Tobacco Documents Online

Journal ofConsuiUnti and Climeal Psveholo{Y CopYright 199: b, the American Ps~enoioucai assurae^n lac `„-, t9G:. voi b0. ~0 5.6E9-697 , I I I I I i I I I I I I I I I I I Tobacco Withdrawal in Self-Quitters John R. Hughes Departments of Psychiatrv Psychology, and Family Practice University of Vermont Self-reported and observer-rated signs and symptoms of nicotine withdrawal were assessed preces- sati.on and 2, 7,14, 30, 90, and 180 days postcessauon in smokers who quit on their own for 30 days. Anxiety, difficulty concentrating, hunger, irritability, restlessness, and weight gain increased, and heart rate decreased, postcessation (p <.001). Except for hunger and weight gain, these symptoms returned to precessation levels by 30 days postcessation. Craving, depression, and alcohol or caf- feine intake did not reliably increase. Postcessation depression, but not withdrawal symptoms. craving, or weight gain, predicted relapse. These results are consistent with prior studies. Cessation of smoking typically produces irritabilitx anxietx nocturnal awakening, depression, difficulty concentrating, hunger, restlessness, impatience, and a strong desire (i.e, crav- ing) for nicotine (American Psychiatric Association, 1987; Hughes, Higgins, & Hatsukami, 1990). Cessation also de- creases heart rate and increases weight (American Psychiatric Association, 1987; Hughes et a1,1990). These signs and symp- toms are reliable (Httghes, Hatsukami, Pickens, & Svikis, 1984), observable by others (Hughes, Gust, Skoog, Keenan, & Fenwick,1991; Hughes & Hatsukami, 1986), and may be clini- cally significant jHughes & Hatsukami, 1986). Many of these appear to be withdrawal effects in that they are time limited (Hughes et a1,1991; Hughes et a1,1990). Many also appear to be due to deprivation of nicotine in that they are relieved by nicotine replacement (Fagerstrom, 1988; Hughes et al, 1991) and occur with cessation of nicotine gum (Hughes, Hatsukami, & Skoog, 1986; West & Russel1,1985). With one exception (Gritz, Carr, & Marcus, 199 1), prospec- tive studies of abstinence effects have been restricted to the fewer than 5% of smokers who quit through treatment pro- grams (Fiore et a1,1990). Such smokers differ from average U.S smokers who quit on their own in several respects (Fiore et al, 1990). Importan* they appear to be more dependent than self-quittets: for example, they smoke more cigarettes (Fiore et al,1990) and score higher on dependencx scales (Russell, Peto, & Pate1,1974). Prior prospective studies of withdrawal have also failed to include adequate control groups of nonsmokers or continuing This study was funded by Grant HL-39220 from the National Heart, Lung, and Blood Institute and Research Scientist Development Award DA-00109 to John R. Hughes from the National Institute on Drug Abuse. I thank Kevin Cruser, Pam Shea, William Valliere, Sara Pepper, Laura Solomon, and Brian Flynn for their help in conducting the studyc I thank Greg Goodwin, Gary Badger, and Jim Fenwick for their help with the analyses. Correspondence concerning this article should be addressed to John R. Hughes, Department of PsychiatrX Human Behavioral Pharmacol- ogy Laborztory4 University of Vermont, 38 Fletcher Place, Burlington, Vermont 05401-1419. smokers (Hughes et al, 1990). These groups are necessary to test "withdrawal" versus "self-medication" interpretations of postcessation effects (Goldstein, 1987; Mangan & Golding, 1984). For example, consider anxiety scores. Withdrawal would be inferred if precessation values among smokers were similar to those of nonsmokers and abstinence values of smokers were greater than those of nonsmokers. Self-medication would be inferred if precessation scores among smokers were less than those of nonsmokers and abstinence values of smokers were equal to those of nonsmokets. The present study examined the validity reliability, inci- dence, magnitude, and time course of nicotine withdrawal in persons who attempted to stop smoking without treatment. Control groups of continuing smokers, never-smokers, and long-term ex-smokers were included. The study also examined reasons for variability in the severity of nicotine withdrawal and whether severe withdrawal prospectively predicted relapse. Method Another article reports the recruitment methods, subject flow, sam- ple characteristics, and cessaiion rates in mote detail (Hughes et al, in press). The present article focuses on withdrawal effects. Subjects Subjects were obtained through newspaper and radio advertise- menu throughout Vermont and nearby upstate New York, western New Hampshire, and western Massachusetts. The ads recruited smokers who were about to quit on their own and offered brief tele- phone counseling and payment (S20 per hour) for filling out forms. Subjects were unaware of the amount of payment until after entering the studx Because this was a study of the effects of abrupt cessation in regular smokers, we excluded individuals who were not smoking daily, had n:ceatly reduced their smoking by 50%, had recently switched to a low-nicotine cigarette, were planning on stopping through gradual re- duction, or had littk motivation for cessation. Among the 1,396 indi- viduals screetted. 566 (41%) were excluded by one or more of these criteria. Those included and excluded did not differ on demographics or smoking history variables. The 830 remaining potential subjects gave informed consent. They named two observers who were 18 years old or older and who saw the 689

