Tobacco Documents Online

PREVENTIVE MEDICINE 22. 261-271 (1993) I I I I I I I I I I I I I I I I I Predictors and Reasons for Relapse in Smoking Cessation with Nicotine and Placebo Patches JESPER NORREGAARD, M.D.,t PHILIP TONNESEN, M.D.. DR.MED.SCI., AND LARS PETERSEN, M.D. Department of Pulmonary Medicine P. Bispebjerg Hospital. Copenhagen. Denmark Background. The reasons for relapse to smoking are not fully understood but several factors are of importance. Addiction to nicotine seems to play a prorrunent role but there are several other contributing factors. Methods. To determine predictors of outcome in smoking cessation. we performed uni- vatiate and multivatiate analyses in a large smoking cessation trial comprising 289 subjects. Weight gain and withdrawal symptoms were analyzed separately as predictors. To deter- mine self-perceived reasons for relapse we created a questionnaire, which was answered by 132 relapsers. Results. Previous attempts to quit smoking and a low saliva cotinine concentration were significantly associated with abstinence in the nicotine-treated group. A trend toward higher abstinence rates was found among males and among subjects with a low nicotine depen- dency score. Logistic regression analysis showed higher success rates in subjects with the largest weight gain during the first weeks of quitting in contrast to higher relapse rates in subjects who had the greatest weight gain after 3 months. A high score on withdrawal symptoms was not predictive of relapse. Subjects with "slips" had a markedly increased relapse rate. Craving for cigarettes was the most often seif-reported (48%) reason for re- lapse. Conclusions. The relation between weight gain, withdrawal scoring, and outcome seems rather complex. Craving for cigarettes was the most reported reason for relapse. C 1s93 Academx Press. lx. INTRODUCTION The average 1-year sustained abstinence rate in most smoking cessation trials is between 10 and 20%. The highest percentages are reached when behavioral ther- apy is used in combination with nicotine substitution (1, 2). Thus approx 80°!o fail to succeed in quitting smoking, even though most claim to be motivated to quit at the start. The reasons for relapse are not fully understood but several factors are of importance. Addiction to nicotine is one of the most prominent factors (3). Earlier studies have determined different predictors of outcome in smoking cessation (4-8). Although discrepancies exist between the different studies, there has been constancy in reporting a higher incidence of relapse in subjects of lower socio- economic status and with a high daily cigarette consumption, high nicotine de- pendency score, and high alcohol intake. Higher success rates have been reported among subjects who live with a nonsmoking spouse, who have tried to quit pre- Viously, who have heart diseases, and who believe they will succeed in quitting. ' To whom reprint requests should be addressed. 261 0091-7435/93 S5.00 Copyttsht O 1993 by Aeademw Press. Iac. All n&t of ropodwnoe m aaY form rexned. I

