Tobacco Documents Online

I I I I I I I I I I I I I I I I I Originai Contributlons Predicting Smoking Cessation - Who Will Quit With and Without the Nicotine Patch Susan L. Kenford, Michael C. Fiore, MD, MPH; Douglas E. Jorenby, PhD; Stevens S, Smith, PhD; Oavid Wetter, MS; Timothy B. Baker, PhD ObJective -1'o identify predictors of smoking cessation success or failure with and without transdermai nicotine patch treatment. Design --Two Independent randomized, double-blind, placebo-controlled stud- ies using the nicotine patch assessing outcome at the end of treatment and at 6-month follow-up; each study used a different mode of adjuvant counseling. Patlentr: -Subjects were daily smokers (> 15 cigarettes per day), aged 21 to 85 years with expired air carbon monoxide levels of at least 10 ppm, and motivated to quit. Eighty-eight subjects participated In study 1, and 112 subjects participated in study 2. interventlon.-Study I consisted of 8 weeks of 22-mg nicotine patch therapy with Intensive group counseling, Study 2 oonsisted of 4 weeks of 22-mg nicotine patch therapy and 2 weeks of 11-mg nicotine patch therapy with brief Individuai counseling. Main Outcome Meaaures.-The prediction of smoking cessation (at end of treatment and after 6 months) based on pretreatment and intratreatment measures in smokers using active or ptacebo nicotine patches. Resutta.-Pretreatment rnarker9, such as the Fagerstrom Tolerance auestlon- naire score, numberof cigarettes smoked per day, years smoked, expired aircarbon monoxide levei, or baseline blood nicotine and cotinine levels, showed no consistent reiationshipwtth successful smoking oessation across both studies. Of the Intratreat- ment markers examined, withdrawal aeverity and nicotine replacement levels also were not consistently predictive of cessation success, However, any smoking during the second week of treatment was a consistent and powerful predictor of failure at the end of treatment and after'6 months. Among active nicotine patch patients who smoked at all during week 2 after quitting, 83% and 9711/6 (studies 1 and 2, respec- tively) were smoking at 6-montt, follow-up, Conversely, abstinence during the sec- ond week of treatment predicted successful smoking cessation. Among active nico- tine patch patients who were totally abstinent durin9 week 2 after quitting, 46% and 41 % (studies I and 2, respecttvely) were abstinent at 6-month follow-up. Of all nico- Gne patch patients In both studies who were smoking at 8-month follow-up, 74% be- gan smoking during week 1 or 2. Among all ptacebo patch patients who were smok- ing at 6-month foliow-up, 86% began smoking durtng week I or 2. Conclushsns,-Smoking status (abstlnent or smoking) during the first 2 weeks of nicotine patch therapy, particularty week 2, was highly oorrelated with clinicat out- come and can serve as a powerful predictor of smoking cessation. Early smoking behavior also predicted outcome among placebo patch users. Traditional measures of dependence are not consistently predictive of cessation suceess. Ciiniclans are advised to emphasize the Importance of total abstinence after a quit attempt and to follow-up with patients within the first 2 weeks of quitting; smoking during this or'itlcal time should be assessed and treatment may be altered as appropriate. (JANA. 186{;.°71:689-SG+) From /ha C.ntN k~, TNbaCeo PlNKOh and Intar• vennon (M. Y..nrord. Dre F:or., Joranby, SmNn, and Saw, and Mr war.a) and th. Saction ot G.n.ral tnt,arnat Madlcme. Oeparvnent of Medicina (Ors Fora end Joranby), UNwufIN o( Wleeonain Madical Schood, end tha Univeraty o/ W aooM(n ComprMA"iw Can- cwr Center (Dra Ftryre, JorenOy. and 6akar), Maoyon .nd oepartrnKa M PtychobW, Unw.r.lty of Wi.eon• .in. Madieon (M. Kanto~tl. Ore Sntitn and 8aksr, and Mr Wartu), lks Fara, Jorertby, StNttt. und 8aker, M. KaNOrd, ano Mr Watur haw eonduetetl reaearen fcmdad in pan by CIBA-GEIGY Coro, Ed~fon, NJ: 'e;an ?narmac.utlui. Ltd, Athbna, Irel.nd; and La6.(N I.aqOratorlaW. W.ynM. NJ. of Ffora nas afso rocoivad Itonorarla fo( aduoauonal,scUwWa from CIBA-GEtGY Coro, Laderta Labaalorree. Mar{on MarraU Dow Inc, Kaneae Gty, Mo, and Parke-OavN. MorrN Alatna. NJ. R.pmt reqiKtf to Cantar for Tooacco Roeaarch ana Intorvor+don, UnKvNt?ry of wifooneln M.dicai Seeioot, 7275 Mao-eal Seianees Cantar, 1300 UNvar- aay Aw. Madison, Wi 53706-1532 (0( Fqr*). JAMA, February 23. 1p94-vo( 271, No. 6 h UbIE ROUS clinical trials have shown that the transdermal nicotine patch sye- tem helps people quit smoking.' Quit rates, however, vary• Mdely across these trials. This may be due in part to the marked heterogeneity of smokers in the United Statee today, If one could iden- tify those smokers likely to succeed or fail with or without nicotine patch treat- ment, it would be posaible to make treat- ment decisions on a more rational basis and increase each patient's odds of quit- ting successfully. For example, smokers judged to be at risk for failure might be given a more potent adjuvant treatment, such as in- tensive group counseling, to boost treat- ment effectiveneas.- Other smokers might be switched to a wholly different treatment, such as nfcotine fadtng or averaive arnoking.