I I I I I I I I I I I I I I I I I I 690 JOHN R. HUGHES subject at least 3 hours per day. 5 days per week. One of the observers in each pair was called to verify how much he or she saw the potential subject. to obtain the observer's consent to periodically rate the subject on several behaviors, and to verify abstinence or smoking, for which he or she was paid 53. Among the potential subjects. 630 (76% of 830) were reached for precessatton interviews and still had plans to stop smoking in the next 2 to 10 days. Those reached and not reached for precessation interviews did not differ on demographics or smoking history variables. Among the 630 individuals reached, the gender percentages and the means, and standard deviations (in parentheses) for subject chancter- istics were the following: 65% were women; age - 38.7 (11.9) years; cigarettes smoked per day = 26.7 (12.3 ): nicotine yield - 0.85 (0.31) mg per cigarette; duration of smoking = 21.6 (11.8) years; percentage v, tto tried to quit in the past = 92%; and Fagerstrom score (Fagerstrom & Schneider, 1989) = 6.2 (1.8). The Fagerstrom score consists of seven items concerning nicotine intake (e.g., numberof cigarettes smoked per day) and behavioral indexes of dependence (e.g., time to first cigarette after awakening). As delineated in a prior article (Hughes et al., in pt ess), these subjects were similar to subjects in other studies of self- quitters (Cohen et al" 1989), except that the proportion of women was greater in the present study. Procedures At the precessation interview, subjects were asked to rate a set of symptoms and behaviors composed of nicotine withdrawal effects (see Table 1) plus some filler symptoms (e.g, feel better about myself). These were listed in random order and labeled as a"behavior check- list" rather than as a withdrawal scale. Observers completed a similar scale, but several items were deleted as pilot work indicated observers could not rate these reliably (Hugheset a1.1991; Hughes, Hattukami, Pickens. Krahn, et al.,1984). The behaviors were rated for the last 24 hours from 0= none, 1- mild, 2= moderate, to 3- seven. These self-report and observer scales previously have been shown to be sensi- tive to the effects of abstinence and of nicotine replacement (Hughes & Hatsukami, 1986; Hughes et al, 1990, Hughes, Hatsumaki. Pickens, Krahn. et al.. 1984; Hughes et a1.,1991). Subjects were also mailed a two-paragraph instruction sheet and an illustration on how to take one's right carotid pulse. At each telephone contact, subjects took their pulse for 30 s on two occasions. In a pilot study of 25 subjects using this technique, intrarater stability was 0.77 and interrater agreement with a trained technician's simultaneous pal- pation of the radial pulse was also 0.77. In addition, subjects did not systematically report higher or lower pulse rates than did technicians. Subjects were given 15 min to have any questions about quittittg answered by trained counselors over the telephone. Subjects had pteviy ously been sent a stop-smoking booklet (Quit and Win; Pechacek, Ar- ken. & Johnson, 1982). Specific answers to the more common ques- tions were developed a priori in pilot testing and were similar to those described elsewhere (Htuba dc Knttks,1986). Subjects were given no information concerning witbdrawal symptoms and were not advised in methods that might abate withdta.val (e.g., relaxation). Subjects were told that no more counseling would be availabk as this was a study of self-quitting. Subjects set a quit date and informed the interviewer of this date. Subjects and observers were contacted by telephone at 2, 7, 14, 30, 90, and l80 days after their quit date, and the behavior checklist and smoking status were obtained. All subjects whose data are included were reached within 2 days of these follow-ups. If the subject reported abstinence at all follow-ups and this report was not refuted by the observer, a research assistant arranged a home visit within a month of the I 80-day follow-up. At this visit, a breath sample for carbon monox- ide and a saliva sample for cotinine were obtained (Jarvis. Tunstall-Pe- doe. Feyerabend. Vesey, & Saloojee. 198,7). Control groups of 56 long-term ex-smokers (abstinent for more than f,vear), 67 current smokers (currently smoked daily), and 61 never- smokers (never smoked daily for I month) were recruited from the observers. These subjects were tested initially and then 7 and 30 days later. These control groups did not differ from each other or from the subjects, with the exceptions that experimental subjects were slightly younger than control group subjects (36.1 vs. 36.5-40.1 years, p <.05) and that never-smokers were slightly more educated than smokers, current smokers, and experimental subjects (14.3 vs. 13.0-13.6 years, p < .05). Data Analysis The major data analyses used only smokers abstinent at the 2-, 7-,14-, and 30-day follow-ups. Many smokers who eventually ars abstinent smoke a few cigarettes on a few days after their quit date. When we divided subjects into those who smoked some but averaged 1 cigarette or less per day between fol low-ups (n - 33; M cigarettes per day - 0.20) and compared them with complete abstainers (n - 145), the total with- drawal discomfort score (sum of the valid withdrawal symptoms see Results section) did not differ between the two groups (10.4 vs. 11.1, p= .38); thus, for this article, subjects who smoked 1 cigarette or less per day were included. An earlier publication (Hughes et al., 1991) discussed the pros aad cons of examining withdrawal effects using cross-sectional