- Email: [email protected]
Nicotine Patches and Uninsured Quitline Callers
Nicotine Patches and Uninsured Quitline Callers A Randomized Trial of Two Versus Eight Weeks Timothy A. McAfee, MD, MPH, Terry Bush, PhD, T. Mona Deprey, MS, Lisa D. Mahoney, MPH, Susan M. Zbikowski, PhD, Jeffrey L. Fellows, PhD, Jennifer B. McClure, PhD Background: State-level tobacco quitlines are integrating nicotine replacement therapy (NRT) into service. Because of funding limitations some provide short courses of NRT. No randomized trial has evaluated the relative benefit of short versus standard treatment. Design:
A two-cell randomized trial comparing 2 weeks of NRT to 8 weeks.
Setting/ Uninsured callers to the Oregon Quit Line during a free-patch initiative from October 18, participants: 2004, to May 5, 2005, who were 18 years or older, smoked five or more cigarettes per day, did not have a medical contraindication to NRT use, and were interested in quitting in 30 days. Data were collected from April to November 2005, and analyzed in 2006 –2007. Intervention: Participants were eligible for two phone counseling sessions. 1154 participants were randomized to receive via the mail either 2 or 8 weeks of nicotine patches. Measures:
Primary outcome was self-reported complete abstinence from tobacco for 30 or more days at the 6-month phone survey. Secondary outcomes were 7-day point prevalence and 90-day abstinence, satisfaction, and patch use. ORs and CIs were computed. Cost per quit and incremental cost per additional quit were computed based on program costs.
Results:
Intent-to-treat 30-day abstinence was 14.3% in the 2-week group, and 19.6% in the 8-week group (OR 1.45 [CI⫽1.01, 2.12]). Average cost per quit was $1156 for 2 weeks and $1405 for 8 weeks, with an incremental cost effectiveness of $2068. Satisfaction increased from 90% to 97% with 8 weeks. Those receiving 8 weeks of NRT took more calls (2.0 vs 1.6) and used more patches (6.3 weeks vs 4.3 weeks), but were less likely to purchase patches (16.2% vs 39.3%).
Conclusions: Eight weeks of patches improved quit rates compared with 2 weeks, and was cost effective. (Am J Prev Med 2008;35(2):103–110) © 2008 American Journal of Preventive Medicine
Introduction
S
tate-level tobacco quitlines grew rapidly in the past decade, with over half a million people accessing evidence-based cessation treatment through them in 2005.1–3 A strong evidence base supports quitline efficacy and effectiveness, including multiple metaanalyses.4 –7 Nicotine replacement therapy (NRT) is being rapidly integrated into state tobacco quitlines in the U.S., Canada, Europe, and Australia/New Zealand. As of 2006, over 20 states reported some form of NRT distribution, with most either time-limited, populationlimited, or involving less-than-recommended durations of therapy.8 NRT has been introduced for two main From the Department of Health Services, School of Public Health, University of Washington (McAfee); Clinical and Behavioral Sciences, Free & Clear (McAfee, Bush, Deprey, Mahoney, Zbikowski); Group Health Center for Health Studies (McClure), Seattle, Washington; and the Center for Health Research, Kaiser Permanente Northwest (Fellows), Portland, Oregon Address correspondence and reprint requests to: Timothy A. McAfee, MD, MPH, Free & Clear, 999 Third Avenue, Suite 2100, Seattle WA 98104. E-mail: [email protected].
purposes: (1) as a promotional aid increasing calls to the quitline; and, (2) to improve quit rates. Several evaluations have established that offering free NRT will result in dramatic increases in calls to state-level,9,10 employer/union-based,11 and health plan12 quitlines. Using NRT as a means of promotion may be considerably more cost effective than the more traditional use of mass media.13 Because of limitations in state funding, a number of quitlines have been experimenting with courses of NRT that are considerably shorter than the manufacturer’s and U.S. Food and Drug Administration (FDA)’s recommendations (8 –12 weeks).14 Courses of 2 and 4 weeks are common, and even 1 week has been offered.15 The amount of NRT offered seems to be a minimal issue in terms of creating a promotional impact (Deprey M, et al., Free & Clear, unpublished finding).16 However, much less is known about the impact of shorter courses on quit rates. Cummings et al.17 compared different amounts of NRT and their effect on quit rates, and did not find a significant difference between 2 weeks and 8 weeks.
