking behavior, we may have placed too much emphasis on increasing positive co-worker interactions and not enough on decreasing negative smokingrelated interchanges. To address these issues, we conducted a second study (Glasgow, Klesges, & O'Neill, in press), this time employing significant others as partners. It was reasoned that the use of subject-identified significant others (usually spouses) would address the first two concerns because spouse support would not be expected to be at maximal levels and subjects did not have the dual role of both changing their own behavior and of being a "partner." To address the third concern we placed greater emphasis on decreasing well-intentioned but detrimental social interactions (e.g., nagging about smoking in the house). Twenty-nine subjects were randomly assigned to our basic treatment condition or to basic treatment plus social support. Significant other support procedures involved two group meetings of partners, semi-weekly mailings of sections of a revised partner support manual, and phone calls from therapists on alternating weeks to discuss progress. As in the first study, partners were provided with a list o f support behaviors that their " s m oker " identified as most helpful. In addition, individualized suggestions, drawn from the modified PIQ, were provided
SOCIAL SUPPORT IN SMOKING
613
as to which specific partner behaviors would be perceived as unhelpful. The importance of partners seeking feedback from subjects was also emphasized. Results were quite similar to those in the first study (see Table 1). Again, there were no differences between conditions at either posttest or followup on abstinence rates or on CO reductions among nonabstinent subjects. In fact, subjects in the basic treatment condition experienced nonsignificantly higher abstinence rates and greater CO reductions than subjects in the social support condition. There were again no between groups differences on the modified PIQ, but analyses of individual PIQ items targeted for intervention suggested that partners did engage in desired support behaviors. Despite the lack of between-groups differences, this investigation also revealed consistent correlational findings suggesting a relationship between social support, as measured by the modified PIQ, and treatment outcome. Specifically, the presence of "'positive" social support was not related to treatment outcome, but at posttest the absence of negative smoking related interactions (e.g., "expressed doubt about your ability to quit") was significantly correlated with concurrent self-report and biochemical measures of more successful treatment outcome (rs = .25 to .49 in Study 1 and .32 to .63 in Study 2).
The Brown Program Abrams, Pinto, Monti, and colleagues (1985) investigated three relapse prevention approaches in randomized trials across three worksite settings. Within each worksite, subjects (ns = 27 to 36) were randomly assigned to (a) cognitive behavioral management, (b) social support/social skills training, and (c) health education and nonspecific support (total N = 90). Groups of between 9 and 12 smokers met at the worksite with a malefemale co-therapist team for 11/2hours weekly for 8 weeks. The first four weeks consisted primarily of a standard nicotine fading program for smoking cessation using the Foxx and Brown (1979) procedure. Subjects set a quit day at Week 5. From Week 3 onwards the relapse prevention training was gradually introduced, with the last 4 sessions (Weeks 5-8) exclusively devoted to one of the three conditions described below. Cognitive behavioral management. This condition focused on intrapersonal coping only. All high-risk-for-relapse situations ultimately were identified as leading to intrapersonal reactions (stress, negative moods, withdrawal, urges to smoke, fatigue) and subjects were trained to use intrapersonal coping skills. The program used relaxation training, cognitive restructuring (thought-stopping and substitution of positive selftalk), and training in coping with "slips" (Abstinence Violation Effects). Within each session a skills acquisition model was used including instructions, modeling, practice, feedback, and homework. Social skills/support treatment. This relapse prevention program focused exclusively on social skills and social support network interventions (buddy systems at the worksite and at home) to prevent relapse. All high
614
LICHTENSTEIN, GLASGOW, AND ABRAMS