analyses of subjects abstinent at each follow-up versus longitudinal analyses of the subset of subjects able to remain abstinent for a long period. In the present studX we chose to do a longitudinal analysis of the 178 subjects who reported abstinence at 2-, 7-, 14-, and 30-day follow-ups, whose abstinence was observer verified, and who had complete data for self- reported withdrawal symptoms. Among these subjects, 110 also bad complete observer-rati.ng data. For data on 90- and 180-day follow-ups, we used the 78 subjects who continued to be abstinent at both of these times as wel l. Expiorntory analyses indicated the incidence and magni- tude of signs and symptoms were slightly higher (typically <I0% and <0.5 units, respectively) using cross-sectional rather than longitudinal amlYses. The withdrawal scores were ordinal and most of the scores were skewed with many zero scores: thus, nonparunetric analyses were of- ten used. Two-tailed p values aie given. Results Yalidity and Reliability A valid symptom was defined as one that was significantly different during abstinence than during smoking. To test this, the precessation and follow-up scores for each symptom were entered into a five-cell (precessation, 2-, 7-, 14-, and 30-day fol- low-up) one-way Friedman's test. After cessation, anxietx diffi- culty concentrating, hunger, irritability, restlessness, and weight gain increased, and heart rate decreased (see Table 1 and Figure 1). Observer ratings confirmed the findings of anxiety, irritability4 and restlessness. Post hoc tests comparing postces- sation scores with precessation scores were performed for those variables for which the Friedman's test was significant. Bon- ferroni adjustments were used to control Type I error. AnxietX difficulty concentrating, irritabilitx and restlessness differed from precessation only at 2 days postce.ssation; heart rate dif- I

I I I I I I I I I I I I I I I I I I TOBACCO WITHDRAWAL IN SELF-QUITTERS - Table I Means and Standard Deviattons of Symptom Scores Among 178 Subjects Abstinent at 2-. 7-. 14-, and 30-Day Follow- Ups Days postcessauon Symptom Precessation 2 7 14 30 Friedman's test Self-ratings (0-3: N - 178) Anxiety M 1.5 2.1• 1.8 1.4 1.2 59.6" SD 1.2 1.0 1.0 1.1 1.0 Depres4on M 1.2 1.1 1.2 1.0 0.8 11.6 SD 1.2 1.1 1.0 1.0 0.9 Difficulty concentrating M 0.9 1.40 1.0 0.7 0.7 32.3"' SD 1.0 1.1 1.1 0.9 0.9 Hunger M 0.7 1.6• 1.6• 1.4• 1.3• 42.9" SD 1.0 1.1 1.2 1.2 1.1 Irritability M 1.6 2.1 • 1.8 1.6 1.3 31.7" SD 1.1 1.1 1.0 1.0 1.1 Poor memory M 0.7 0.8 0.8 0.6 0.6 3.7 SD 1.0 1.0 1.0 0.9 0.9 Restlessness M 1.4 2.0• 1.8 1.4 1.2 56.9° SD 1.2 1.1 0.9 1.1 1.0 Other self-ratings Craving(0-100) M 61.8 57.5 49.2 35.2• 27.6• 111.6" SD 24.6 26.6 27.0 25.6 25.0 No. of nocturnal awakenings M 1.3 1.7 1.3 1.2 1.1 8.7 SD 1.7 1.9 1.8 1.4 1.5 Weight (kg) M 71.8 71.9 72.3' 72.7' 72.9• 43.3" SD 1.5 15.7 15.7 16.1 16.0 Heart rate (bpm) M 82.0 76.8• 75.1' 75.7• 77.6 27.7" SD 13.1 11.3 10.9 11.1 11.3 Alcohol (drinks/day) M 0.2 0.2 0.2 0.2 0.2 0.2 SD 0.7 0.8 1.0 0.7 1.1 Caffeine (drinks/day) M 2.3 2.1 2.3 2.1 2.1 7.2 SD 1.7 1.6 1.9 1.7 1.9 Total withdrawal discomfort score' M 7.4 10.5• 9.0' 7.7 6.6 82.8" SD 4.2 4.2 4.1 4.2 4.0 Observer ntings(0-3 N ~ 78) Anxiety M 1.1 1.7• 1.2 1.0 1.0 19.8" SD 1.0 1.0 1.0 0.9 1.0 Deptession M 0.6 0.8 0.6 0.7 0.6 3.0 SD 0.9 0.9 0.8 0.9 0.8 ~ry M 1.0 1.7• 1.3 1.2 1.3 12.0" SD 1.1 1.1 1.1 1.1 1.1 Restlessness M 1.0 1.6• 1.3 1.2 1.0 19.4'°' SD 0.9 1.0 1.0 0.9 1.0 Nae. Abstinence was de5ned as self-reported 1 ci8aitttt or less per day, not refutad by observer. Kg ~ kilograms; bpm = beats per minute. • ' Calculated as sum of ratings of anxiety, difficulty concentrating, hunger, irritability, resdessness, and nocturnal awakenings. . p<.01. "p<.001. 691 I

~ 3 Anxiety 3-1 Dl/llculty Concentrating 31 Irrllablllly 3-1 Rostl.san.ss ;° 2 a 2 21 a 2 a C C a C « C w ~ w ~OC w ~ "~'----. ¢ 11 r""~--~-----.------ i e 0 ~ ~...... ae 1 1 JJ' 0 ~ , ..,_, Y,,_.. 0 ....ee.. -.-.---y----•--••---.---•~.-.- .,--. _._.._, ~.._ _.- 0 2 7 14 30 90 1ll0 0 2 7 14 30 •0 /00 0 2 7 14 30 90 100 0 271/ 30 90 101 Awakening at NiOht 3 s a Z e a C .w. C .' . 1 ~ w 1R s 0, . -, 0 2 7 14 30 00 10o « 100 . a H.art Rat. 0 ,., ..,_ ,--•----,-..____.._ 0 27 10 3o tlo t0o 027 14 30 •0 /00 0 2 7 14 -- •----• 30 90 180 Total Withdrawal Score Cr.rlno Depression ,i Ca/lsln.led Serttspes 16 100 3 12 ~ w 3 ~. c a 2 °~ -c so ° ' ~ w C ~ , 20 ~ O TTT'T T 0 .-y ---~--..---..-~- 1 ~r'~••~•~ o~ 7 14 lo vo I eo -^~ 0 2 7 11 30 90 100 0•= 7 14 30 00 /00 0 0 2 7 14 30 90 180 Fiaurr 1. Self-reported symptoms prc- and postcessation of tobacco use. (The abscissa on each itraph indicates precessation (0) and days postcessa- tion. Solid circles and solid lines - subjects abstinent for 30 days[rt-17E1; solid circks and dotted lines - subjecta abstinent for 180 days (n - 511; open cincks - current smokers (n - 611; trianj.ks - never smokers [n - 611; open squat+a w bo0-term ex-smokers [n - 561) ~M6E9V99