$ Despite much research, accurate and consistent predictors of successful smok- ing cessation have not been identified. Variables that have proven predictive in some studies do not predict in others, and when predictive relations are found, they are raodeat. For example, gender has been found to predict treatment suc- cess in some atudies" but not in an- other.' Similarly, measures that are hy- pothesized to reflect nicotine dependence (eg, number of cigarettes smoked per day, blood nicotine levels, and expired air carbon monoxide (CO.] levels) have predicted treatment success weakly and inconsistently.'a One reported measure of nicotine dependence, the Fageretrom Tolerance Questionnaire' has predicted long-term success in some studiese•la but has not been consistently related to out- come among nicotine patch users."a=The inconsistency in predictive relations aug- gest.a that the accurac,y and sensitivity of any given predictor is affected by factors that vary from study to study, such as the target population and type of eeasation treatment. We examined predictors of treatment succeaa and failure among smokers with and without the nicotine patch. The re- search included two studies with two independent samples of smokers. More- over, two different adjuvants were used SnOfdng Cessation Nhitt and Witt'tout NiGOGne Patch-Kenford at at 6ae I

I I I I I I I I I I I I I I I I I I with nicotine patch therapy and placebo patch therapyt (1) intensive group coun- seling and (2) brief indMduai counsel- ing. Independent samplec and different types of adjuvants provided the oppor- tunity to assess the generalizabtlity of our findings. To identify predictors of successful cessation with and u'ithout the nicotine patch, we assesapd a number of vari- ables that could l.w quickly and easily measured by a clinician and therefore would have broad clinical utility, We evaluated measures collected prior to quitting (pretreatment variables) and early in the quitting attempt (intratreat- ment variables). METHODS Two studies using nicotine patch therapy and two different modes of ad- juvant therapy were conducted sequen- tially. In study 1, patch therapy was paired with state-of-the-art intensive group counseling; in study 2, patch therapy was paired with brief individual counseling that could be conducted in a clinician's office. The methods used in both studies and efficacy results are de- scribed in more detail elsewhere.' SubJects Subjects were recruited through me- dia announcements in Madison, Wis, a city of 190 700 residents. Inclusion crite- ria consisted of age 21-to 65 years, a history of smoldng at least 15 cigarettes per day during the past year, expired C0, level of at least 10 ppm (as deter- mined at screening), and motivation to quit smoking. Exclusion criteria included presence of cardiovasculardisease, preg- nancy or lactation, regular use of pay- chotropic druRs, current symptomatic psychiatric disorder, alcohol or other drug abuse, chronic dermatologic disorders, and/or use of any experimental medica- tion within the paet 30 days. Eighty-eight subjects participated in study 1, and 112 subjects participated in study 2. Study Dealqna After enrollment, the subjects and the clinician agreed on a quit date. Subjects were told to stop amoking on this date at which time nicotine patch and adjuvant counseling treatments commenced. On the quit date and once per week during the treatment period the following data were gathered at the clinic site: self-re- ported smoking status for the previous week (confirmed by expired C0, assay). ~_uital signs, and questionnaire and study drug information. Between visits, sub- jects kept a smoking diary each day that contained an eight-item withdrawal sur- vey1d Se+vm samples for nicotine and cotinine were coUected at a screening 640 JA1rW, February 23. 14Gd-Vof 27t. No. 8 visit before the quit date and at 4 weeks following the quit date. Both studies w ere double blinded and placebo controlled. Study 1 consisted of 8 weeks of 22-mg nicotine patch therapy (or equivalCnt placebo patch), and study'L consisted of 6 weeks of nicotine patch therapy--t weeks of'2'2-mg patch therapy followed by 2 weeks of 11-mg patch therapy (or placebo). In both studies, eounselingses- sions occurred for 8 weeks, and subjects were contacted at 6 months following their quit dates to determine their smok- ing status. Self-report was confirmed by expired C0, assay. Si.e-month follow-up data were used to assess predictors of success and fail- ure becauae the nicotine patch was li- c:ensed by the Food and Drug Admin- istration at this end point of study 1, and we wanted to prevent confounding of results due to the extraordinarily high rates of off-study prescription patch use immediately after licensing. Trsnsdermal Nlcotino The transdelTnal nkotine deli z'ery sys- tem (PROSTEP, Lederle Laboratories, Wayne, NJ) used in these studies was a hydrogel matrix reservoir containing niccr tine. The patch was applied once a day, worn for 24 hours, and delivered a total absorbed dose of either ?2 mg or 11 mg of nicotine. Placebo patches contained no nicotine. Active patches and placebo patches were supplied by Elan Pharma- ceuticals Ltd, Athlone, Ireland. While a comparison of phannacoldnetic proper- ties shows differences between the dif- ferent brands of patches across the initial days of treatment, wzthin 3 days all brands achieve comparable nicotine levels and maintain these throughout treatment. Adjuvant Tr+eatmenc In the