Am J Prev Med 2008;35(2) © 2008 American Journal of Preventive Medicine • Published by Elsevier Inc.
0749-3797/08/$–see front matter doi:10.1016/j.amepre.2008.04.017
103
However, the comparison groups were not randomized, had variable response rates, and were not comparable populations, with different groups varying in insurance status, geographic location, and recruitment method. Determining the impact of offering 2 weeks of NRT rather than 8 weeks on quit outcomes and satisfaction has financial, clinical, and theoretical importance. If additional NRT does not improve quit rates, providing shorter courses could free up additional financial resources to treat more people, provide more-intensive services such as proactive counseling calls, or provide increased promotion. For example, twice-a-day dosing of bupropion SR does not cost effectively increase 1-year quit rates compared to once-a-day, but proactive phone calls do when provided via a health plan quitline.18,19 However, if increased duration of NRT does improve quit rates, then policies developed assuming it does not may decrease an individual’s chances of succeeding. It is also expensive to drive calls to a state quitline; thus, providing less-optimal service may actually not be cost effective from a systems perspective. The uninsured use state quitlines disproportionately,20 and have fewer options for accessing evidence-based cessation support. Finally, if 2 weeks works as well as 8 weeks, the primary mechanism of action for NRT may be its mitigation of short-term withdrawal symptoms, and its role in long-term relapse prevention may be less important.15 The state of Oregon conducted a time-limited campaign in which all tobacco users in the state were eligible to receive free nicotine patches. The primary purpose of the overall campaign was to test the impact of the availability of free patches on call volumes to the Oregon Quit Line (ORQL). All residents were eligible for 2 weeks of patches during the campaign. This amount was chosen due to funding limitations and to avoid giving the health plans a disincentive from providing NRT support for their insured patients. Pre–post comparison in insured callers of quit outcomes,21 cost effectiveness,13 and promotional impact (Deprey M, et al., unpublished finding) are reported elsewhere. Because uninsured callers do not have any possibility of receiving additional NRT from a health plan, the state of Oregon was interested in determining what added benefit 8 weeks provided compared to 2 weeks. This paper presents results of a randomized trial for uninsured callers embedded in this larger initiative, comparing 2 weeks of NRT versus 8 weeks, with both groups getting two telephone counseling calls. Six-month quit rates, use patterns, satisfaction, and the cost effectiveness of additional NRT are reported.
Methods The study design was reviewed and approved on September 15, 2004, by the Western IRB and on September 24, 2004, by
104
the Department of Health Services, Public Health Division/ Multnomah County Public Health IRB. Data were collected from April to November 2005, and analyzed in 2006 –2007.
Eligibility Callers to the ORQL were eligible for this study if they called between October 18, 2004, and May 5, 2005, and were (1) uninsured, (2) 18 years or older, (3) spoke English, (4) had a working phone number, (5) smoked five or more cigarettes per day, (6) were interested in quitting within 30 days, and (7) provided informed consent over the telephone including consenting to randomization to receive different service offerings and to an additional studyrelated follow-up survey call. Eligibility also required callers to have an interest in using NRT and be medically appropriate for NRT based on FDA criteria.
Recruitment, Consent, and Randomization Availability of free nicotine patches was promoted via a press conference covered by most television, radio, and newspapers in the state, as well as through announcements to community and healthcare provider organizations. Word-of-mouth promotion was encouraged by inclusion of referral cards in materials sent to callers. Callers to the ORQL were screened initially by registration staff to determine eligibility and interest. Eligible, interested callers were transferred to a cessation counselor and formally consented. The consent process included informing potential participants that if they participated in the study, they would receive 2 weeks of NRT and a second counseling call (in addition to the standard service single call), and that some participants would be randomly assigned to receive up to 8 weeks of NRT. After consent, the counselor randomized the participant, using an application embedded in the counseling support database. Randomization occurred in blocks of 12 so that after every 12th person recruited the cells balanced.