I I I I I I I I I I I I I I I I I I TOBACCO WITHDRAWAL IN SELF-QUITTERS fered at 2. 7. and 14 days postcessation: hunger at 2. 7.14. and 30 days postcessation, and weight at 7,14, and 30 days postcessa- tion. A total withdrawal discomfort score was calculated by add- ing scores for anxiety, difficulty concentrating, hunger, irritabil- ity, restlessness, and nocturnal awakenings. Post hoc tests indi- cated that the score was significantly greater than precessation at 2- and 7-day follow-ups, did not differ from precessation at 14-day follow-up, and was lower than precessation at 30-day follow-up (see Table 1 and Figure 1). Unexpected}X craving and depression never showed a signifi- cant increase over time, and craving decreased over time. Alco- hol and caffeine intake did not change over time. Although craving and depression were not valid withdrawal symptoms, because they have been of recent interest (Anda et at., 1990; Glassman et al, 1990; West & Schneider, 1987), they are dis- cussed at certain points in this article. Interrater agreement between the 110 pairs of complete sub- jects and observers was determined using the 2-day follow-up scores. To assess agreement across the range of scores, Spear- man rank order correlations were used. To assess agreement on whether a score did or did not increase with cessation, kappa coefficients were used. Kappas were used in preference to chi- square tests as the former corrects for differences in base rates. Neither the correlations nor the kappas were large (rs <.30, Ks <.22). Although subjects tended to rate themselves as hav- ing a greater symptom intensity than did observe=s, this differ- ence was not statistically significant. Table 2 f>srcentage of Subjects Whose Symptom Scores Were Greater at 2-, 7-, 14-, and 30-Day Follow-Ups Than at Precessation Among Subjects Abstinent For 30 Days Days postoeaation Symptom• 2 7 14 30 Self-ratings (N - 178) Anxiety 49 37 30 27 Craving 37 24 17 16 Decreased heart rate (a 1 bpm) 61 64 65 37 Deptnssion 31 38 28 22 Difficulty concenttatina 43 30 20 20 Hunger 53 51 47 45 Irritability 38 38 31 24 Nocturnal a.vakeaings (2 l/night) 39 31 33 27 Restlessness 46 39 35 25 Decreased heart rate (>2 bpm) 61 64 65 5E WeiEht gain (Z 1 lb or 0.45 kg) 30 53 56 56 Increased affeine (z 1 drink/day) 25 28 27 24 Increased alcohol (x 1 drink/day) 3 7 4, 4 Observer ratings (n - 78r Anxiety 58 41 29 26 Deptession 38 29 33 28 Irritability 56 39 40 37 Restlessaess 61 45 41 32 Note. Abstinenct was defined as self-reported I cigatette or less per daX not refuted by observer. Symptoms and follow-ups with percent- ages in boldface are those whose mean scores differed from ptscessa- tion means. Lb - pound; kg - kilogramx bpm - beats per minute. 'Ml are rated 0-3 unless otherwise noted. Incidence, Magnuude, and Y'ariability 693 The peak increase in most self-reported symptoms averaged from 0.5 to 0.9 of a possible 3 units (Table 1). The magnitude of the increase in symptoms after cessation was similar across observer- and self-ratings. The greatest decline in mean heart rate was 5.2 beats per minute (bpm) at 14-day follow-up. The greatest increase in mean awakenings at night was 0.5 occur- rences at 2-day follow-up. Incidence was calculated for valid symptoms as the percent- age of subjects whose follow-up score was greater than his or her precessation score by at least 1 unit for the symptoms, by I awakening for insomnia, by 2 bpm for heart rate, and by 1 pound (0.45 kg) for weight gain. Incidences for the valid symp- toms ranged from 37 to 62% across the follow-ups (see Table 2). Forty-nine percent of abstainers fulfilled Diagnostic and Statu- tical Manual ojMenta! Disorders (rev 3rd ed4 DSM-III-R) crite- ria for nicotine withdrawal (American Psychiatric Association, 1987) at 2 days postcasation, 37% at 7 days, 28% at 14 days, and 18% at 30 days postcessation (The DSM-111-R criteria of occur- reace within 24 hours was assumed to be present in all subjects.) Relationship Among Signs and Symptoms Interrelationships among symptoms was examined at 7 days postcessation, because this was when the largest number of effects occurred. At this point, increases in anxietX difficulty concentrating, irritabilitx and restlessness were intercorrelated (rs =.46-68). Changes in heart rate, hunger, weight, and in- somnia were not highly correlated among themselves or with changes in other withdrawal variables (rs <.26). Similar interre- lationships occurred at 2- and 14-day follow-ups. Factor analysis of withdrawal at 7 days postcessation used a principle-components analysis for initial factor extraction fol- lowed by a varimax rotation. This analysis indicated three fac- tors: Mood (anxietyy, difficulty concentrating, irritabilitX and restlessness); Appetite (hunger and weight gain); and Insomnia (awakenings at night). The Mood factor explained 39% of the variance. The Appetite and Insomnia factors each added 12%; thus, a total of 63% of the variance could be explained. Increases in craving were correlated with increases in the other mood symptoms and depression (rs =.34-50) but not with changes in heart rate or weight. Increases in depression were correlated with increases in irritability and anxiety (rs = .41-45) but not with changes in other symptoms. Time Course Anxiety difficulty concentrating, irritabilitX restlessness, and nocturnal awakening showed a time-limited increase (see Figure 1). By 30 days postcessation, these had returned to pre- cessation levels and were similar to values for the control groups. Hunger showed a similar time-limited effect but re- mained elevated at 30 days postcessation. Weight showed a grad- ual increase toward weights similar to those of long-term ex smokess. Heart rate showed a decrease at 2-, 7-, and 14-day follow-ups. Although heart rate appeared to increase again at 30- and 180-day follow-ups, whether the increase was real is unclear. - I

