Tobacco Documents Online

I I I I I I I I I I I I I I 1 I I I Patterns and Predictors of Smoking Cessation Among tlsers of a Telephone Hotline CARLOS ROBERTO JAEN, MD, PhD K. MICHAEL CUMMINGS, PhD, MPH MARIA ZIELEZNY, PhD ROBERT O'SHEA, PhD Dr. Jabn was a doctoral student in epidemiology when this research was conducted. He is now Assistant Professor of Family Medicine at the School of Medicine and Biomedical Sciences, State University of New York at Buffalo (SUNYAB). Dr. Cumminoe is Director of Smoking Control at Roswell Park Cancer Institute in Buffalo. Doctors Zielezny and O'Shea are both Associate Professors of Social and Preventive Medicine at SUNYAB. This research was supported in part by grant No. CA36265 from the National Cancer Institute, Public Health Service. Tearsheet requests to Carlos Roberto JaEn, MD, PhD, De- partment of Family Medicine, School of Medicine and Biomedi- cal Sciences, State University of New York at Buffalo, 462 Grider St., Buffalo, NY 14215; tel. 716-898-4743; fax: 716-898-4750. The authors report results of a prospective co- hort study of 1,552 smokers who called a stop smoking hotline to request self-help smoking cessa- tion information. The participants were classifted into three groups based on reports at the 6-month followup: 242 quitters, 497 recidivists, and 813 nonquitters. Baseline and followup data were used to evaluate three comparisons: quitters nonquitters, quitters versus recidivists, and vists versus nonquitters. versus recidi- Nonquitters appear to be less motivated and more doubtful of their abilities to quit successfully compared with the other two groups. Quitters appear to live in a supportive environment for smoking cessation. Heavier smokers are more hesi- tant to try to quit, but once they make an attempt they are as likely to succeed as lighter smokers, when other factors are kept constant. Synopsis .................................... Most former cigarette smokers in the United States have stopped without formal assistance. However, a large proportion of smokers desire and seek help other than by attending formal programs. It is important to recognize what factors are likely to influence the effectiveness of smoking cessation attempts among these persons. EVERY YEAR MILLIONS OF AMERICANS attempt to stop smoking cigarettes, but only a fraction are successful in maintaining their newly acquired non- smoking status (1,2). What distinguishes those who are successful from those who try and fail or those who do not try? Answers to this question are likely to help all those interested in improving the effec- tiveness of smoking cessation interventions. Most of what we know about stopping smoking is derived from research on subjects attending formal treatment programs, despite the fact that 92 per- cent of ex-smokers have quit on their own (3). Among those who quit on their own, there is a group of smokers who ask for help without neces- sarily attending formal cessation programs. Little is known about this group of help-seekers. It has rn .,r.a a.aa, A.oau Efforts to promote environments supportive of smoking cessation are likely to result in a larger number of successful quitters. Similarly, efforts to strengthen motivation and belief in personal ability to quit are likely to encourage more nonquitters to attempt to stop smoking. Finally, it appears that some smokers need a previous quit attempt before they are able to maintain cessation successfully. been found that the majority of smokers who are motivated to stop smoking are less interested in formal programs than in do-it-yourself methods (3-5). Thus, there is a need to investigate smoking cessation among those outside formal treatment programs. One approach to studying smoking cessation is to compare successful quitters, recidivists, and nonstoppers in terms of factors that are potentially associated with smoking cessation. Ockene and colleagues made this comparison in a sample of 169 smokers enrolled in the Multiple Risk Factor Inter- vention Trial (MRFIT) (6). They evaluated stress, personal security, belief in personal control, social support, demographic variables, and smoking ,rate as potential factors. Successful quitters had higher 2046399792 I