group counseling intervention used in study 1, groups of eight to 12 members met for approximately 60 min- utes per week for 8 weeks. Sessions were designed to provide information in a sup- portive atmosphere and teach coping skiUs appropriate to the subjects' place in the quitting process. This was de- signed to be a high-intensity counseling intervention. Individual counseling in study 2 consisted of weekly meetings for 8 weeks with individual subjects. Sessions lasted 10 to 20 minutes and targeted topics relevant to the individu- al's place in the quit process. This coun- seLing waa designed to be a moderate- intensity counseling intervention. Both counselinginterventie+ts stressed the importance of abstinence, along with the identification of high-risk situations and potential coping behaviors. Both counseling interventions were conducted by psychologists or psychology graduate students who followed a treatment manual developed for each study. Statistlcal Analysea Baseline subject characteristics in the two groups (active patch vs placebo patch) were evaluated wzth two-tailed independent-group t testy for continu- ous-level dependent variables; k= tests of independence were computed for categorical outcome variables. For con- tinuous variables displaying significant heterogeneity of variance, separate-vari- ance independent-group t tests were computed. Efficacy was tested with XL tests of independence, The relations among pretreatment measures of nico- tine dependence and intratreatment measures (blood nicotine and cotinine levels and withdrawal severity) were evaluated with Pearson product-moment correlations. Prediction of end-of-treat- ment abstinence or smoking from in- tratreatment abstinence or smoking was examined both descriptively (by cross- tabulation) and by means of odds ratios (ORs) calculated separately for active patch subjects and placebo patch sub- jects in each study, Additionally, sub- jects classiIIed ae smoking at 6 months (relapsers) were examined to determine what percentage of subjects resumed smoking at each week of patch treat- ment. We use the eategorieal variable abstinence or smoking as our major out• come variable because it is the variabte of chief clinical importance. All analyses were conducted on an intent-to-treat ba- sis with those unavailable for foUow-up or without biochemical confirmation of abstinence classified as smokers. RESULTS Study 1 Table 1 depicts the baseline charac- teristics of subjects in the active group and placebo group. No statistically sig- nificant differences were noted between the two groups. Of the 88 subjects who were enrolled and randomized, 77 com- pleted the 8-week active treatment phase of the study. Of these, one subject in the placebo patch group used nicotine gum throughout the study and was ex- cluded from all analyses. Of the 11 who failed to complete the active treatment phase (the first 8 weeks on study), 10 were using placebo patches, and one was using an active patch. Efficacy Abstinence was defined as a self-re- port of zero cigarettes smoked in the preceding 7 days confirmed by an ex- pired CO. value of less than 10 ppm. At the end of patch therapy (8 weeks), 26 active patch subjects (59,190) were claa- c in9 CessaGon Witt1 and Without Nicdine Patch-Kenford et al I

I I I I I I I I I I I I I I I I I sified as abstinent, while 17 placebo patch atitrjcCLb \i rre abaLinent (X'(~1f=1)='.3; P_ U7). Of the 87 study subjects, biochemical conf)rmation of smoking status was avatlable for 62 at the 6-month follow-up mark. Of the re- maining 25 subjects, five refused tu be interviewetl,ll failed to complete treat- ment, and nine «'ere unavailable for fol- luw-up. AL 6 months, 15 (34.1%) of 44 active patch subjects were abstinent, compared with nine (20.970) of 43 pla- cebo patch subjecta Mdf=1)=1.9; P-,17 ). Survival analyses indicated statistically significant (P<.O5) lower relapse among active patch subjects durine thp 6-month follow-up period.= Study 2 Table I depiets the baseline charac- teristics of subjects in the active group and pllcebo Kt•oul). Tl,~: two gi'oup5 did not differ reliably on any of these char- acteristics. Of the 112 subjects who were enrolled and rstndomized, 79 completed the treatment phase of the study. Of the 33 who failed to complete the treatment phase (the ffrst 6 weeks on study), 18 were using placebo patches, and 16 were using active patches. Efficacy Abstinence was defined as in study 1. At the end of patch therapy (6 weeks), 21 active patch-subjects (36.85i;) were abstinent, while 11 placebo patch sub- jects (20.0%) were abstinent (xz((lf=1)= 3.9; P=.05). Of the 112 subjects who par- ticipated in thia study, biochemical veri- fi(xtion of self-reported smoking status was obtained for 72 at the 6-month fol- low-up mark. Of the remaining 40 sub- jects, 33 failed to complete treatment, and seven were unavailable for follow- up. At this time, 10 (17.5%) of 57 active pauch subjects were abstinent, while only four (7.3%) of 55 plac.ebo patch subjects were abstinent (X1(df=1]=2.7; P-.10). Survival analysis indicated statistically significant (P<.05) lower relapse among active patch eubject3 during the 6-month follow-up pericul.2 Baael.ine characteristics did not differ reliably across the two studies. The dif- ference in attrition and efficacy seen ba- tween the two studies is hypothesized to be related to the nature of the adjuvant treatments (eg, group vs individual coun- seling and length of treatment sessions). Pretreatment Maaaurea Baseline expired CO, level, blood nico- tine and cotinine levels, cigarettes smoked per day, number of years smoked, and ): agerstrom Tolerance Questionnaire score were as6esse,'' in relation to end-of-treatment abstinence and 6-month abstinence. Point-biserial JAMA. February 23, 1994-VoI 271, No. 8 TiUfe 1.