Intervention Participants in both cells were eligible for two phone counseling sessions, and could also call in on an ad hoc basis. The first counseling session occurred immediately after randomization, and the second was scheduled for the week after their quit date. Counselors’ training included techniques to manage call length, but time limits were not set. Content of the phone counseling sessions was based on recommendations of the U.S. Public Health Service Clinical Practice Guideline. The two calls were an abbreviated version of a five-call program (Quit for Life™) found to be effective in multiple randomized trials.18,22,23 Counselors also provided education on the proper use of NRT, including side-effect management and ongoing use. Experienced telephone tobacco counselors provided the interventions, with computer-driven protocol support. A subset of calls was taped and reviewed with feedback to the counselors for quality assurance monitoring. All participants received identical printed support materials. Participants randomized to Cell 1 received a 2-week supply of patches. Individuals were mailed either 21 mg if the participant smoked more than ten cigarettes a day, or 14 mg if they smoked ten cigarettes a day or less, as recommended by the FDA. Participants were encouraged to save money not spent on cigarettes, in order to purchase additional patches
American Journal of Preventive Medicine, Volume 35, Number 2
www.ajpm-online.net
(a cut-out cardboard “piggy-bank” was included in the mailing). Participants randomized to Cell 2 were sent an initial 4-week supply of patches and were eligible to receive a refill including FDA-recommended lower doses for a second four weeks. Participants had to complete their second call, or call back, in order to receive a refill. Counseling and medication protocols are described in detail elsewhere.20,23
(OR) with confidence intervals for quit rates between Cells 1 and 2 were calculated. Reported patch use in each cell was tested for trend using the Cochran–Armitage test for trend, which tests for trend in binomial proportions (quit or not-quit) across levels of a single factor (patch use). Patch use was divided into 0 to ⬍1 week, 1–2 weeks, 3– 4 weeks, 5– 6 weeks, and 7⫹ weeks. The analysis was generated using SAS/STAT® Software 9.1.3.
Data Collection
Cost Effectiveness
Basic demographic information was collected during the routine registration process at the initial call. Participants in the study were called back 6 months after registration by an independent survey research firm, Marketing Strategies, Inc. Two weeks before the call window opened, a letter was sent reminding the participant of the upcoming call, and that they would receive $5 for completing the survey. For those who still had working phone numbers, 12 attempts were made over a 4-week window. The window opened 2 weeks before the due date of the call and closed 2 weeks after the due date. Responders were asked about their smoking status, quitline and additional service use, satisfaction with quitline services, experience with patches and other medications, and out-of-pocket expenses for additional cessation services and products.
The cost of each intervention strategy, average cost per quit, and the cost effectiveness of counseling plus 8 weeks of NRT compared to counseling plus 2 weeks of NRT was estimated assuming a state-program perspective. The program perspective included state-paid intervention costs for telephone counseling, mailed self-help (quit kit) materials, and NRT. Quitting was defined as 30-day abstinence at 6 months assuming adjusted intent-to-treat (as described above). Cost per quit and incremental cost-effectiveness results were also estimated for responders only and for all sampled callers (even those without valid phone numbers). For the latter analysis, it was assumed all callers not reached were treatment failures. This provided a high and low range for intervention cost effectiveness. The incremental cost per additional quit of counseling with 8 weeks of NRT compared to counseling with 2 weeks of NRT was estimated by dividing the difference in intervention costs by the difference in quit rates. Quitline promotion costs during the initiative were the same for each strategy and were excluded from the cost-effectiveness analysis.