I I I I I I I I I I I I I I I I I 694 71 0. 9• 14 Long-Term Svmptoms In the 78 subjects who remained abstinent for 180 days, weight remained an average of 3.2 pounds above precessation values at 180 days postcessation (Wilcoxon signed ranks test, p<.04). and 39% of these abstinent smokers had gained 5 or more pounds since cessation. Craving persisted but declined over time and did not approach that of never- and ex-smokers (i.e., zero ratings) until 180 days postcessation. At 180-day fol- low-up, 51 % of the abstinent smokers reported some craving for tobacco in the past 24 hours. Determinants of Abstinence Effects The precessation items of sex, education, occupation, mari- tal status, income, cigarettes smoked per dax nicotine yield of cigarette, duration of smoking, total and item scores for the Fagerstrom Tolerance Questionnaire, and baseline symptom scores were correlated with (a) the total withdrawal (i.e., change) score at 7-day follow-up, (b) total withdrawal score at 30-day follow-up (to examine duration of withdrawal), (c) craving at 7- and 30-day follow-ups, and (d) weight gain at 30- and 180-day follow-ups. The only substantial covariation (e.g., r>.20) was By 2-day Depression Scor. ,a, JOHN R. HUGHES 1N By 14-day Depression Score IN, a M Ma m m . O t•• that precessation scores of the individual signs and symptoms were inversely related to the increase in scores at 7 days postab- stinence (rs =-.49 to -.69, ps < .0 1). The one exception was weight (r = -.12, ns). Abstinence Effects and Smoking Cessation The ability of the individual withdrawal signs and symptoms and the total scores to predict abstinence at subsequent follow- ups was tested. Although some of the withdrawal symptoms predicted outcome at some follow-ups, none consistently pre• dicted outcome across follow-ups in a meaningful manner. Al- though not a valid withdrawal symptom, depression did consis- tently predict abstinence across follow-ups (see Figure 2). To illustrate this, at each follow-up, abstinent smokers were di- vided into those whose depression scores had increased post- cessation and those whose follow-up scores were unchanged or had decreased since precessation. I then compared survival curves between these two groups. Subjects whose depression score increased at 2-day follow-up were more likely to later relapse than those whose depression score remained the same or decreased at 7-day follow-up (log rank X2 (1, N= 290) = 6.8, p < .009). Similar effects occurred for the increase in depres- By 7-day Depression Scor. IN 14 m 1N By 30-day Depression Scor. ta, M .i M M M Days Post-Cessatton ts• e•.P-+ a r.«.dM ---fl-- rW...w .•.4.mr r pfew•Mw I Ffgure 2. Depression scores in abstinent smokers who later relapsed or remained abstinent.

I I I I I I I I I I I I I I I I I TOBACCO WITHDRAWAL IN SELF-QUITTERS sion at 7-day (X' (1, 236) = 4.0. p < .05 ), 14-day (X2 (1. ,V = 198)=6.0,p<.01),and30-day(X2(1,N=171)=3.9,p<.05) follow-ups. However, the difference in the relative rates of re- lapse in subjects who did and did not have increases in depres- sion was small (10-15%). When the few subjects with more ex- treme increases in depression scores (increased 2 or 3 units; n= 40) were examined, the ability of depression to predict relapse did not increase substantially. When absolute precessation de- pression scores or absolute postcessation depression scores were used (instead of change scores), a similar but again less robust predictive effect occurred. Craving, weight gain, and total withdrawal discomfort did not consistently predict re- lapse. Discussion Anxietx difficulty concentrating, hunger, irritabilitX rest- lessness, weight gain, and decreased heart rate were valid signs and symptoms of nicotine withdrawal in self-quitters. In addi- tion, awakening at night showed a nonsignificant increase post- cessation. The increases in anxietx irritabilitX and restlessness were confirmed by observer ratings. Craving, depression, im- paired memory; and changes in alcohol or caffeine intake were not verified as abstinence effects in this sample. Prior prospective studies of nicotine withdrawal have used the 5% of smokers who attend formal smoking cessation pro- grams'(Hughes et al, 1990). The one exception is a study that followed self-quitters at 1, 2, 7, 30,180, and 365 days postcessa- tion (Gritz et af, 1991). Valid symptoms in the present study correspond well with the seven most prevalent symptoms in this prior study (i.e., anxiety or tension, difficulty concentrat- ing, excessive hunger, eating more than usual, irritability or anger, restlessness, and impatience). The prior study used abso- lute postcessation scores rather than diffetencx scores (absti- nence minus precessation) and cross-sectional rather than lon- gitudinal analyses. However, if the present data are reanalyzed using postcessation scores and cross-sectional analyses, inci- dence rates similar to this prior study are obtained. The present results are also comparable with two prior stud- ies that differed in subjects and settings but used very similar experimental procedures, withdrawal scales, follow-ups, and data analysis (Hughes et a1.,1991: Hughes & Hauukami,1986). The major differences were that in the prior studies subjects were volunteers for double-blind studies of nicotine gum, were randomized to placebo gum, and completed mailed question- naires rather than tdephone interviews. One study consisted of 50 volunteers wbo attended a behavior therapy cessation pro- gtam (Hughes & Hatsukami,1986). The other study examined 105 general medical practice outpatients who received only brief advice regarding quitting (Hughes et a1.,1991). W hen simi- lar time points are compared, withdrawal symptoms appear to be slightly less prevalent (usually < 