risk situations were identified as either originating in the social environm e n t or as requiring social support for coping. Participants were first asked to identify their " p r o x i m a l social networks," that is, the five individuals that they saw and confided in the most frequently, and to classify each person as potentiaily supportive, neutral, or unsupportive. Selection o f at least one supportive individual who had already successfully quit smoking was r e c o m m e n d e d . Participants were instructed how to deal effectively with persons who might act as smoking triggers and how to initiate an assertive request for someone else not to smoke. Conversational skills, the seeking out o f support when stressed, and handling praise or criticism from friends and relatives was practiced. Finally, subjects were trained to encourage neutral individuals to provide more active nonsmoking support (e.g., arranging exercise with friends). Once again, an intensive behavioral skills acquisition program was used, including instructions, modeling, feedback, role-playing, and homework. Health education and nonspecific support. Subjects in this condition received no behavioral skills training for relapse prevention. Instead, they were given information about the health consequences o f smoking and especially facts about "withdrawal" effects. A series o f nonspecific group discussions were held to aid the participants' efforts to quit smoking and to equate for contact time with the other two conditions. Participants sometimes discussed high-risk-for-relapse situations, but therapists provided no coping skills training per se. Overall results (see Table 1) revealed no statistically significant differences across the three treatment conditions or the three different worksite replications at posttest or follow-up (Abrams et al., 1985). Abstinence at follow-up was verified by CO in all cases. Thus, neither the social support nor the cognitive behavioral conditions were superior to the discussioncontrol condition at follow-up. Despite the negative overall results o f the social support program, correlational data based on pre- and posttreatment measures point to the importance o f social support variables in treatment. Based on a recent item-and-factor analysis o f the Partner Interaction Questionnaire (PIQ), a revised version o f the scales (consisting o f a 10-item " s u p p o r t i v e " subscale and a 10-item " u n s u p p o r t i v e " subscale) was administered pre- and posttreatment (Kamarck & Lichtenstein, personal c o m m u n i c a t i o n , Nov e m b e r 14, 1984). At pretreatment the supportive subscale o f the revised PIQ was significantly related to 6 - m o n t h follow-up smoking rate, r = - . 2 1 , p < .05). The ratio o f smoking friends to total friends in the social network o f the participants was also positively correlated with smoking rate at 6 - m o n t h follow-up, r = .33, p < .01). Analyses o f posttreatment measures o f social support revealed two subject characteristics correlated with 6 - m o n t h follow-up: a generalized social support for nonsmoking scale (Abrams & Habif, 1981; r = - . 4 6 , p < .01), and the PIQ unsupportive subscale, r = .24, p < .05. The PIQ results replicate previous research by Mermelstein et al. (1983) and the N o r t h Dakota program reported above. It was o f clinical interest to note that m a n y consumers
SOCIAL SUPPORT IN SMOKING
615
did not find aspects o f the social support training program to be useful; for example, they resisted suggestions such as the need to assertively request that other smokers not smoke in their presence.
DISCUSSION The results across the three different research programs are remarkably consistent. In five separate studies across the three projects the addition o f a social support component to a standard behaviorally-based cessation program in no case resulted in significant improvements in treatment outcome. Only in one of the two Oregon studies and only at the end of treatment was there an encouraging trend in favor o f the social support condition. The Oregon data, taken alone, could be viewed as encouraging evidence for partner support but they were limited by insufficient power to detect a small but real effect. The North Dakota and Brown studies also have limited statistical power but the patterns o f means and the consistency of the results across the three programs make it unlikely that we are committing a Type II error with respect to social support as conceptualized in these studies. Although there was some variation in support enhancement procedures, we recognize that only a limited range o f theoretical and operational social support options have been tested. It must also be acknowledged that the largely negative outcome data in these studies could be owing to the irrelevance o f support, as conceptualized in these studies, for smoking cessation o r to the inadequacy of our social support training. The basic hypothesis that increased support would facilitate cessation and maintenance may not have been tested adequately owing to our failure to substantially increase social support in the experimental groups. The PIQ data from the Oregon and North Dakota studies are consistent with this notion. The failure to increase social support may be related, as we discuss below, to the stability or "'traitlike" nature o f social skills or to dyadic support exchanges. All three programs, however, found correlational results consistent with the hypothesis that experienced or perceived social support is associated with successful cessation and/or maintenance. In several cases, degree of social support was prospectively related to later smoking status, suggesting that these results are not artifactual (e.g., unsuccessful subjects attributing their failure to lack of support). Thus, the importance of social support in smoking cessation is consistently confirmed in these three sets o f studies. These correlational data also suggest that it would be premature to abandon searching for therapeutic uses o f social support. The Oregon and North Dakota programs engaged key members o f the smokers' natural support system (spouses or co-workers) and attempted to train them in how to be more supportive or, especially in the North Dakota program, less likely to sabatoge. The data indicate that it may be difficult to change the behavior o f spouses or significant others within the constraints o f a smoking cessation program. Neither program showed significant treatment effects and there were at best only marginal effects on social support as assessed by the PIQ, a valid measure o f experienced