262 NORREGAARD. TONNESEN. AND PETERSEN I I I I I I ~ I I I I I I I I Other studies have analyzed situational factors of relapse from questionnaires or interviews (9, 10). Early relapse has been reported most often to be caused by the urge to smoke while psychosocial aspects of the.,smoking habit (i.e.. stress. other smokers, alcohol intake) are the cause of late relapse. These results might have been confounded due to the retrospective nature of the data. Only two of four prospective studies have found recidivism to be related to the intensity of the withdrawal symptoms (11-14). The mean weight gain after smoking cessation from a large cohort study has been reported to be 3-4 kg (15) arid weight gain has been suggested to be an important reason for relapse in smoking cessation (16). As we do not find the mentioned predictors to be fully explored especially in relation to nicotine substitution. we analyzed possible predictors of outcome from a placebo-controlled nicotine patch trial. In that context we examined whether the withdrawal symptoms and the weight gain were predictors of outcome. These prospective data were supplemented with an analysis of withdrawal symptoms during the period of relapse and the self-reported reasons for relapse. SUBJECTS AND METHODS Subjects Two hundred eighty-nine subjects participated in a smoking cessation trial; 145 were allocated to receive nicotine patch treatment and 144 to receive placebo patches (Table la). For details of this study see (17). Thirty subjects were sus- tained abstainers after 1 year. All subjects were instructed to attend the clinic when they relapsed. At the clinic the relapsers were asked to complete a simple questionnaire concerning reasons for relapse. A large proportion of the relapsers, however, never attended the clinic, but reported relapse by phone or letter. Therefore, only 132 of 259 (51%) completed the questionnaire (see Results for the specific questions). Assessment A sample of at least 3 ml of unstimulated saliva was collected at the initial visit. TABLE la BASELINE CHARACTERlSTiCS OF 289 SMOKERS Item Percentage females 71 Age (years) 45.3 (23-74) No. of cigarettes/day 21 (10-40) Nicotine yield (mg/cig) 1.6 (0.9-2.1) Tried to quit previously 88% Duration of smoking (years) 26 (4-50) Saliva cotinine (nglmi) 439 (71-1290) Exhaled air CO (ppm) 29 (&-66) Fagerstrdm TQ (max 11) 7.4 (4-11) Horn-Russell Scale (max 27) 15.7 (5-27) Note. Mean values; range in parentheses.

I i I I I I I I I I I I I I I PREDICTORS OF SMOKING RELAPSE 263 Samples were collected in the afternoon, and within I hour the sample was frozen at - 20°C. The saliva samples were analyzed for cotinine by chromatography (18). A single saliva cotinine concentration value is a good measure -of daily nicotine intake during smoking and nicotine patch therapy (19). The number of cigarettes smoked daily as well as the nicotine content per cigarette was registered initially. The Fagerstrom Tolerance Questionnaire (20) and a modified Horn-Russell Scale (21. 22) were used as an indirect measure of the subjects' degree of nicotine dependence. Subjects were weighed on a balancing beam scale at initial visit and then at each subsequent visit to the clinic. Seven visits were scheduled for the 52-week period (entry and weeks 1, 3, 6, 12, 26. and 52). At each visit subjects were asked if they smoked now and if they had smoked since last visit. Between two visits a single slip was allowed. A slip was defined as unlimited smoking for 24 hr. followed by up to 5 days of smoking up to 15% of the numbers of cigarettes smoked at entry. Carbon monoxide (CO) in end- expiratory air was measured using a CO analyzer (Bedfont Monitor). Levels of 10 ppm or less were used to indicate a nonsmoker and all subjects with a CO value of 11 ppm or more were counted as relapsed. During the initial weeks of the study, each subject was requested to record nine withdrawal symptoms on a 100-mm visual-analogue scale on a daily basis. The lowest scoring was worded "Nothing," the highest scoring "Very severe." The diary from the first week was returned by 259 subjects (89%a) and those from the second and third weeks by 170 subjects. Treatment consisted of daily use of a nicotine or placebo patch. The nicotine patch was 30 cm'- in size, with a nicotine content of 0.83 mg/cm2; it released 15 ± 3.5 mg (mean - SD) of nicotine during a 16-hr period. The maximal plasma concentration of nicotine of 14.2 ± 3.9 ng/ml was attained 5-10 hr after applica- tion. After 16 and 24 hr the nicotine concentrations were 8.9 ± 1.8 and 2.5 ± 0.8 ng/ml, respectively (Kabi Pharmacia, unpublished data). About 95% of all absti- nent nicotine-treated subjects used the patch during the first week. From the cotinine concentrations at the following visits, it can be estimated that 80-85% used the nicotine patch regularly for at least 6 weeks but only 50-60% used the patch at the 12-week visit (17, 19). Each visit lasted 20 to 60 min. No group support or behavioral therapy was administered; thus most of the time was devoted to collection of data and assess- ments. Ethics The study was conducted in accordance with the Declaration of Helsinki. The study was approved by the local ethics committee and informed consent was obtained from' each participant. Statistical Methods Ordinal data were correlated by the Spearman rank coefficient of correlation. The single items, weight gain, and each withdrawal symptom were compared in relapsers and quitters by nonparametric tests (Mann-Whitney, Kruskal-Wallis, I

I I 1 I a I I I I I I I 264 NORREGAARD. TONNESEN, AND PETERSEN Fisher's exact. and X=). The predictive importance of the different parameters was analyzed using stepwise logistic regression. In this the predictors were converted to binary variables using the median as the cutoff point. The main outcome mea- sure was chosen to be success at the 6-week visit, because of a high patch com- pliance at that time and because a sufficient number of abstainers could be in- cluded to attain a reasonable strength of the analysis. In the analysis of withdrawal scoring and weight gain in relation to outcome, a logistic regression analysis adjusting for treatment and sex was performed. All P values are two-tailed. RESULTS Predictors Table lb shows the different pretreatment characteristics in relation to outcome at the 6-week visit. Three indirect measures of nicotine dependency were higher among relapsers in the nicotine-treated group. The same trend was observed in the placebo group. A stepwise multiple logistic regression analysis showed nicotine treatment to be by far the most important predictor of outcome (P < 0.001) after 6 weeks (Table 2a). We therefore performed two separate analyses for nicotine- and placebo- treated subjects (Tables 2b and 2c). The smoking saliva cotinine concentration was the most important smoking-related parameter in the nicotine-treated group and subjects who had never tried to quit before had lower abstinence rates. A similar analysis of the placebo group showed significantly higher success rates among males and among subjects with a low Fagerstrom score. TABLE lb UNIVARIATE ANALYSIS OF BASELINE CHARACTERISTICS OF 289 SMOKERS IN RELATION TO OUTCOME AT 6 WEEKS Nicotine group Placebo group Item Abstinent (n - 77) Relapsed (n = 68) Abstinent (n = 24) Relapsed (n = 120) Percentage females 70 75 50 75' Age (years) 47 43 46 45 No. of cigarettes/day 20 2-" 19 22' Nicotine yield (mg/cig) 1.5 1.6 1.6 1.6 Tried to quit previously (S6) 96 77" 91 89 Number of tries 3.0 2.7 2.2 3.4' Duration of smoking (years) 24 27 20 22 Saliva cotinine (ng/ml) 383 453" 393 439 Exhaled air CO (ppm) 26 29 26 28 Fagetstrom TQ (max 11) 6.9 7.3 6.6 7.6' Hotg-Russell Scale (max 27) 15 16 13 16' Weight (kg) 67 70 71 67 BMI (kg/m=) 23 24 24 24 Note. Mean values (% values). ' P < 0.05. "' P < 0.01. Mann-Whitney or Xz (% items).