I I I I I I I I I I I I I I I I I I initial expectations of success, had easier prior cessation attempts, and had a higher degree of personal-security than recidivists and nonstoppers (6). The-middle-age, male smokers involved in MRFIT are not directly comparable to the general population of smokers. Thus, research that would compare successful quitters, recidivists, and non- stoppers in a larger and more heterogeneous popu- lation of smokers is needed. The purpose of this study was to identify demo- graphic, attitudinal, socioenvironmental, health sta- tus, smoking history, and use-of-cessation-help fac- tors that distinguish between successful quitters, recidivists, and nonstoppers in a population of smokers who asked for help during their smoking cessation attempt. Methods A prospective cohort study design was used. The 1,895 subjects were recruited from callers to the toll-free Roswell Park Stop Smoking Hotline in Buffalo, NY. The subjects were enrolled in the study from August 1, 1984 to November 22, 1985. They were smokers (one or more cigarettes a day) who requested self-help smoking cessation informa- tion. On average, these smokers smoked 28 ciga- rettes a day. These subjects were also enrolled in a randomized trial testing the efficacy of five differ- ent self-help smoking cessation booklets. This trial tested format (high structure versus low structure) and quitting instructions (cold-turkey versus grad- ual reduction) as factors affecting effectiveness of self-help booklets. Extensive evaluations of this trial have been published elsewhere (7,8). In short, at followup, there were no differences between the groups in terms of quit attempts, proportion of quitters, use of the booklets, and multiple compli- ance measures. Trained hotline operators collected information on demographic, attitudinal, smoking history, and health status variables at initial contact. At the 6-month followup 1,552 (82 percent) were inter- viewed by telephone to obtain information on smoking status and socioenvironmental and use-of- cessation-help variables. The 343 persons not com- pleting followup interviews included 33 who re- fused and 310 who either moved to an unknown address or changed to an unlisted telephone num- ber. Unless otherwise specified, results reported are based on the 1,552 participants. Independent variables. Demographic variables meas- ured included age, sex, race, education, and mari- `One approach to studying smoking cessation is to compare successful quitters, recidivists, and nonstoppers in terms of factors that are potentially associated with smoking cessation. ' tal status. Attitudinal variables were measured using four-item Liken's scales. These attitudinal variables included perceived threat from smoking, perceived benefit from quitting, perceived ease or difficulty quitting, and perceived likelihood of quitting 6 months from baseline. Socioenviron- mental variables included presence of smokers at home, number of smokers among five closest associates, helpful support during quitting, and nagging during quitting. Perceived health status at baseline was measured by asking subjects to rate their present health as excellent, good, fair, or poor. Smoking history variables measured included number of cigarettes smoked daily, nicotine content of brand smoked, amount of nicotine smoked daily, duration of cigarette use, presence of a previous quit attempt, and duration of the longest previous abstinence from smoking. Amount of nicotine smoked daily was calculated by multiply- ing number of cigarettes smoked daily by the nicotine content of the brand smoked. Use-of- cessation-help variables included enrollment in a stop smoking clinic during the study, use of nico- tine gum, and group assignment for the random- ized trial (7,8). Dependent variables. Subjects were divided into three mutually exclusive categories based on their self-described smoking behavior during the study. The 813 nonquitters (52 percent) were those who reported that their behavior is best described by the statement: "Have not tried to quit smoking" or "Have tried to quit, but was not able to stay off cigarettes more than one day." One-third of this group (254 persons) reported not trying to quit during the study. The 497 recidivists (32 percent) chose the statement "Quit smoking for more than a day, but smoking again now" or described themselves as nonsmokers at followup but their self-reported quit period was less than 30 days before the interview. The 242 quitters (16 percent) described themselves as nonsmokers at followup and their self-reported quit date was at least 30 days before the interview. On the average, these rwr«+r..