10% difference in incidence) and less intense (usually <0.5 units in change scores) in the present study than in the first study of program attendees. Re- sults for the medical outpatients study fell between those of the present study and those of program attendees. As in prior studies of tobacco withdrawal, most of the signs and symptoms in the present study showed a time-limited course in which symptoms first increased and then decreased (i.e., a pattern consistent with a drug withdrawal symptom: Hughes et al, 1990). Depending on the symptom. the mean scores returned to precessation scores between 7 and 30 days postcessation (see Figure 1). In the prior study of self-quitters (Gritz et al., 1991), withdrawal was measured at 7 and 30 days. By 30-day follow-up, most seemed to stabilize. Other studies of smokers in cessation programs found that most withdrawal symptoms declined to precessation levels by 21 to 30 days post- cessation (Gross & Stitzer.1989; Hughes et al.. 199 I: West. Ha- jek, & Belcher,1989b). In summarr; the mean duration of nico- tine withdrawal appears to be 2 to 4 weeks. Heart rate, hunger, and weight gain appeared to be the excep- tions to this time coutse. In prior studies, whether heart rate showed a time-limited, biphasic "withdrawal" pattern or a sim- ple uniphasic decline postcessation is debatable (Nemeth-Cos- lett, Sampson, & Henningfield, 1986; Schneider & Jarvik,1985; Ward, Garvey4 & Bliss, 1992; West & Schneider, 1988). Unfortu- naiel}4 our results (see Figure 1) do not help resolve the time pattern for heart rate postcessation. Hunger appeared to exhibit a prolonged but still time-ii- mited pattern. This result is consistent with the prior study of self-quitters (Gritz et a1,1991) and the prior study of medical outpatients (Hughes et a1.,1991); however, other studies of pro- gram attendees have found that hunger persists unabated for 4 weeks (West, Hajek, & Belcher, 1987,1989b) to 10 weeks (Gross & Stitzer,1989). Weight gain continued throughout the study until at 6 months the mean weight of abstinent smokers appeared to be similar to that of ex smokers but greater than that of never- smokers or current smokers. The mean weight gain of 3.2 pounds (1.4 kg) over 6 months is similar to that reported in prior studies of self-quittets (Klesges, Meyers, Klesges, & La- Vasque, 1989). The uniphasic pattern of weight gain suggests that weight gain is due to a simple loss of nicotine's weight-sup- pressing effects (Grunberg, 1990; Klesges et al. 1989) rather than to a withdrawal effect. On the other hand, the rate of weight gain appeared to be greatest in the first 2 weeks and decreased thereafter, suggesting a time-limited (i.e., with- drawal-like) potentiation of weight gain. Craving did not increase after abstinence. As in prior studies (Hughes et al, 1991; Hughes & Hatsukami, 1986), craving rat- ings were high precessation: however, the present results do not appear to be due to ceiling effects as the mean precessation score was 62 on a 0- to 100-point visual analog scale. These negative results could be due to the use of the phrase "craving for cigarettes" rather than "desire for a cigarette" (Kozlowski, Mann, Wilkinson, & Poulos, 1989) or because the term "crav- ing" has a different meaning to smokers during smoking than during abstinence (West & Schneider, 1987). Whatever the rea- son, in this studx craving failed to show a time-limited increase characteristic of withdrawal phenomena. Protracted withdrawal states have been described for alcohol (DeSoto, O'Donnell, Allred, & Lopes, 1985) and opiates (Mar- tin & Jasinski, 1969). Most of the symptoms of nicotine with- drawal showed no evidence of protracted withdrawal. However, hunger took 3 months to return to baseline, weight gain contin- ued for 6 months, and craving, although greatly diminished, persisted in many subjects for 6 months. As in prior studies (Hughes et al, 1991; Hughes & Hatsu- I

I I I I I I I I I I I I I I I I I I 696 JOHN R. HUGHES kami. 1986). observer ratings were not highly correlated with self-ratings yet both proved sensitive to abstinence effects. One possible interpretation of this pattern is that observer and self- ratings tap different but still valid sets of behaviors. Several investigators have stated that nicotine withdrawal has a large between-subjects variability. Comparison of the variabil- ity in symptomatology in the present study with that found in studies of alcohol or drug withdrawal is difficult given the dif- ferent scales used and the different populations studied (i.e., self-quitters vs. those who attend a clinic). Thus, whether nico- tine withdrawal is more variable than other drug withdrawal syndromes is unclear. It was hoped that with a larger and more heterogenous popu- lation than used in prior studies, correlates of individual differ- ences in nicotine withdrawal could better be detected. How- ever, as in prior studies (Hughes et al, 1991; Hughes & Hatsu- kami, 1986), the study was not able to replicate the finding that the Fagerstrom score predicts more severe withdrawal (Fager- strom & Schneider,1989). In addition, as in most prior studies (e.g, Hughes et al, 1990), withdrawal severity did not differ between men and women or between heavy and light smokers. A major hypothesis of the nicotine dependence theory is that nicotine withdrawal prevents abstinence. If this is true, then subjects with more severe withdrawal should have a greater probability of relapse. Although some earlier trials have not confirmed this relationship (Hughes et a1,1991; Hughes & Hat- sukami. 