I I I I I I I I I I I I I I I I 1 I PREDICTORS OF SMOKING RELAPSE TABLE :a PREDICTORS OF ABSTINENCE AT THE 6-WEEK FOLLOWUP 26-5 Item N Odds ratio .CI° Pb Treatment <0.001 Placebo 137 1 Nicotine 135 5.3 3.0-9.4 Saliva cotinine concentration (ngiml) <0.01 1-424 135 1 >425 137 0.4 0.3-0.8 Previously attempted to quit 0.02 No 33 1 Yes 239 3.8 1.3-10.8 Horn-Russell Scale 0.03 1-16 153 i 17-27 119 0.5 0.3-0.9 Smoke years 0.04 2-25 150 1 More than 25 122 1.8 1.0-3.4 Sex 0.13 Female 195 1 Male 77 1.7 0.9-3.2 Daily cigarette consumption 0.26 10-20 182 1 More than 20 90 0.7 0.4-1.4 Fagerstrom Tolerance Questionnaire 0.34 3-7 147 I 8-11 125 0.7 0.4-1.4 Body-mass index (kg/m2) 0.78 <23 136 0 >23 136 0.9 0.5-1.7 Age (years) 0.90 <44 139 ( >44 133 1.1 0.6-1.7 Note. Seventeen subjects with missing values not included. ° CI: 95% confidence interval. b Significance levels for testing odds ratio 1. The abstinent subjects at 6 and 12 weeks were analyzed separately with regard to abstinence at 1 year, using the same mentioned predictors including nicotine treatment. None of the above predictors reached significance in the stepwise logistic regression analysis (which included 85 and 61 subjects, respectively). Abstinent with Slips Slips were a highly significant predictor of relapse during the following period. At the 3-week visit 32% of 56 abstainers with slips at the 1-week visit had relapsed compared with 13% of the completely abstinent subjects (P < 0.01). At the 6-week visit 44% of 55 subjects with slips at the 3-week visit had relapsed compared with 9% of the completely abstinent subjects (P < 0.01). At the 3-month visit 72% of 29 with slips at the 6-week visit had relapsed at the 3-month visit compared with 9% in the totally abstinent group (P < 0.01). l

I I I I I I I I I I I I I I I I I 266 NORREGAARD. TONNESEN. AND PETERSEN TABLE :b PREDICTORS OF ABSTINEtiCE AT THE 6-WEEK FOLLOWUP IN NICOTINE•TREATED SUBJECTS Item .ti' Odds ratio CI' P° Previous attempted to quit 0.01 No 18 1 Yes 117 6.7 1.8-24.7 Saliva cotinine concentration Ingrml) 0.03 1-424 70 1 >425 65 0.4 0.2-0.9 Smoke years 0.06 2-25 80 1 More than 2_5 55 2.1 1.0-4.4 Horn-Russell Scale 0.06 1-16 79 1 17-27 56 0.5 0.2-1.0 Body-mass index (kg/m2) 0.17 <23 68 0 >23 67 0.6 0.3-1.3 Daily cigarette consumption 0.29 10-20 89 1 More than 20 46 0.6 0.3-1.5 Age (years) 0.59 <44 73 1 >44 62 0.7 0.2-2.5 Sex 0.96 Female 98 1 Male 37 1.0 0.4-2.5 Fagetstrom Tolerance Questionnaire 0.99 3-7 79 1 8-11 56 1.0 0.4-2.4 ° CI: 95% confidence interval. b Significance levels for testing odds ratio 1. Weight Gain as Predictor No significant differences were observed between the weight gain of those who relapsed and that of the patients who stayed abstinent during the following period using univariate analysis (Fig. 1). However, a logistic regression analysis of the complete abstainers, including weight gain during the first week, treatment, and sex, showed that high weight gain during the first week was a significant predictor of success at the 6- and 12-week visits (P < 0.05, n = 85). A similar trend was observed for weight gain at the 3-week visit using the 3-month visit as outcome measure (P = 0.07, n = 75). However, a reverse trend was observed for weight gain after 6 weeks and outcome at 6 months (P = 0.06, n= 74), and high weight gain after 3 months was a significant predictor of relapse at the 6- and 12-month visit (P = 0.02, P = 0.04, n= 60). Withdrawal Symptoms Many of the scores for withdrawal symptoms were scored significantly higher among relapsers (Table 3). A trend toward higher scoring among subjects with