-wo.nw.. 1.p. vo/. 10a. de. • r» I 2C46399'793

Table 1. General companson of demographic and attltudrnal vartables by smoking status I I I I I I I I I I I I I I I I I I U*en o. oereanr _ NonQumers Pectawsrs putn+s Vanade (N.813) (N.49T) (N.242) Total P•vehue Demographic Age (years) ................................. 44.5 43.2 44.0 44.0 0.326 Sex: female ................................ 64.5 68.2 64.0 65.6 0.345 Race: nonwhite............................. 7.8 12.6 5.4 9.0 0.002 Education: college graduate'•2 ............... 17.8 18.7 29.4 19.9 0.011 Marital status: married ...................... 60.9 60.5 59.3 60.5 0.969 Atvtudinal Perceived threat from smoking: very likely .... 59.3 62.7 59.2 60.4 0.557 Perceived benefit from quitting: very likety'•' .. 63.3 71.3 72.0 67.2 0.017 Perceived ease or difficulty quitting: very difficuft'-43 .................................. 60.2 47.1 39.6 52.8 <0.001 Perceived likelihood of quitting in 6 months: very likety'.2.' ............................... 25.2 35.3 43.2 31.3 <0.001 ' P<0.05 bt nonmR!«I vrarsus Oftn. = P<0.05 br rrrarhvsb wnus qurtt.rf. 3P<0.06 /a norquRtm v.rwe nodN+us. Table 2. General comparison of socioenvironmental and health status variables by smoking status Are.n a a«oenc Nonoumw R.aeWts Orwan VunDN (N.t13) (N.49n (N.242) TOta/ PwMw $OCJOAnVIfOnn/enta/ Smokers at home'.z ......................... 46.1 43.2 29.1 42.5 <0.001 Number of smokers among 5 closest friends. .. 2.7 2.7 2.5 2.7 0.197 Helpful support during quitting'•' ............. 31.4 51.5 55.3 41.5 <0.001 Nagging duhng quitting" .................... 11.5 7.8 4.2 9.2 0.001 Health status Perceived statns at baseline: excellent" ..... 13.5 17.9 22.7 16.3 0.014 ' P<O.OS 1a nmquntm v.rwa quAMrs. = P<0.05 br natlnsts v.nus q~. 3 P<o.OS /or nwan"" vwwe r.admsb. Table 3. General comparison of smoking history and use-ofcessation-help variables by smoking status M.m w Pere«rt VranraON fNN.A1J) ( ROCKWim (N.~242) To4/ PvM» Smoking history Amount smoked (cigarettes per day) .......... 29.9 25.6 26.2 27.9 <0.001 Nicotine content (mg per cigarette) ............ 0.87 0.88 0.86 0.87 0.732 Daily nicotine use (mg per day) ............... 26.3 22.4 227 24.5 <0.001 Years of cigarette use' ...................... 25.1 23.0 23.4 24.1 0.012 Ever quit smoking'-2 ......................... 87.4 92.9 93.0 90.0 0.001 More than 30 days of previous abstinenoe ..... 33.3 46.2 49.8 40.0 <0.001 Uae of cessation help Self•help booklst group: 0.017 High-structure, cold turkey ................. 17.5 17.0 13.1 16.6 ... Higtrstructure, gradual reduction ........... 18.2 15.8 13.9 16.8 ... Low structure, cokd turkey .................. 14.9 19.4 25.0 17.9 ... Low structure, gradual reduction............ 16.9 17.6 13.9 18.6 ... Control ............................. 32.6 30.2 34.0 32.0 Clinic attendancei~a ~ ........................ 3.1 6.0 12.2 5.4 <0.001 Use of nicotine gum......................... 11.9 15.8 10.1 12.9 0.047 ' P<0.05 /a mrquRlMS vawt r.can~W. 2P<o.06 /a naquRlM vasus pr"ta. 3 P<0.05 br r.atlN+W v.eWI q~. 774 rubpe Mwtl~ Mpm+. 2C46999'794