Outcome Measures and Analysis The primary outcome was self-reported complete abstinence from tobacco for 30 or more days at the 6-month survey. Six or more months is the North American Quitline Consortium’s recommended time interval for telephone quitline quit rate follow-up.8 The primary analysis used for costeffectiveness computation was based on an adjusted intent-totreat analysis where those randomized to receive treatment, regardless of whether they actually completed the two calls or used the patches, were included in the analysis. Those with working phone numbers who could not be contacted were included in the analysis, imputed conservatively to be smoking. However, it was decided a priori to exclude those who had nonworking phone numbers at 6-month follow-up from the primary analysis. This was done for three reasons: (1) study participants were uninsured, with a high rate of mobility and phone disconnects20; (2) phone disconnect rates are unlikely to be related to quit status; and (3) the study intervention involved only two sessions of brief phone-only contact conducted in a real-world environment with minimal opportunity to update contact information. Because there is wide variation in how quit status is reported in trials, 7-day point-prevalence and 90-day continuous abstinence at 6 months were also measured. In addition to the adjusted intent-to-treat quit-rate calculations, 7-, 30-, and 90-day quit rates were computed for survey respondents only, and for the total random sample (intent-to-treat analysis including those without working phone numbers imputed to be smokers). Satisfaction and patterns of NRT use were compared based on those responding to the survey. Thirty-day adjusted intentto-treat quit rates were compared between the two cells based on reported use or non-use of NRT. Chi-square two-tailed comparisons for statistical significance were made. Fisher’s exact test was used when a comparison included a cell with less than five counts. Odds ratios
August 2008
Results The overall response to the free-patch initiative is detailed elsewhere.13,21 Briefly, the initiative earned media attention, influenced six health plans to change their cessation benefits, and increased overall calls to the ORQL to greater than 10,000 calls in the first month (a 20-fold increase over baseline), with a continued doubling of baseline call volumes for 6 months after the public campaign of free NRT to all residents was concluded. There were 2934 uninsured callers registered with the quitline during the study period, of a total 10,596 registrants (27.7% compared to state uninsured prevalence of 17%); of those, 2092 were formally assessed for study eligibility by the program staff (Figure 1), and 1154 were randomized. The 6-month uninsured survey response rate was 49.4% for those offered 2 weeks of NRT (Cell 1) and 53.0% for those offered 8 weeks (Cell 2) (p⬍0.30), excluding those with nonworking phone numbers. Cell 1 and Cell 2 participants completing the survey did not differ significantly on demographics and tobacco history (Table 1). There was no difference between Cells 1 and 2 in the rate of nonworking phones (30% vs 31%, p⬍0.72). Those with nonworking phones also showed no significant differences for any baseline demographics and tobacco history between Cell 1 and Cell 2. There were significant differences across the cells beAm J Prev Med 2008;35(2)
105
Analysis
Follow-up
Allocation
Enrollmennt
tween those with working 7879 initially excluded and nonworking phones. 10,596 Oregon Quit Line 7662 insured Those with working registrants during study 217 uninsured enrollment window phones (both respond67 smoked <5 ers and nonresponders) cigarettes per day 59 non-English were more likely to be 625 not screeneda speaking older (41.3 vs 36.1, 91 not ready to quit ⫺7.16, p⬍0.001); women (65.3% vs 54.7%, 11.55, 2092 uninsured p⬍0.001); and white (90.2 938 not enrolled screened 300 declined enrollment vs 85.9, 4.61, p⬍0.03) 55 medical contrathan those with nonworkindication for NRT use ing phones. Having less 1154 randomized 583 recruited to a than a high school educaseparate study tion did not affect phone status. 578 received 2 weeks of NRT 576 received 8 weeks of NRT Baseline characteris576 NRT mailed 575 NRT mailed tics of responders and 2 NRT not mailed 1 NRT not mailed nonresponders were compared (Table 2). Responders were 2 576 total population ITT 578 total population ITT years older on average, more likely to be high school graduates, and 405 adjusted ITT 398 adjusted ITT more likely to have 200 responders 211 responders smoked more than 20 205 nonresponders 187 nonresponders years. 