I I I I I I I I I I 1 i 1 I I PREDIC'TORS OF SMOKING RELAPSE 267 TABLE :c PREDICTORS OF ABSTINENCE AT THE 6-WEEK FOLLOwUP IN THE PLACEBO GROCP Item N Odds ratio ` CI' P° Sex <0.01 Female 195 1 Male 77 0.3 0.1-0.7 Fagerstrom Tolerance Questionnaire 0.04 2-7 60 1 8-11 65 0.3 0.1-0.9 Age (years) 0.26 <44 73 1 >44 62 1.9 0.7-5.0 Saliva cotinine concentration (ngiml) 0,34 1-424 70 1 >425 65 0.6 0.3-1.6 Horn-Russell Scale 0.35 1-16 79 1 17-27 56 0.6 0.2-1.7 Daily cigarette consumption 0.50 10-20 89 1 More than 20 46 0.7 0.2-2.1 Body-mass index (kg/m=) 0,48 <23 68 0 >23 67 1.5 0.5-4.7 Smoke years 0.81 2-25 80 1 More than 25 55 1.2 0.3-5.5 Previous attempted to quit 0.92 No 18 1 Yes 117 1.1 0.2-6.3 ° CI: 95% confidence interval. b Significance levels for testing odds ratio 1. slips could also be observed. The predictive value of the withdrawal symptoms in relation to outcome was analyzed using different methods. A univariate analysis of the withdrawal symptoms during the first week for abstainers without slips (n = 86) and outcome at the following visits at 3 weeks, 6 weeks, or 3 months showed no significant predictive value of the withdrawal scores. A similar analysis of the withdrawal symptoms during the second and third weeks in relation to outcome during the following visits did not show any differences. If. however, subjects with slips were included in the analysis, high craving during both periods did predict later relapse (P < 0.05). A nonparametric correlation analysis (Spearman) of the withdrawal symptom score of the completely abstinent subjects in relation to time of abstinence did not show any negative correlation between length of abstinence and withdrawal scor- ing. In fact, high scoring for frustration. anxiety, and difficulty in concentration during the first week and hunger during the second and third weeks were signif- icantly (P < 0.05) positively correlated to length of abstinence. Neither a logistic regression analysis including the significant predictors from the above analysis

I ~ I I I I I I I I I I I I I 268 51 NORREGAARD. TONNESEN. AND PETERSEN (Nu+ter of subpots aDove eacn caurtn) 12 30 Ftc. 1. Weight gain in relation to abstinence in the next following period, nor the withdrawal symptom scoring showed any tendency of high withdrawal score predicted relapse. Self-Reported Reasons for Relapse Craving for cigarettes was the most often reported reason for relapse during the TABLE 3 MEAN WITHDRAWAL SYMPTOMS DURING THE FIRST WEEK iN RELATION TO SMOKING STATUS AT THE END OF THE FIRST WEEK (MEDIAN VALUES) Abstainers Item Complete With slips Relapsers Nicotine group n= 57 n= 33 n= 41 Craving for cigarettes 39 38 61" Irritability 11 12 20 Frustration 5 6 14** Anger 5 6 6 Anxiety 6 7 7 Difficulty concentration 6 10 15 Restlessness 21 20 32 H unger 27 17 23 Depression 5 8 9 Placebo group n= 29 n= 23 n= 75 Craving for cigarettes 37 51 62" Irritability 17 18 23 Frustration 4 6 11'• Anger 6 5 9 Anxiety 5 9 13" Diffsculty concentration 7 19 l7• Restlessness 14 31 29 Hunger 15 22 16 Depression 4 8 9" ' P < 0.05. •' P< 0.01: Mann-Whitney test for scoring of complete abstainers vs relapsers.