~ subjects were off cigarettes for 4 1/2 months (mean Table 4. Significant preoictors of quitters versus nonqumers in = 20 weeks, median = 22 weeks). logistic regression analysis I I I I I I I I I I I I I I Analysis plan. All analyses were performed using SPSS and FREND software packages. The 343 nonresponders were compared to the 1,552 re- sponders in terms of variables collected at baseline using analysis of variance and chi-square tests. During the first step of the analysis of responders, each potential predictor was compared over the three outcome groups using chi-square or analysis of variance. Factors were considered to be signifi- cantly associated with outcome if the P-value was less than 0.05. If there were no significant associa- tions, the variable was not considered further. The next step involved contrasts between each of the pairs of outcome variables, that is, quitters versus nonquitters, quitters versus recidivists, and recidivists versus nonquitters. These contrasts were evaluated in two ways. First by bivariate analysis and then by multiple logistic regression analysis for each contrast. It was assumed that subjects in the study were members of a cohort with a finite probability of moving to the next step of quitting. The outcome in the risk analysis was the step which represented greatest progress in the quitting proc- ess. The exposure (predictor variable) category was presented in a way that would produce an estimate greater than 1. This was done in order to allow for direct comparisons of the relative effects of each of the factors evaluated. In order to make the risk estimates obtained in the logistic regression analyses comparable, all variables found to be significant in the three-group comparisons were included in the models for each of the contrasts. Because three comparisons were made using the regression models, in a 95 percent confidence interval a 98.3 percent level was used to take into account the effect of multiple compari- sons (100 percent - [5 percent = 3] = 98.3 percent) (9). Each stratum of categorical variables was introduced as an independent regression vari- able (dummy variable). Coefficient and standard error estimates derived from logistic regression models were utilized to calculate odds ratios and 95 percent confidence intervals (10). This analysis allowed the evaluation of the effect of each poten- tial predictor in the presence of all other potential predictors. Results Compared to responders, nonresponders were more likely to be younger (mean age = 38 versus Risk Conhdncti van.a.' rano Mr.w.r~ Demographic Education: Less than college ............... 1.0 ... College graduate ................ 1.8 1.1,2.8 Attitudinal Perceived ease or difficulty quitting: Very difficutt .................... 1.0 Other .......................... 2.1 1.4,3.2 Perceived likelihood of quitting in 6 months: Unlikely ........................ 1.0 ... Likely .......................... 1.7 1.1,2.7 Very likely ...................... 2.9 1.5,5.8 Socioenvrronmentaw Presence of smokers at home: Yes ............................ 1.0 ... No ............................ . 2.3 1.3,3.8 Helpfui support du ring quittting: No ............................. 1.0 .. Yes ............................ 2.0 1.3,3.1 HeaKh status Psrceived status at baseline: Other .......................... 1.0 Exael lent ....................... 1.6 1.1,2.8 Smokwro history Longest previous abatinence: None ........................... 1.0 ... 1-30 days ...................... 1.6 1.1.23 Nlore than 30 days .............. 1.8 1.2,2.8 use•of.oessadorHre/p Cessation clinic attendance: No . 1.0 Yes ............................ 4.4 1.9,10.1 ' M~dOlefon b 1lna var+.CW tlw bpMfc r•qrrbbrr modM MukiOMC rfa. o«ea.e b.rt.ik Nan qOrq, n.opKq aurr,q awak,p. N- .nak.e. a,nnon oa eq.rre. Un. w« aw arn"+p. uooba vr«+a "WWW'M ana w. oa rnootr+. gum. 29popflt COnAdMY~ ttlrfv~/ uM1p 9l.3 p~=M 1&rN b Yq vMO *OCONnf Muncie cog wr+.«r 44 years), nonwhite (16.3 percent versus 8.9 per- cent), male (41.2 percent versus 34.4 percent), and slightly less experienced with cigarettes (years of smoking = 20 versus 24). In these comparisons age and number of years smoking were highly corre- lated (Pearson's r = 0.872). Table I shows the three-group comparison of demographic and attitudinal variables. Four factors were eliminated at this stage of the analysis: age, sex, marital status, and perceived threat from smoking. Quitters were less likely to report their race as nonwhite compared with nonquitters. Quit- ters were more likely to be college graduates compared with recidivists and nonquitters. Quitters were more likely to perceive benefit from quitting and more confident of their ability to be off cigarettes 6 months from baseline compared with nonquitters and recidivists. These two factors dis- I I /M~r.po~ia 1111111131 ve/. tOt, Ms. • 77111 2C46399'795