616
LICHTENSTEIN, GLASGOW, AND ABRAMS
helpfulness. The "costs" of involving significant others or spouses in cessation programs, at least in the ways attempted in these studies, clearly seem to outweigh the treatment gains. Social support exchanges between significant others may be stable or "trait-like" in nature and thus difficult to modify within time-limited smoking cessation programs. The structured treatment program (e.g., attendance at treatment sessions) may produce temporary increases in spouse supportive behaviors but the dyad then reverts to its customary mode of operation. It is possible that, in dyads which are naturally supportive of one another, smoking cessation is more likely to occur irrespective of the nature of treatment. The correlational data from Oregon and North Dakota are consistent with these conjectures: partner support training did not produce significant between group differences on the PIQ whereas PIQ scores were-across conditions--related to outcome. This is indirect evidence, however, because PIQ scores were only obtained after training. A valid pretreatment measure of partner support could be useful in demonstrating training effects and possibly in selecting couples more likely to accomplish behavior changes. By contrast, the Brown program focused primarily on the smokers themselves and attempted to train them in how to deal adaptively with both smoking and nonsmoking persons in their social networks. The Brown approach would have the advantage of being less costly or invasive because the active cooperation of a spouse or significant other was not required (cf. Abrams, Elder, Carlton, Lasater, & Artz, 1986). However, this approach was also unsuccessful. Further, there was both empirical and anecdotal evidence from the North Dakota and Brown programs to suggest that subjects found these support programs not to their liking, an important bit of consumer information. This dislike was reflected in the differences in credibility ratings obtained in one of the North Dakota studies and in anecdotal observations of the Brown subjects' resistance to following social support program suggestions. These observations suggest another preintervention assessment issue. A smoker's social network features and the perceived availibility of support may moderate his/her interest in, and need for, additional social support. Assessment of such individual differences could permit better matching of support interventions to smokers' needs. There may be other important individual differences in responsivity to or need for social support. Social support may primarily make a difference only for persons experiencing high levels of stress--the buffering hypothesis (Cohen & Wills, 1985). The analyses reported here do not consider interactions between social support and stress. The Oregon program has searched for a buffering effect and has found some evidence of it, but only at the end of treatment and only with a support measure other than the PIQ. More attention to buffering possibilities would be useful. There is also the possibility of sex differences; women may be more responsive to social support when quitting smoking. Data from recent studies by Pechacek et al. (1986) and Fisher and Bishop (1986) suggest
SOCIAL SUPPORT IN SMOKING
617
that, in smoking cessation programs, women are more responsive to social support manipulations then men. The Coppotelli and Orleans (1985) study of quitting among wives is also consistent with this notion although quitting husbands were not included for comparison. The present studies did not have sufficient numbers to investigate this possibility but it should be considered in future research. There are several additional issues in the application of social support to smoking cessation that merit consideration. More detailed analysis of the specific function that social support plays in smoking cessation would be useful. The functions (e.g., tangible assistance, emotional support) described by Cohen and Wills (1985) may be useful in this effort. Recent findings by Coppotelli and Orleans (1985) exemplify this approach. They found, for example, that, when husbands were perceived to minimize stress by avoiding interpersonal conflict and by taking over some of the quitter's usual chores and to be empathic and tolerant of moodiness, wives were more likely to achieve short-term cessation. The instrument used in that study appears to be useful for wife-perceived husband support; other instruments may be needed for other dyads