I I I I I I I I I I I I I PREDICTORS OF SMOKING RELAPSE 269 initial period and a common reason in the placebo-treated group (P < 0.05). Relapse due to external cues was the most important reason for late relapse and was more common in the nicotine-treated group (P < 0.05). From Table 4 it appears that the weight gain was of minor importance in re- lapsing. although more important for late relapsers compared with early relapsers (P < 0.01). One-half of the subjects reported increased appetite and one-third reported changes in food intake (Table 5). Subjects reporting increased appetite after smoking cessation did not differ in smoking cotinine concentrations com- pared with subjects who did not. DISCUSSION Our analysis of predictors of outcome focus on objective measurements and two commonly used questionnaires to measure nicotine dependence. Our results show that nicotine treatment was by far the most important predictor. The pre- dictive value of previous attempts to quit was confirmed (6). This supports the view that smoking cessation is a process in which the smoker passes through different stages (23). Smoking cotinine concentration was also a significant pre- dictor of outcome. This parameter is an indirect measure of nicotine intake and was found to be significant even in the nicotine-treated group. This could indicate either that behavioral factors were of importance in addition to nicotine addiction or that a sufficient nicotine substitution was not attained for those subjects with the highest nicotine intake during smoking. Our results do not confirm the findings by others of a relationship between the severity of withdrawal symptoms and later relapse (11, 12). In fact, our results indicated the opposite. It is possible that this was caused by a correlation between withdrawal scoring and other confounding variables such as alcohol intake or social status. It is not stated clearly in the two cited studies whether abstainers with slips are included. Our study shows that smoking very few cigarette is a strong predictor of relapse and is correlated to high withdrawal scoring. This emphasizes that slips should be carefully avoided; this should be emphasized in all TABLE 4 REASONS GIVEN FOR FAILURE TO QUIT Statement Early relapse Late relapse Nicotine Placebo Nicotine Placebo (n - 24) (n-64) (rt-32) (n - 9) The timin` was wrong The weight gain was too much Craving for ciprettes was too much External causes (illness, social. other problems) I could not function at work or at home J relapsed by chance and could not stop again i.ite is not worth living without cigarettes No cause 13 25 6 0 13 3 31 33 46 61 34 44 8 30 56 11 33 17 6 22 17 13 28 22 0 6 6 ll 16 3 6 11 Note. More than I answer allowed. Data given as percentage of column number. I

1 270 NORREGAARD. TONNESEN. AND PETERSEN I I I I I I I I I I I I I I I I I TABLE 5 ANSWERS ON DIFFERENT WEIGHT-RELATED QCESTiONS AFTER CESSATION OF SNOKING (n = 132) Question Yes (`1c) No (hc) Do not know (%) No answer (`~c) Did you change food composition when stopping smoking? 25 (rt 8 3 Did you eat more between meals? 31 48 18 2 Did you eat more sweets. cakes. etc.? 38 50 9 3 Did your appetite increase? 48 30 16 7 smoking cessation programs. It is also essential to register lapses in abstinence (slips) in smoking cessation trials. Many of the withdrawal symptoms were significantly higher among relapsers than among completely abstinent subjects. and craving for cigarettes was the highest scored withdrawal symptom. This suggests that subjects relapse due to withdrawal symptoms; however, as these symptoms were registered during the period of relapse, scoring might be influenced by the subjects' emotional reaction to the failure of quitting. Craving a cigarette was also the most important self-reported reason for re- lapse, especially for early relapse in the placebo group. Many subjects reported to have relapsed due to external cues. a statement more often reported among the late relapsers. These findings confirm that withdrawal symptoms are more prom- inent in the early phase after quitting and that psychosocial factors are a more common reason for late relapse (9). Increased appetite and food intake were often reported, but weight gain was reported to be the reason for relapse by only 20% of the subjects. We found a slight but significant correlation between weight gain in the initial period and later success in the multivariate analysis, which suggests that prevention of weight gain should not be emphasized during this period. The subjects with the greatest weight gain might have increased caloric intake and intake of sweets. We spec- ulate that the mechanism might be that food intake decreases withdrawal symp- toms, maybe by increasing blood sugar as suggested in a small pilot study (24). Our results, however, indicate that after the initial period weight gain should be prevented due to a higher relapse rate among subjects with the largest weight gain. CONCLUSIONS Further studies are needed to establish whether a higher nicotine substitution would negate the relationship between outcome and indicators of high nicotine dependency such as cotinine concentrations. As permanent abstinence often is achieved after several attempts, further smoking cessation studies should include recycling of relapsers. Potential quitters should be strongly advised to avoid all smoking. Further studies should try to confirm the relationship found between weight gain and outcome and its underlying mechanism. REFERENCES 1. Prignot J. Pharmacological approach to smoking cessation. Eur Respir ! 1989: 2:550-560.