I I I I I I I I I I I I I I I I I I I Table 5. Significant predictors of quitters versus recid vists m logistic regression analysis vanade ' F1r5k iAflhdllfCy ratio interval 2 Demographic Education: Other .......................... 1.0 College graduate ................ 1.8 1.1,2.9 Race: Nonwhite ....................... 1.0 White .......................... 2.4 1.1,5.6 Socioenvironmental Presence of smokers at home: Yes ............................ 1.0 No ............................. 1.8 1.1.2.7 ' In adtlRlOn to tl1dM vf1rHl11N, tM IoQtStIC togty,{qn rrqyM inctUOetl pNc*,yb betNfR from quKhnq. WrcMNO 11111106 oW ddfi0tAty Qummq, pereen'W Iik.Mihood of qutntng in 6 month., h.1pfW wpport during qummq, nspSprq Curing qunWq. p.reav.d nwm wlus tt tfW1/N.. amouftt .mac.d. dufatlon Of c9&-fl. us.. Mr auA srtfoKrq, bngNt pflVDUa abtaMMnq, booklet Qroup aTSgnrtMrM. ertaaoon dawe att.ndanert. and us. of mootnm gum. 295 04100 n eaMiO.rK1 mtuv.t wwq 98.3 t>•ra.rtt MvM to tax. Kno aesouvu motopt. Cant»rtaoro. . Table 6. Significant predictors of recidivists versus nonquitters in logistic regression analysis vaff~l~ ' R;a+< FM c«,fld.nc. w"'r 2 Attltudinal Perceived likelihood of quitting in 6 months: Unlikely ........................ .0 Likely .......................... 1.5 1.0,2.1 Very likely...................... 1.5 1.0,2.4 3molang history Amount smoked (per pack)3 .. . . . . . . 1.5 1.2,2.0 Socioenvironmental Helpful support during quitting: Yes ............................ 1.0 No ............................. 2.1 1.5,2.9 ' In addition to tn.w v.nabNs, the bystrc r.grKpon moaa mohidetl.aueauon1 rae., p.•cm.a benefit from a+rttx+g• a.rc.n.a wn or d+ffieuny awronq, a«. carvetl pxMthooG of a+Atuq in 6 monMS. MtdW wppon " Qumng, nagging during qurtmg. peee.nW h.aRh ftatua at tsawu». Caxatton ol aqarett. us.. .r.r quit smou,ng. i«~q.w pr.VIous aCSer+.ne., oookl.t proup usprnrm, ppwon elcme amnoane., and use ot n4adne gum. 295pMY.wtt ca+titl.Rei wtterHai WUfg a Ye.3 ptlrcetM NvM t0 tafa• Kqo acootXx mu/!pN campar»ons. ~ For ttws pta.1faon anotxw smab0 m aqarata t~•• ear .r.n groUp.0 1n pr1GU a« aay (20 aqar.nes M podc) to .rauaa tt» tt.a a..on .aomm,a a.ac of atWaw anaae. tinguished recidivists from nonquitters in the same direction of effect. There was a dose-response relationship for the expected extreme difficulty of quitting (nonquitters 60 percent, recidivists 47 per- cent, and quitters 40 percent) . Table 2 compares the socioenvironmental and health status variables of the three groups. Only one variable, the number of smokers among the respondent's five closest associates, was not differ- entially distributed among the three groups. Non- quitters reported less helpful support and more 7" .reMe Medut R.oata nagging during a quit attempt compared with quitters and recidivists. For example, 12 percent of nonquitters reported nagging during quitting whereas qnly 4 percent of quitters reported this perception. Compared with quitters, nonquitters and recidi- vists were more likely to have other smokers at home. Compared with nonquitters, quitters and recidivists were more likely to perceive their health status as excellent at baseline. In table 3 smoking history and use-of-cessation- help variables of the three groups are compared. The nicotine content of the brand of cigarettes -smoked was not associated with a differential distribution between the groups. Nonquitters re- ported less experience with smoking cessation in terms of previous quit attempts and previous days of abstinence compared with quitters and recidi- vists. Although 5 percent of study subjects attended a formal cessation program during the period of observation, there was a dose-response effect with formal cessation clinic attendance and quit status (quitters 12 percent, recidivists 6 percent, and nonquitters 3 percent). Table 4 shows variables that were able to dis- criminate between quitters and nonquitters in the multivariate