-Bac9lina Sub,act Chqrflcl6nstics for Study t ano t;tudy 2-  Study t ' btuCy 2 Actlve Patoh Placebo Patch Actlve Patch Placebo Patcn Var)able (n-44) (n-43)1 (n~S7) (nr56) age, y a33(1.5r :26i' -) ;3.:(1.21 da2(1 Si Fem3ie QdntlOr 'e 56 8 55 i 88.4 57 ;l Clgarertes per day 23 3( t 1) 30.3 (1.5) 29.8 (1..3) 30 8 rt.31 Explred a.r corbon monoz,oe ppm 32.3 (1.8) 31 4(2 61 32 5(1 d) 3a 3 (z 6; Nicotlne, nQ/mL ?t.o (1.1) 18.7 (i o) 2 1.0 11.3) :1.4 (1 1) Commne,ng/mL 328.2(184) 231.6(16.5) 3113(1rlg) 3)14(14.4) Yearo smoMng 25.2 (1.3) 24 3(1.a) 24 3 3 (1,2) 259 f1.a) Fagerstrom Tolerarv:n Qucsnonnaire u.ore 7.3(02) 6.9 (0.21 7.2 i0.2) 7,7 (0.2) We.,qM. w7 79 3 (2.3) 80 1 (3 21 729(23) 72 5 r7.3) eecn Oepression Invantory acore'' 5.t (0,7) 5.3 (0.7) 6.3 (0.6) d2t08) -Seanaard 4rrorG are in pareMhhW9nn. Nu gruup drtterqnGs were signf/icant (P>.05) in e;tner itudy tOne auoiect in me placeoo Patch Qyoup cnYweo nfeotine gum throuQhout It1e patcn treatmtnt pnaSa ol tne itu0y. This M.ortot wea elimrnated from an enatyses. Tabltl 2.-CorrslfltWns t3etwaen Ba6eIIr18, W@Zk 1, antl WABk 4 Wrthorawal Severity and Abstin2nC6 31 End of Treatment and 6 AAonth=• ( study 1 Study 2 Atuqnenoe .t 1 Weeks Abatlnence at a Months AbatMence at 0 Weeka Abstlnence at 6 Montns wrtnarawat Actnra Patch Ptecabo Patch Aotive Patch Placebo Patch AcUw Patch Place Patc bo Acttve h Patch Placebo Patch t?aseline - 36t (40) O8 (38) - 35t (40) .04 (3d) .1 a (49) -.19 (4 9) .06(49) -.20 (49t Week I -.385 (44) ,18 (40) - 422 (44) .15(40) -.02 (5fS) -.13 ( 53) -.04 (56) - 13 (53) WaeK 4 -.351' (43) -.25 (34) -.35t' (.f3) -.05 (34) - 10(51) -• 26(3 8) - 17 (5S) - 21 (38) -MIr)nertCCres on the wdhdrawet meaaura inotcg te pNater vnlhcNawa) symplom¢. Abst.nenci, vdnaWes aro cooad: 1 Indicatec abetrnence: and 0. srnokmq. The number ot tuo)ect! on whi0h correlation was based la rn parantlreses. The numb4r changod as a lunction of tne avadab+4ty ot usable wirhdrawat ratinga YP<.o1, correlations were conducted scparately for active patch subjects and placebo patch subjects for each of there vari- ables. None of these pretreatment mea- sure5 showed a consistent relationship with posttreatment succeas across stud- ics 1 and 2. None of the 32 correlations was statistically sipnificant, and none exceeded -.26, Intratreaunent Measures Precesaation serum nicotine and coti- nine levels have been proposed as mea- sures of tobacco dependence and as de- terminants of successful amoking cessa- tion.t` Additionally, the degree to which pretreatment serum nicotine and coti- nine levels are matched or maintained via replacement therapy may be impor- tant in determining cea3ation success.3 To exau)iue these relatiuru(, we com- pared the following three measures of nicotine and cotinine with successful smoking cessation outcome among ac- tive nicotine patch users: (1) baseline blood nicotine and cotinine levels, (2) absolute blood nicotine and cotinine lev- els achieved during nicotine replacement treatment (as measured at week 4 of patch treatment), and (3) percentage of baseline blood nicotine and cotinine lev- els replaced with nicotine replacement therapy. These point-biserial correla- tions revealed no consistent relation with outcome. For instance, the only signitl- cant predictor of 6-month abstinence was baseline nicotine level in study 2(puint- biaerial correlation=-,.''.7) (higher base- line nicotine levels were associated with frmonth relapae). The same correlation was small and opposite in sign for study I (point-biserial correlation=.1fl). Finally, we assessed whether the se- verity of withdrawal symptoms during treatment was correlated with success- ful cessation. A withdrawal severity in- dex was constructed by calculating the mean score for the daily eight-item with- drawal scale. Withdrawal severity at baseline, week 1, or week 4 after quit- Ling did not consistently correlate with abstinence (Table 2). To assess the pos- t:ibility that some of these intratreat- ment messures were confounded by in- tratreatment smoking, we separately analyzed data of subjects who were com- pletely abstinent; none of the correla- tions was significant. intratrsatnnent Smoking In contrast to the measures described herein, intratreatment smoking did dem- onstrate a consistent relationship «-ith posttreatment success. Specif)cally, we analyzed the relation between in- tratreatment smoldng by week and post- Strolcing Ct>seatirxl W~th and Without Nicotine Patctr-+Centoro et at 591 I