or larger social networks. The timing of support--that is, when in the process of cessation and maintenance social support is most useful--is another issue to consider. Data from both Oregon (Mermelstein et al., 1983; Mermelstein, Cohen, Lichtenstein, Baer, & Kamarck, 1986) and Brown suggest that experienced support from significant others is important in short-term maintenance although the ratio of smokers to nonsmokers in one's social network is more important for long-term maintenance. The role of the smokers in one's environment suggests that extra-treatment factors should be considered in future research, possibly in much the same way as Moos and his colleagues (e.g., Cronkite & Moos, 1980; Moos & Finney, 1983) have approached the treatment of problem drinkers. In this expanded framework, posttreatment support--for example, family or marital functioning and life stressors--is seen to play a major role in maintenance or relapse (Moos & Finney, 1983). Smoking in the social network is difficult to modify and presents a constant source of smoking cues in social situations. One way to address the issue of reducing the number of smoking cues in the environment is to advocate stronger no-smoking policies and other environmental manipulations at the worksite and in public or civic settings. This suggestion argues for going beyond the traditional behavioral approaches to smoking cessation and for looking at the separate and interactive effects of organizational, community, and public policy interventions with behavioral treatment approaches (Abrams et al., 1986). Furthermore, social support treatments may interact synergistically with "systems level" interventions such as competitions between worksites or changes in a smoking policy. Attempts should be made to capitalize on such synergistic interaction. Finally, it may be that the small group treatment format employed by our three programs, or any counselor-guided program (Janis, 1983), pro-
618
LICHTENSTEIN, GLASGOW, AND ABRAMS
vides sufficient social support so as to obscure the contribution of additional components. If this conjecture is valid, social support should be more important for minimal interaction programs or in the context of unaided quitting. We realize this paper raises questions more than it provides answers. We are still persuaded that the social environment is a key factor in smoking cessation and maintenance, and we think that our experience will alert researchers and practitioners to the complexity of the issues so that they will not repeat our disappointing intervention findings. A more comprehensive survey of the role of social support in smoking cessation can be found in Cohen, Lichtenstein, Mermelstein, Mclntyre-Kingsolver, Baer, and Kamarck (in press). We hope to encourage other investigators to take different and more powerful approaches to social support that will take into account both individual (or dyadic) differences and larger social networks and systems.
REFERENCES Abrams, D., Elder, J., Carlton, R., Lasater, T., & Artz, L. (1986). Health and industry: A behavioral medium perspective. In M. Cataldo & T. Coates (Eds.), Behavioral medicine in industry. New York: Wiley. Abrams, D., & Habif, V. (1981). The social support for non-smoking scale. Unpublished manuscript, The Memorial Hospital/Brown University, Providence, RI. Abrams, D. B., Pinto, R. P., Monti, P. M., Jacobus, S., Brown, R., & Elder, J. P. (1985, March). Health education vs. cognitive stress management vs. social support training for relapse prevention in worksite smoking cessation. Paper presented at the annual convention of the Society for Behavioral Medicine, New Orleans, LA. Brown, R., Lichtenstein, E., McIntyre, K., & Harrington-Kostur, J. (1984). Effects of nicotine fading and relapse prevention in smoking cessation. Journal of Consulting and Clinical Psychology, 52, 307-308. Brownell, K., Heckerman, C., Westlake, R., Hayes, S., & Monti, P. (1978). The effect of couples training and partner cooperativeness in the behavioral treatment of obesity. Behaviour Research and Therapy, 16, 323-333. Chesney, M. A., & Feuerstein, M. (1979). Behavioral medicine in the occupational setting. In J. McNamara (Ed.), Behavioral approaches to medicine: Application and analysis, pp. 267-290. New York: Plenum. Cohen, S., Lichtenstein, E., Mermelstein, R. J., McIntyre-Kingsolver, K. O., Baer, J. S., & Kamarck, T . W . (in press). Social support and smoking cessation and maintenance. In B. H. Gottleib (Ed.), Creating support groups: Formats, processes, and effects. New York: Sage. Cohen, S., & Syme, S. L. (1985). Social support and health. Orlando, FL: Academic. Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering hypothesis. Psychological Bulletin, 98, 310-357. Cotletti, G.,&Brownell, K. (1982). The physical and emotional benefits ofsocial support: Application to obesity, smoking, and alcoholism. In M. Hersen, R. Eisler, & P. Miller (Eds.), Progress in Behavior Modification: Vol. 13 (pp. 110-178). New York: Academic. Coppotelli, H. C., & Orleans, C. T. (1985). Partner support and other determinants of smoking cessation maintenance among women. Journal of Consulting and Clinical Psychology, 53, 455-460.