analysis and the strength of these associations. Table 5 shows similar comparisons between quitters and recidivists, and table 6 com- pares recidivists and nonquitters. Discussion Participants in this study are not a representative sample of the general population of smokers. Nevertheless, they differ from persons who are involved in formal cessation programs since very little effort was required on their part to be included and remain in the study. They represent a sample of help-seekers who are likely to participate in public health programs for smoking cessation. Subjects in this study needed only to call a stop smoking hotline and identify themselves as smok- ers. Since participants needed to acknowledge smoking as a problem and take the steps to seek help, they may be similar to the help-seekers seen by primary care providers (11). This sample has a relatively high percentage of women (65.6 percent) and a relatively high percentage of college gradu- ates (19.9 percent). At baseline, only 8.8 percent reported their health status as poor, 51.2 percent reported their health status as good, and 16.4 percent as excellent. Attitudes seem to distinguish between nonquitters `2G46399'796

I I I I I I I I I I I I I I I I I I and the other two groups. IVonquitters appear more fearful of the difficulty of smoking cessation and more doubtful of their abilities to quit successfully than the other two groups. The construct of being more or less doubtful of one's ability to quit is similar to self-efficacy about cessation. Several workers have reported a strong effect for self- efficacy in predicting smoking cessation (6, 12-15). Those who did not quit live in a social environ- ment less supportive of smoking cessation. Quitters appear to be different from recidivists and nonquit- ters in terms of their socioenvironmental vari- ables: presence of smokers at home, helpful sup- port during quitting, race, and education. Others have reported similar results (12,16,1 7). Smoking is more common among those who are black and less educated (18). Perceived helpful support during quitting appears to be a stronger predictor of success than nagging. These two variables are related and likely to be influenced by the outcome of the cessation effort. Nevertheless, results sup- port a stong influence of the immediate environ- ment in the outcome of cessation attempts. Amount smoked discriminated between nonquit- ters and recidivists but not between quitters and recidivists. Thus, it appears that heavier smokers are more hesitant to try to quit, but once they make the attempt, they are as likely to succeed as lighter smokers, when other factors are kept con- stant. Length of previous abstinence was a factor that discriminated between quitters and recidivists. Similar results have been reported for other groups (6,19). Perhaps, some amount of success in quitting is necessary for some smokers before they are successful at becoming nonsmokers. Health status at baseline discriminated between nonquitters and quitters. Those who perceived their health status as excellent were more likely than the others to be quitters. This finding emphasizes the need for early intervention, before the negative health consequences of smoking are apparent. Cessation clinic participants were more likely than others to be quitters. Clinic attendance proba- bly represents a measure of motivation and addic- tion combined. When clinic attenders were com- pared with nonattenders, they were found to be heavier smokers and to perceive a greater health threat from smoking (7). In contrast, Fiore and colleagues reported that those using cessation pro- grams, a sample of the U.S. population, had lower rates of success than those quitting on their own (3). A likely explanation is that participants in this study represent a different population of smokers. They represent help-seekers who would benefit `Heavy smokers can be told with confidence that they are as likely to succeed as light smokers, when other factors are kept constant. Finally, it appears that a previous quit attempt is needed for some smokers before they are able to maintain cessation successfully. ' more