'aule 3-Smoklno n Weeks 1 tind 2 of Patcn Treatment and Abstlnence Status at End of Treatment and 6 Months in Actlve Pstch Subjects I I Abstinent m Woek t ? t 3ntl 2 $moklnq In Week 1 2 1 0'2 Table 4.-Smokin9 in WBAkS 1 antl 2 01 PBtOh Traatment antl AbsllrlBnCB Status at En0 01 Treatrri9nt 2n0 b tv10nt11d in P18cebo Patt:n Subjects No. of Subjects Smokinq at End of Treatment, % Smokin9 at 6 Months, % Noo of Subjects Snrokktg at End ot Trewtment % Smoking at 8 Monthri, % 17 70.8 76.5 34 82.4 97.1 1tt 68.7 63.3 35 85 7 97.1 2 2 8 530 77.9 At 78.0 96. t I I I I I I I I I I I I I I study 1, Acttve Petch Subjects (n.41) • Study 2, Actfve Ptton 8ub}eots (ns7) No, of Subjects Ab,tlMnoe at Eno of Treatmant, Y% Abstinence at 6 Months,'4 No of BubJecte Abstfn.ncs at K nd of Treetment, e4 t Abstlnence at 8 Months,'4 27 n.a 40.7 23 85.2 1 39 26 769 18 2 22 72.7 . 40.D 2? 81 8 45.5 16 75.0 jO.o study 1. Ptscebo Patsh Subjects (n=43) 8moking No. of Stttoking at End Sntotttng Abstinence at AbatInenoe at Noo of End Abtttn.nne No. of End .. tl 1 a nanoe Aoetlnent In WeeFt Sub{eets of Trortmant Y% at 8 Months, % SubJaota of Treatment, Y% at 6 Months, 1L t 11 79.0 385 12 6oo 83 2 , e 81.3 .3.e 10 70,0 30.0 1 and 2 12 83.3 41.7 5 50.0 0 No. of 9moktng at End Smoking In Week 4ubjeeto of Treatment, % at 8 Months, % 6ub{ectt of Treaiment, K at 6 IAonths, % 1 30 76.7 ea.7 43 884 t _.. s 30 2 27 85.2 92.6 43 91.1 97.8 1 nr 2 31 77.4 87.1 50 84.0 92.0 treatment cessation success. These analyses demonstrated that smoking during any week of treatment was highly and negatively associated with success- ful cesSation. ' Becauoe it would be most useful clini- cally to predict long-term autcomes early in the quitting process, three time peri- crds from the initial phase of treatment were analyzed intensively. These were (1) any smoking in the firAt week of treat- tnta;t, (2) any smoking in the second week of treatment and (3) any smoking in ei- ther the first or second week of treat- ment. Tables 3 and 4 show the relation of intratreatment smoking an short-term and long-terrn smoking ce.4sation rates using these three predictor time periods. Any smoking during the fitst 2 weeks of treatment was highly associated with a poor prognosis. Howeti•er. smoking or ab- stinence during week 2 waa the most ac- cvrxte pretlietur of outcutne (Table 5). Table 3 illustrates the relation between early intratreatment smoking and clini- cal success at the end of treatment and the 6-month follow-up mark among ac- tive nicotine patch users for each study. Smoking was defined as any tobacco use at a given time point. Among nicotine patch users, if a patient smoked at all =-during week 2, there was a6i.79u chance or 55.7% chance (atudies I and 2, reapee- tively) that the patient would be smok- ing at the end of treatment. This rela- tionship was also predictive for long-term outcomes. If a patient smoked in week 2, there was an 83.3% chance or 97.1% $62 .JAMA, February 23. 1991--vol 271. No. e chance (studies 1 and 2, respectively) that the patient would be smokin¢ 6 months later. Similar findings were noted among placebo patch users (Table 4). Among placebo patch users, if a patient smoked during week 2, there was:u185,2% chance or 91.196 chance (studies 1 and 2, respea tively) that the patient would be smok- ing at the end of treatment and a 92.6% chance or 97.8% chance (studies 1 and 2, respectively) that the patient would be smoking 6 months later. Abstinence during week 2 was also highly predictive of both short-term and lung-te:Tn abstinent.•e. Fur example, ar ttong active nicotine patch users, if a patient were totally abstinent during week 2, there waa a 76.99b chance and r^Z,7qo chance (studies 1 and 2, respectively) that the patient would be abstuunt at the end of treatment and a 46.2% chance and 40.9% chance (studies 1 and 2, respectively) that the patient would be abstinent 6 montha later (Table 3). Among placebo patch us- ers, if a patient were totally abstinent during week 2, there was an 81.8% chance and 70.09ro chance (studies 1 and 2, re- spectively) that the patient would be ab- stlnent at the end of treatment and a 43.8% chance and 30.0% chance (studies 1 and 2, respectively) that the patient would be abstinent 6 months later (Ta15te 4). We examined the issue of whether amount of stnoking in week 2 added to the predictive validity of our prediction rule. The results of a variety of anah sea suggested that including intonnation about the amount of smoking did not al- Study 2, Plecebo Patch Subjects (n.66) ter the predictive accuracy of the rule. Another test of the accuracy of a predic- tion rule is to determine the proportion of treatment failtire4 the rule iaentiIIe.s--ie, of all subjects smoking at follow-up, what percentage smoked by week 2? Among active nicotine patch users in studies 1 and 2, respectively, of those smoking at 6 months, 59% and 8396 had first amoked by the end of week 2(749c across both studies; Figure). Among placebo patch users in studies 1 and 2, respectively, of those smoking at 6 months, 79% and 9)95 had first smoked by the end of week 2 (869'a across both studies; Figure). Odds ratios constitute another index of association between early smoking and long-term outcome and reflect the elassification power of a prediction rule (ratio of correct to incorrect decisions yielded by the prediction rule). When smoking in week 2 of treatment was used as a marker for outcome among nicotine patch users, successful predic- tion was greatly enhanced at end of treat- tttettt (OR, 6.7 and 16.0 fur studies 1 and 2, respectively) and at six months (OR. 4.3 and 23.5 for studies 1 and 2, respec- tively) (Table 5). Although smoking dur- ing any time within the fYr3t 2 weeks is predictive of failure to quit, the ORs demonstrate that the week 2 rule is as- sociated aith the most accurttte predic- tions and the least number of miscias- sifications. The power of the week 2 rule to detect the observed association with 6-tnonth outcome was .53 in study 1 and .95 in study 2. Similar findings are ob- Srndting Cessauan With and Witnout Nicotme Patch-Ken(oro et ai I I