SOCIAL SUPPORT IN SMOKING
619
Cronkite, R., & Moos, R. (1980). The determinants of posttreatment functioning of alcoholic patients: A conceptual framework. Journal of Consulting and Clinical Psychology, 48, 305-316. Fisher, E. B., Jr., & Bishop, D. (1986, March). Gender, social support, and smoking cessation. Paper presented at the meeting of the Society of Behavioral Medicine, San Francisco, CA. Foxx, R. M., & Brown, R. A. (1979). Nicotine fading and self-monitoring for cigarette abstinence or controlled smoking. Journal of Applied Behavior Analysis, 12, 111-125. Glasgow, R. E., Klesges, R. C., Godding, P. R., Vasey, M. W., & O'Neill, H. K. (1984). Evaluation of a worksite smoking control program. Journal of Consulting and Clinical Psychology, 52, 137-138. Glasgow, R. E., Klesges, R. C., & O'Neill, H. K. (in press). Programming social support for worksite smoking modification: An extension and replication. Addictive Behaviors. Janis, I. L. (1983). The role of social support in adherence to stressful decisions. American Psychologist, 38, 143-160. Krantz, D. S., Grunberg, N. E., & Baum, A. (1985). Health psychology. Annual Review of Psychology, 36, 349-383. Levy, R. L. (1983). Social support and compliance: A selective review and critique of treatment integrity and outcome measurement. Social Science and Medicine, 17, 13291338. Lichtenstein, E. (1982). The smoking problem: A behavioral perspective. Journal of Consulting and Clinical Psychology, 50, 804-819. Lichtenstein, E., Mermelstein, R. J., Kamarck, T. W., & Baer, J.S. (1986). Partner support and smoking cessation: Replication and extension. Manuscript in preparation. Malott, J. M., Glasgow, R. E., O'Neill, H. K., & Klesges, R. C. (1984). Coworker social support in a worksite smoking control program. Journal of Applied Behavior Analysis, 17, 485-495. McIntyre-Kingsolver, IC, Lichtenstein, E., & Mermelstein, R. J. (1986). Spouse training in a multicomponent smoking cessation program. Behavior Therapy, 17, 67-74. Mermelstein, R., Cohen, S., Lichtenstein, E., Baer, J. S., & Kamarck, T. (1986). Social support and smoking cessation and maintenance. Journal of Consulting and Clinical
Psychology. Mermelstein, R., Lichtenstein, E., & McIntyre, K. (1983). Partner support and relapse in smoking cessation programs. Journal of Consulting and Clinical Psychology, 51, 465466. Moos, R., & Finney, J. (1983). The expanding scope of alcoholism treatment evaluation. American Psychologist, 38, 1036-1044. Pechacek, T. F., Sillers, C., Glasgow, R. E., Murray, D., Blake, S. M., Mittelmark, M. B., Loken, B., & Luepker, R.V. (1986, August). Effect of social support on gender differences in smoking cessation. Paper to be presented at the annual meeting of the American Psychological Association as part of a symposium titled "Gender differences in smoking cessation" (B. Grim, Chair), Washington, D.C. Stachnik, T. J., & Stoffelmayr, B. E. (1981). Is there a future for smoking cessation programs? Journal of Community Health, 7, 47-56. Weisz, G., & Bucher, B. (1980). Involving husbands in treatment of obesity: Effects of weight loss, depression, and marital 'satisfaction. Behavior Therapy, 11, 643-650. RECEIVED:November 4, 1985 FINAL ACCEPTANCE:March 24, 1986