from an intensive cessation program. In summary, this study demonstrates the utility of separating smokers into groups according to the stage of smoking cessation that they experience during the intervention. Findings from this study substantiate the notion that efforts to promote environments supportive of smoking cessation are likely to result in a larger number of successful quitters. Similarly, efforts to strengthen motivation and belief in personal ability to quit are likely to encourage more nonquitters to attempt to stop smoking. Heavy smokers can be told with confi- dence that they are as likely to succeed as light smokers, when other factors are kept constant. Finally, it appears that a previous quit attempt is needed for some smokers before they are able to maintain cessation successfully. This information is likely to benefit all health practitioners promoting smoking cessation. References .................................. !. Hatziandreu. E. J., et al.: Quitting smoking in the United States in 1986. J Nat! Cancer lnst 82: 1402-1406 (1990). 2. The health benefits of smoking cessation. A report of the Surgeon General. DHHS Publication No. (CDC) 9a8416, Public Health Service, Centers for Disease Control, Center for Chronic Disase Prevention and Health Promotion, Office on Smoking and Health, Rockville, MD, 1990. 3. Fiore, M. C., et al.: Methods used to quit smoking in the United States: do cessation prograrns help? JAMA 263: 2760-2765, May 23-30, 1990. 4. Schwaru. J. L., and Dubitzky, M.: Expressed willingness of smokers to try 10 smoking withdrawal methods. Public Health Rep 82: 835-861, October 1967. 3. Public Health Service: Adult use of tobacco, 1975. DHEW (NCI), June 1976. 6. Ockene, J. K., et al.: Relationship of psychosocial factors to smoking behavior change in an intervention program. Prev Med 11: 13-28 (1982). 7. JaEn-Ciudoba, C. R.: Smoking cessation in a cohort of smokers interested in self-help. Doctoral dissertation, State University of New York at Buffalo, Publication No. qe~w~..-wqn~b« tsp, vel. t0a. qe. • m 2C46399'79'7

I 1 I I I I I I I I I I I I 8823995. Michigan Dissenatlon Index, vol. 99, issue 1113, 1988. 8. Cummings, K. M., Emont, S. L., Jaen-, C. R., and Sciandtf: R.: Format and quitting instructions as factors influencsag the impact of a self-administered quit smoking program. Health Educ Q 15: 192-216 (1988). 9. Netter, J., and Wasserman, W.: Applied linear statistical models: regression, analysis of variance and experimental design. Richard D. Irwin, Inc., Homewood, IL, 1974. 10. Kleinbaum. D. G., Kupper, L. L., and Morgenstern, H.: Epidemiologic research: principles and quantitative meth- ods. Lifetime Learning Publications, Belmont, CA, 1982. 11. Aday, L., and Shortell, S. M.: Indicators and predictors of health service utilization. In Introduction to health services, S. J. Williams and P. R. Torrens, editors. Ed 3, New York, 1988, pp. 51-81. 12. Curry, S., Thompson, B., Sexton, M., and Omenn, G. S.: Psychosocial predictors of outcome in a worksite smoking cessation program. Am J Prev Med 5: 1-7 (1989). 13. Eisinger, R. A.: Psychosocial predictors of smoking recidivism. J Health Soc Behav 12: 355-362 (1971). 14. DiClentente, C. C.: Self-efficacy and smoking cessation maintenance: a preliminary report. Cognitive Ther Res 5: 175-187 (1981). 15. Condiotte, M. M., and Lichtenstein, E.: Self-efficacy and relapse in smoking cessation programs. J Consult Clin Psychol 49: 648-658 (1981). 16. Mermeistein, R., et al.: Social support and smoking cessation and maintenance. J Consult Clin Psychol 54: 447-453 (1986). 17. Cohen, S., and Lichtenstein, E.: Partner behaviors that support quitting smoking. J Consult Clin Psychol 58: 304-309 (1990). 18. Public Health Service.: The health consequences of smok- ing: nicotine addiction. A report of the Surgeon General 1988. D~iHS Publication No. (CDC) 88-8406, Office on Smoking and Health, Rockville, MD, 1988. 19. Jackson, P. H., Stapleton, J. A., Russell, M. A. H., and Merriman, R. J.: Predictors of outcome in a general practitioner intervention against smoking. Prev Med 15: 244-253 (1986). ~ ~ ~ 779 wewa Narn+ w.«4 ~ On I