I I I I I I I I I I I I I I I I served when the ORs tt)•(, rHlculatecl for Nlacebo patch ;1xbJPCt_ (Tnble S). acldi- tionally, c5 cuefticient. i al carl•elattion co- efficient for t«•o clichotomous vHi~ahles) of the (ieciAion rule for euc•h timc point were StaListicaliy ;it,mificant. Finally, we )•e!•omputeri the OR.:uLer statisticaliy contrulling for a variet,V of other potentlul llreclictc r ~:+t iub ies, i n pa)•- ticular, we exantinet 1 the relation between week 2 smoking tuul abtitinettce, both at end of treatment and at R-month follow- up, aS.er contml I i ng tot• rneaaurey of ph,yai- cal dependence (Fagerrt)•om Tolerance Questionnaire score, pretreatment blood nicotine level, number of cigarettes smoked pretreatment, and pretreatment expired C:O. Ir,:L•e1) and withdrawal dur- ing the first week of treatment. These analyses revealed that in all cases the adjusted ORe were actually higher than the re8pectlL•e unadjusted ORs. Prediction of Week 2 Smoking Because smoking abstinence in week 2 was such an accurate predictor of long- term success, we performed simulta- neous logistic regression analyses to identify predictors of week 2 stnokirlg. In particular, we examined the relation of week 2 smoking and a variety of other predictor vxriablea: withdrawal sever- ity of week 1, pretreatment Fagerstrom Tolerance Questionnaire score, pretreat- ment biood nicotine level, pretreatment expired C0, level, and mean number of cigarettea smoked daily at pretreatment. These analyses revealed that no vari- able predicted week 2 smoking in study 1, but two did in study 2: pretreatment Faserstrom Tolerance Questionnaire score and pretreatment blood nicotine level. However, their level of associa- tion with week 2 smoking u•as low, with no e value exceeding .19. Therefore, no variable consistently and silrnificantly predicted week 2 smoking across both studies 1 and 2, and when associations were found, they were low. COMMENT The intent ofthis 1•e~earch a•a.; to iden- tify a simple yet powerful pt-atjiction rule that could be used easily in a clinician`a office to predict smol:ing ce::ation suc- ce>s or failure with or without the nico- tine h;ttch. Anal}•ses were based on two sep u ate ~tuclie~, u:ing two independent samples of smokers, t«•o different ad- juvant counseling therapies, and active patch therapy vs placebo patch therapy. In this way these two studles provide an oppo)•tunity to test the generalizability of our findings. The analyses reveal one powerful pre- diction rule for clinicians: any smoking in the first 2 week-a of treatment predicts both short-ter)n and long-term failure, with week 2 smoking being particularly predictive of outcome (Table 5). (:on- versely, total abstinence during the first 2 weeks of treatment was consistently correlated with sustained smoking ces- satton success. By assessing the pres- ence or absence of any smoking during the second week af treatment, clinicians can predict with good accuracy whether the patient will or will not quit smoking. 1~arly detection of high-risk patients may also allow the clinician to modify treat- ment to increase the likelihood that at- risk patients will aucceed. These analyses also identified a num- ber of factors that did not consistently predict smoking cessation outcome. 5pe- cifically, theae analyses cast some doubt on the ability of numerous acx.epted or hypothesized measures of tobacco depen- dence to predict success or failure con- sistently for patients using the nicotine patch.17 Baseline indicators, such as the number of cigarettes smoked per day, Fagerstrom Tolerance Questionnaire s,core, and years of smoking, were poorly and inconsistently correlated with suc- cessful cessatlon. Biochemical measures also fared poorlyas predictors: pretreat• ment expired C0, value, pretreatment blood nicotine and cotinine levels, abso- lute blood nicotine and cotinine levels achieved on the nicotina patch, and the percelitage replacement of pretreatment nicotine "or cotinine with nicotine patch therapy were poorlr and ineonsistently correlated with successfui smoking ces- sation• These me,eures might have shown Significant retatic,ns, ~rith outcorne had we huii many more subjects in our two stud- ifs, but we do not believe that this would change the relative superiority of the week 2 rule to predict outcome. Our current analy-ses failed to identify an easy-to-use clinical measure to iden[ity a priori (ie, before treatment begins) 3mok- ers «ho are most likely to 4;uccecd or fail with or t~ithout the nicotine patch. One caveat to keep in mind is that our results may merely reflect unreliable assessment or thr selection of 12nproper measures of e Active Petch (n-29) ~ 1so ~ e Plac.t» Potch (n .3a) r o- e_ -o, ~~ ~ao 30 20 10 j~ 0 Quit 1 2 3 4 5 6 7 8 Six•Month Day Week Foltow•up 100 90 . ao ~ 70 so so 40 ¢ o 10 0 Ouoi 1 2 3 4 S 8 7 a S x-MOntn Day Week Follow-up TM cumulatNa proponion od sub{ects amoktnq at e-morltn foUOw-up for study 1(top) and stutly 2 (DOftom), procanted aa a tuncrion of tfhe week at whtttl ttey smoked their first dqarotte (1e, wnen re- tapun atarletl to amodca). TBDIe 5.