1986). other trials have (US. Department of Health and Human Services [USDHHS], 1990). The robustness and consistency of the predictive power of withdrawal varies sub- stantially across these studies and across follow-ups within the studies (USDHHS.1990). It was hoped that the largerand more heterogenous sample would more likely detect a withdrawal- abstinence relationship; however, none of the valid withdrawal symptoms consistently predicted outcome. Subjects with in- creases in depression were more likely to relapse- This result replicates and extends the results of two prior studies (Covex Glassman. & Stetner,1990; West, Hajek, & Belcher,1989a) that showed that depression predicted short-term (s4 weeks) absti- nence. Although the present study replicated prior work indicating that weight gain does not increase the probability of relapse (Gross. Stitzer. & Maldanodo,1989; Killen, Fortmann, & New- man. 1990), it failed to replicate the prior findings of others (Hall, Ginsberg, & Jones, 1986; Killen, Fortmann, Newman, & Varad}, 1990) and ourselves (Hughes et al, 1991) that weight gain decreases the probability of relapse. One possible reason for the present failure to replicate this latta finding is that the prior studies used program attendees who gained more weight and were probably more dependent on nicotine than the self- quitters in this studyc The present study had several assets that increased its valid- ity and generalizability: (a) a large sample, (b) a sample of self- quitters, (c) few inclusion criteria. (d) several control groups, (e) previously validated measures, (f) scales sensitive to only large changes in symptoms, (g) withdrawal scores calculated as changes from before cessation, and (h) observer ratings. On the other hand, the study may have suffered from exclu- sion bias (i.e, perhaps subjects with the most severe withdrawal returned to smoking and were excluded from the analysis). If this were true. the present study may have underestimated the magnitude of withdrawal. The way to prevent exclusion bias is to experimentally control abstinence (e.g., through large mone- tary payments): however, this might jeopardize the generaliz- ability of the findings. Exclusion bias is probably unlikely given that subjects with more severe withdrawal were not more likely to relapse in the study. Also, these results may not be generaliz- able to several groups (i.e., adolescents and subjects who stopped smoking by gradual reduction, who were involved in any treatment, or who reported low motivation for cessation were excluded). In summary, the results of this study provide new and confir- matory evidence about nicotine withdrawal; however, the nega- tive results suggest several questions still to be answered: What are the determinants of severity of withdrawal? If withdrawal does not increase the probability of relapse, what does cause early rel.apse? Nicotine replacement is believed to work by re- ducing withdrawal (Fagerstrom,1988), yet withdrawal appears to have little effect on abstinence; thus, how does nicotine re- placement increase abstinence? References American Psychiatric Association. (1987). Diagnostic and statistical manual of inental disorders (3rd ed_ rev.). Washington, DC: Autbor. Anda, R. F Williamson, D. F Escobedo, L. G, Mast, E. E, Giovino, G. A, & Remington, P. L (1990). Depression and the dynamics of smoking: A national petspectm. Journal of the American Medical A.tsociatton. 264, 1541-1545. Cohen, S, Lichtenstein, E, Prochaska, J. O, Rossi, J. S, Gritz, E. R, Carr, C. R, Orleans, C T Schoenbach, V J, Biener. L, Abnms, Q, DeClemente, C, Currx S., Martatt, G. A, Cumming, K. M, Emont, S. L, Giovino, G, & Ossip-Klein, D. (1989). Debunking mythsabout self-cluitting. American Psychologist, 44. 1355-1365. Covey, L S, Glassman. A. H.. & Stetner, F. (1990). Depression and depressive symptoms in smoking cessation. Comprehensive Psychia- tryc 31. 350-354. DeSoto. C. B, O'Donnell, W E, Allred. L. J_ & Lopes, C. E(1985). Symptomatology in alcoholics at various stages ofabstinence. Alco- holism: Clinical and Ezpenmerual Research. 9, 505-512. Fagerstrom, K-Q (1988). Efficacy of nicotine chewing gum: A review In Q F. Pomerleau, C. S. Pomerleau. K-Q Fagerstrom. J. E. Henning- field, & J. R. Hughes (Eds.). Nicotine replacement: A critical evalua- tion (pp. 109-128). New York: Alan R. Liss. Fagetstrom, K-Q, & Schneider. N. G. (1989). Measuring nicotine de- pendence: A review of the Fagerstrom Tolerance Questionnaire. Journal of Behavioral Medicine. 12. 159-182. Fore, M. C, Novomy, T. E, Pierce, J. P Giovino. G. A_ Hatziandreu. E J, Newcomb, P. A, Surawicz, Z S.. & Davis, R. M. (1990). Meth- ods used to quit smoking in the United States. Journal ofthe Ameri- can Medical Association, 263, 2760-2765. Glassman, A. H, Helzer, J. E, Covex L. S, Cottler, L. B, Stetner. F Tipp, J. E., & Johnson, J. (1990). Smoking, smoking cessation, and major depression. Journal of the American MedicalAssocrarion. 264, 1546-1549. Goldstein, A. (1987). Criteria of a pharmacologic withdrawal syn- drome. Arrhives of Gentral Psychiatry 44, 392. Gritz, E. R, Carr, C. R, & Marcus. A. C(1991). The tobacco with- drawal syndrome in unaided quitters. British Journal of Addiction, 86. 57-69. Gross. J, & Stitzer, M. L. (1989). Nicotine replacement: Ten-week ef-