--OCCt AapM (ORS), 95:: ConfitlanCi Intarvata (Cla), and 4, Coefficuents for SmOktnq !n WeekS 1 an0 2 antl AtlsflnCnea 5tatus at End of Trealment and 6 Months in Active Patch Subiects and Placebo Patch Subiezta Bttxty I Study 2 End of TreatmetM 6 Montht End of Treatment e MontM Acffw Patch Ptac.bo Patc h Active Patch Plaoebo Patoh Aettve Paton Plaoabo Pltoh Active Patch Ptae.bo Patcn W.ek (n.4d) (n.4s) (n.44) (n.43) (n=37) (n+.S5) (ns7) (n.s6) i OR (s5et CI) 3.4 (2.1-33.6) +t.0 (2.3-51.2 ) 2.2 (0.6-8.7) 4.1 (0.88-18.0) 8.8 (2.6-20A) 7.6 (t,A•32.0) 21.2 (2.5•ta3.7) t.2 (o.t•t2.a) 9 coentuenl •48' SO' .18 .28 .4a' ,39` ,47' .02 2 OR (05X Cl) 6 7(1 7-25.4) 24 0(t 8-129 .0) 4.3 (1.0-1s.5) 9.7 (1.7-5eF,7) 16.0 (4.2-60.7) 23.0 (4.4-130.7 ) 23.5 (2.7-20i.6) ta.0 (1.7.207.e) 9 cOetitclem .44' .66' .31T .43' .59' ,59e .48- .41' i or2 OR (95% cl) 7.9 (2.0-31.e) 17.1 (3.0-97.3 ) 2.6 (0.e•1o.4) 4.8 (1.0-22.8) 10.7 (2 8-41 2) 7 0(1.1-sd 0) tA e(3 s-,o0 e) 1.1 (1 0•1 2) a cosnicient 46• .56' .24 .31t .49' 3?t 53• .00 'P<A1. TP<.05. JAMA• February 23. 199-1•-Vot 271, No. 8 Srt+o+tin9 Cessation Wrtn and Wrtrbut Nocotine Patci•1--lcenford et al 593 ZND ~j ~y

I I 1 I I I I I I I I I I I I I I dependence. We believe our biochemical connrmation of outcomes, our use of rec- ognized assessment instruments, and the broad array of dependence messtu•es an [,*ue against these pos-qibilities. Another t iiveat is that a restriction of rang+; influ- enced our reaults. Our samples consisted of moderately to heavily dependent stnok- ers; a wider inclusive-sampling of the de- pendence continuum may have shown a btrunger relationship between outcome and measures of dependence. Nioreover, these results may be moat relevant to the 24-hour nicotine patches that are used by 90% of the current patch users. Addition- ally, all our subject's received supportive counseling, Predictive relations might dif- fer for self-quitters .•s those in formal treatment programs. Finally. our inabil- ity to identify a clinically useful a priori mesaure may reflect the. heterogeneity of smokers in the United States attempting to quit. That is, there nuty be subtypes of smokers, and no single predictor will per- form well across these different subtypes. Finally, as our smokers were elvrolled in a university-based clinical research study, these results may not be as powerfW in a typical clinic population. Our results do provide important in- formation for clinicians seeking to aid their patients in quitting smoking. Spe- cifit:ally, the following clinical implica- tions deserve emphasis! 1. Clintcians should-stresa that total abstinence is central to a successful smok- ing cessation effort for most smokers. Optimal smoking cessation counseling should emphasize the importance of early and complete abatinence. Across our two studies, among individuals on the nico- tine patch who were abstinent in the sea ond week of treatment, 47% were suc- cessfully abstinent at 6 months. In con- trast, among individuals who smoked at all in the first 2 weeks of treatment, only 11`:S were abstinent at 6 months. Most smokers who have precrioualy tried to quit are already aware of the danger of any smoking and provide anecdotal sup- port for this clinical rule. A common re- port among smokers who had previously relapsed is that they thought they could have a cigarette "now and then" only to quickly relapse to prequit smoking lev- els1° Hall and colleagtles]' have found that those individuals holding a strict ab- stinence orientation are more likely to achieve long-term success. 2. Clinicians should "front load" smok- i ngcest3atton counseling and support dur- ing the first 2 weeks after quitting. Among the subset of our patients us- ing the active nicotine patch, 74% of those who ultimately relapsed by 6 months began smoking in the first 2 weeks. Among placebo patch patients who had relapsed by 6 months, 86% be- bD4 JAMA, February 23, 1994-Vol 271, No. 8 gan smoking in the first 2 weeks. Cli- nicians need to provide support and coun 3eling during this critical period of high relapse. For this reason, clinicians should schedule a follow-up visit in persol: or make contact by telephone" during tne important first 2 weeks after quitting; review of coping stratel,*ies and the im- portance of total abstinence ahbuld be emphasized during this contact. 3. Assess smoking during the second week after the quit attempt to predict smoking cessation outcome. Smoking during any portion of the treatment period predicted a poor out- come,l-Iowever, we reconutlend using any srnokinQ during week 2 as a predictor of outcome because it occurs early in the quit attempt and requires the determina- tion of abstinence for only a brief period of time. Importantly, it is an easy-to-uae prediction rule for clinicians: follow-up with the patient in person or by tele- phone during the second week of a quit attempt-if the patient has smoked at all during this time period, consider an al- teration of treatnlent. More research may result in re6nement of this prediction rule. but in our studies, any smoking indicated heightened risk for eventual relapse. While our results suggest a reliable prediction rule, they do not reveal how treatment should be altered if patients smoke during the first 2 weeks of treat- ment. Our clinical experience suggests that if a patient is smoking in the second week of treatment, the clinician could con- sider offering the patient more intensive pharmacotherapy and/or additional ad- juvant therapy (perhaps referral to a for- mal cessation program). However, our elinical experience suggests that these steps should be taken only if the patient is stiU motivated to stop smokuig. If the patient reports feeling discouraged or de- feated, treatment may be tenlporw-Dy withdrawn, allowing the clinician, and pa- tient to jointly select a new, future quit data Obviously, more research is needed to determine appropriate treatment al- •terationa and responses to early failtlre. Finally, our reaults point to two other reaearch needs: first, re>aearch aimed at the identification and evaluation of smoker subtypes so that treatment and pretiic- tion cxut be tailored to the individtlal and second, the feasibility of a graded, etepped-care approach to intervention should be determined." This is the ap- proach used for many medical disorders, such as hypertension or hyperlipidemia, where dose or other dime/wons of treat- ment are increased only when an indi- vidual fails to respond to previous treat- ment. If a similar approach proves effec- tive with smoking cessation, it would serve as a model for clinicians as they combat this devastating chronic disease.' This rogt+Hrch wx~ runded in a.vet by Elun Ft+;tr- maCeut:Cwl: Inc. Athlnne, Irilund, ,tr.d Lcdcrlo LAhur,twrie:., U,'Mynr., V.1. Referoncea I. Fim•e 1fC. Jorrnby DP:. Bxket'-CB. Iienfurti SI., T'obaccudrpcndence and tite nitrtir.e pxtch:ellnical pvidClir,e6 for uifecuve u:e, JA.U.d. 1vyY:itit;aSi- :stw. 2. 1'iore N4C, lticnford SL. Jnrenby DE. Lt'rtter Dw, Smith SS. Beker TB. The nicotir,e u,trh: elini- cal etfectireneas: with ditierent cnum,aliny u•ait. ment3. Cbmt. In presi.. 3• 2olmrn A~, RrHntlon TH,Jotrnby DE. F3okcrTR. Sleaeurea otwfYeet end nieotinc dc;xn.ience predict litierent rl re ~ponac u+RrtwklnK tt~:ntion uYK:ments. J Corta,l! Cliu Prur)eul. t 131;i,J:01:1 5•Sy. 4. Tiffany ST, Baker TB. The role of a.•eision and ccutntohng ettategten in tt•esunenLr for uiKttretto amokinQ• In; Efaker'I'E. Cannon US, etlt. .t.,.r,.c• ntenf and 7Yeatment oJ Addicfiur DiNOorlnr,t. Nett• 1'ork, NY: Praeger, ]0b'S:~-?SV. 5. West Dw, Grtth.m 5, 5a'aneon li,'<t'ilkin~on 0. Five year follow-up Of a xmoking withdrawal clinic population. Am J Public Health. 1977:G7:$:15 ',-t4, 6. Horv,•itz Mb, Hinr{i-AlexanJer MT. l4'egner TJ• Pyyehosocial mediaton of abatinence, relapi;e, and continued eraoking: a one-year fulluw-up of a mint- mal inten-ention. Addict Behan. ih5,;lt}:'S 39, 7. Gatvey AJ. Blias RE, Hitchcock JL, Heinhold J W, Rosner Li, Predictors of smoking relal,Ye unong setf-quittcni: a repott from the normative a¢InC study. rtddict Behnv, iM2:17:3d7-a71. 8 Pinto RP, Abrama DR, lfonti PM. Jacoloue 51. Rieotine dependence and likelihood of quittina ymok- ing. Addict B.thav. 1Pii7;12 171•3~i. 9. Fagerrtrom KO, &hnekler NC. lfeaauring nieo- Gne dcpondence: a rsview• of the Faeerstrom Toler- anee Qutytlonruire, J bsl,¢v Sfed 19o9t1"0-2ti? 10. Killen JD. Forutuutn $P, Krwartwr, HC, Varsdy A. Newman B. Who will relapse? sytnptoms of tuco- ttne dependence predict long-term relap+e atter smok- ingeeswtion. J Corvttlt Clin Fqycho!-1992J+tk~J7-P))l. 11. Abelin T, Buehler A, Muller P, Vesanen K,. Imhof PR. Controlled trial of transdermal nicotine patth in tnbaoeo withdrawal. Lancft, 19&4;1:7-10. 12. 'Iranadarmal Nicotine Study Group. Trantder• me] nicotine fer amoking ceaation. JA.4fA. 1991; 206:3133.3198. 13. HugheeJR, Hatsukutti D.9ignaand sylnptoml: of tobacco withdnwal, Arch Cin PayeAtairy, ]gbd; 43:2fs9-29i. 14. Beck AT, Ward CH. alendelson M. Mack J, Erbaugh J. An inventory for meaeuting depns- sion, Arch Gert Psychiatry, 1961:4b81•57t. IS. Hurt RD, Lowell CD, Offord KP, at al. Sarum nicotlne and cotininn lev.le durina nkotine-patch therapy. Clin PAnr+natol Tker. 1999 51:OG10B- 14. Ruesell SlA. Nicotine replacemenu the rote of blood nieotit» levala, their rate of ehanre and nieo- tine tolerance, In; Niaocine Rsplaum.nt• •4 Criti- cul Evaltwtiun, New York. NY: Alan R Liu Inc; 1986:81-94, 17. Hughes Jli. Tobacco withdrawal in self-quit- tara J ConattU Clin PayeAol. 1992:60:889-G97. 18. Bttit]don TH, 7,eltnan DC, Baker TB, Effectu of maintetunca setsions on smoking relap.: delaying the inevitable, J Cona><lii Clin PtytAol. 1437;So: z>i0-7S2- 19. Ha11 8M• Havuiy 8£, W/a3serman DA. Com- tnitment to abstinenee and acute stress in relapae to alcohol, opiate•a, and nicotLu• J Consult Clin P,ychoi, 1490;b8:176-tal. 20. Lando HA, Helleretedt, WL, Pine FL. JdcGov- errl PC. f3rkf suppordve tckphone outlesch ss a reauitment and intervention strttteti:y for tmoking cewation, Ans J Publie Aealth. 1992;Ji:41d6. 21. Orleana CT, Treuing nicotine dependenee in medityJ settings: a stepped-estrt model. In: Orleans CT. Slade J, eds. Vicrotins Addiction: Prineip/es and Manapernent, New York. NY: Oxford Univer• sity Pre"t 19$E:14b-181. 22, Fiore MC. Smoking atatus: Ute new vital tigtt ateeeeing and documerttin,g smoking status. J.LtfA. 1L91~66 31aS-s1Bi Smoking COS3atkxl With arld WittlOttt Nlcotin® Patr,tt--Kenford e,i el I