I I I I I I I I I I I I I I I I I I I TOBACCO wtTHDRAWAL IN SELF-QUITTERS fects on tobacco withdrawal symptoms. Psrchopharmaco/ogi: 98, 331-334. Gross. J.. Stitzer, M. L., & Maldanodo. J. (1989). Nicotine replacement: Effects on postcessation weight gain. Journal oiConsulting and Clin- cal Psvcnologt: S', 87-92. Grunberg. `. E. (1990). The inverse relationship between tobacco use and body waght. In L. T Kozlowski. H. Annts. H. D. Cappell. F B. Glaser, M. Goodstadt, Y. Israel. H. Kalant. E. M. Sellers. & E. R. Vingilis (Eds). Research advances in alcohol and drug problems (Vol. 10, pp. 273-316). New York: Plenum Press. Hall. S. M., Ginsberg, D. & Jones, R. T. (1986). Smoking cessation and weight gain. Journal of Consulting and Clinical Psychology, 54. 342- 346. Hughes, J. R.. Gulliver. S. B, Fenwick, J. W., Cntser, K.. Valliere, W. A, Pepper. S. L" Shea, P. J, & Solomon, L. J. (in press). Smoking cessa- tion among self-quitters. Health Psvchology Hughes, J. R_ Gust, S. W, Skoog, K, Keenan, R., & Fenwick, J. W. (1991). Symptoms of tobacco withdrawal: A replication and exten- sion. Archives of General Psychiatry 48, 52-59. Hughes, J. R., & Hatsukami, D. K. (1986). Signs and symptoms of tobacco withdrawal. Archives of General Psychiatry, 43, 289-294, Hughes, J. R_ Hatsukami, D K., Pickens, R. W„ Krahn, D., Malin, S„& Luknic. A. (1984). Effect of nicotine on the tobacco withdrawal syn- drome. Psvchopharmacology, 83, 82-87. Hughes. J. R_ Hatsukami, D. K., Pickens, R. W, & Svikis, Q S. (1984). Consistency of the tobacco withdrawal syndrome, Addictive Behav- ion, 9, 409-412. Hughes, J. R_ Hatsukami, D. K.. & Skoog, K. (1986). Physical depen- dence on nicotine gum: A placebo-substitution trial. Journal of the American Medical Association, 255, 3277-3279. Hughes, J. R_ Higgins, S. T& Hatsukarai, D. K. (1990). Effeccs of abstinence from tobacco: A critical teview. In L. T. Kozlowski, H. Annis, H. D. Cappeli, F Glaser, M. Goodstadt, Y, Israel, H. Kalant, E. M. Sellers. & J. Vingilis (Eds,), Research advances in alcohol and drug problems (Vol. 10, pp. 317-398). New York: Plenum Press. Hughes. J. R_ & Kottke, T. (1986). Doctors helping smokers: Real world tactics. Minnesota Medicine, 69. 293-295. Jarvis, M. J.. Tunstall-Pedoe, H, Feyerabend, C, Vesey, C„ dc Saloojee, Y. (1987). Comparison of tests used to distinguish smokers from non-smoken, Amencan Journal of Public Health. 77,1435-1438. Killen, J. D" Fortmann, S. P& Newman, B. (1990). Weight change among participants in a large sample minimal contact smoking re- lapse prevention trial. Addictive Behaviors,15. 323-332. Killen. J. D Fortmann. S. P Newtnan. B_ & Varadx A. (1990). Evalua- tion of a treatment approach combining nicotine gum with self- guided behavioral treatments for smoking relapse prevention. Jotu- na1 of Consulting and Clinical Psychology 58, 85-92. Klesges, R. C., Meyers, A. W, Klesges, L. M, & LaVasque, M. E. (1989). Smoking, body weight, and their effects on smoking behavior A comprehensive review of the literature. Psychological Bulletin, 106, 204-230. 697 Kozlowski. L. T.. Mann. R. E.. Wilkinson. D. A.. & Poulos. C. X (1989). "Cravings" are amtbtguous: Ask about urges or desires. .4d- dictive Behaviors, 14, 443-1345. Mangan, G. L" & Golding. J. F (1984). The psychopharmacology of smoking. Cambridge, England: Cambndge University Press. Martin, W R.. & Jasinski, D. R. (1969). Physiological parameters of morphine dependence in man. Tolerance, early abstinence and pro- tracted abstinence. Journal of Psychiatric Research. 7. 9-17. Nemeth-Coslett, R, Sampson, A, & Henningfield. J. E. (1986. Au- gust). Subjective and physiological correlates of cigarette deprivation. Paper presented at the 94th Annual Convention of the American Psychological Association, Washington, DC. Pechacek, T, Arken, R, & Johnson, N. J. (1982). Quit and win. Minne- apolis. MN: Minnesota Heart Health Program. Russell, M. A. H., Peto, J, & Patel, U. A. (1974). The classification of smoking by factoral structure of motives. Journal ofthe Royal Statts- tieal Society: Series A. 137, 313-333. Schneider, N. G, dt Jarvik, M. E. (1985). Nicotine gum vs. placebo gum: Comparisons of withdrawal symptoms and success rates. ln J. Grabowski & S. M. Hall (Eds.), Pharmacological adjuncts in smoking cessation (National Institute on Drug Abuse Research Monograph 53; pp. 83-101: DHHS Pub. No. ADM 85-1333). Washington, DC: US. Government Printing Office. U S. Department of Health and Human Services (1990). Psychological and behavioral consequences and correlates of smoking cessation. In Health consequences of smoking cessation: A report of the US. Sur- geon General (pp. 521-531). Washington, DC: U.S. Government Printing O9'ix. Ward, K. D, GarveX A. J, & Bliss, R. E(1992). Evidence of transient heart rate change after smoking cessation. Psyritopharmacologyt 106, 337-340. West, R. J, Hajek, P, & Belcher, M. (1987). Time course of cigarette withdrawal symptoms during four weeks of treatment with nicotine gum. Addictive Behaviors. 11, 1-5. West, R. J, Hajek, P. & Belcher, M. (1989a). Severity of withdrawal symptoms as s predictor of outcome of an attempt to quit smoking. Psychological Medicine. 19, 981-985, West, R. J Hajek. P& Belcher, M. (1989b). Time course of cigarette withdrawal symptoms. Psychopharmacology 99, 143-145. West, R. J, dt Russell, M. A. H. (1985). Effects of withdrawal from long-term nicotine gum use. Psychological MedicinG 15, 891-893. West. R. J, &Schneider, N. (1987). Craving for cigarettes. British Jour- nal of Addiction, 82, 407-415. West, R. J, & Schneider, N. (1988). Drop in heart rate following smok- ing cessation may be permanent. Psychopharmacologyt 94, 566- 568. Received June 25,1991 Revision t ece>`+ed August 28, 1991 Accepted January 20,1992