173 unreachable (no working 178 unreachable (no working Randomization to rephone) phone) ceive 8 weeks of NRT (Cell 2) increased the Figure 1. Participant flow diagram a Due to high call volumes during intense service delivery periods OR for being quit at 6 months for 30 days or ITT, intent to treat; NRT, nicotine replacement therapy more to 1.45 for adreported at follow-up actually using the patch were justed intent-to-treat. The OR for being quit for 7 days much more likely to quit for 30 or more days than those was 1.52, and for 90 days was 1.63 (Table 3). Including who did not use the patch. In Cell 1, 3/27 non–patch only responders in the analysis produced ORs of 1.43 to users (11%) versus 55/170 patch users (32%) were quit 1.59 (depending on quit definition), with absolute for 30 or more days at 6 months (p⬍0.03). In Cell 2, increases of 8% to 10%. Including those with nonwork1/12 non–patch users (8%) versus 77/197 patch users ing phones counted as smokers (classic intent-to-treat), (39%) were quit for 30 or more days (p⬍0.04). In both produced ORs ranging from 1.40 to 1.56. cells, quit rate increased with amount of patch used, Those randomized to receive 8 weeks were more likely to ranging in Cell 1 from 11% for those using zero to less be satisfied, and more likely to actually use patches (Table 4); than a week of patches to 57% for those using ⱖ7 weeks 66% of participants randomized to Cell 2 requested and (p⬍0.0001, Cochran–Armitage trend test, ⫺4.9146), were sent a second 4-week shipment of NRT. Survey reand in Cell 2 ranging from 12% to 48% (p⬍0.0001, sponders received the second shipment at a higher rate than ⫺3.7532). nonresponders (72% vs 58%, p⬍0.03, chi-square 8.93). Intervention costs per participant were $165.82 for Among responders, those who got the second shipment counseling and 2 weeks of NRT (Cell 1) and $275.40 were much more likely to be quit for 30 days or more at 6 for counseling and 8 weeks of NRT (Cell 2). These costs months than responders who did not (46% vs 15%, adjusted reflect actual NRT delivery. Assuming 30-day abstiintent-to-treat, p⬍0.0001, chi-square 17.1). nence at 6 months (based on adjusted intent-to-treat), Those randomized to receive only 2 weeks of NRT estimated average costs per quit were estimated at were almost 2.5 times more likely to purchase more $1156 for Cell 1 and $1405 for Cell 2. Each additional NRT (Table 4). quit for Cell 2 cost $2068 compared to Cell 1. For Although NRT use was not a precondition for particresponders only, mean costs per quit were $564 for Cell ipation, only three participants were not sent NRT (two 1 and $738 for Cell 2. Incremental cost effectiveness for in Cell 1, one in Cell 2), due to lack of interest. They Cell 2 was $1384 compared to Cell 1. When all nonrewere included in all analyses. Those in both cells who 106
American Journal of Preventive Medicine, Volume 35, Number 2
www.ajpm-online.net
Table 1. Demographic characteristics of responders by cell (% unless otherwise specified)
Mean (SD) age in years 18–29 30–49 50–64 ⱖ65 Women Hispanic Nonwhite Marrieda Employed full- or part-timea ⱕhigh school education Unable to worka,b Mean (SD) cigarettes/day at registrationc Report prior use of patches
Cell 1 2 weeks NRT nⴝ200 (49.4% response rate)
Cell 2 8 weeks NRT nⴝ211 (53.0% response rate)
41.7 (11.3) 18.5 56.0 25.0 0.5 66.5 3.6 10.2 35.0 67.5 54.3 43.1 20.9 (10.2) 41.2
43.5 (11.5) 15.6 52.6 29.4 2.4 64.9 4.8 6.8 42.9 64.8 53.6 42.5 21.6 (11.9) 49.5
p value
Test statistic
0.1** 0.3***
⫺1.68 0.0004
0.7* 0.6* 0.2* 0.1* 0.6* 0.9* 0.9* 0.5** 0.09*
0.11 0.36 1.50 2.66 0.34 0.02 0.005 ⫺0.68 2.87
a
Data obtained from 6-month survey, not available from registration data Among those not employed Among those who reported smoking at least 1 cigarette per day *Chi-square statistic **t-test statistic ***Fisher’s exact test (small n in cells) NRT, nicotine replacement therapy
b c
spondents were assumed to be treatment failures, mean cost per quit was $1658 for Cell 1 and $2040 for Cell 2, and an incremental cost effectiveness of $3131 for Cell 2 compared to Cell 1.
Discussion An improved quit rate was found in uninsured callers to a state quitline randomized to receive 8 weeks of NRT compared to 2 weeks. Adjusted intent-to-treat 30-day quit rates at 6 months showed an absolute increase in quit rate of 5.0%, and a 1.45 OR compared to 2 weeks. The absolute increase was sustained for 90-day abstinence, and the OR increased to 1.63. Looking only at
survey responders, the absolute difference in quit rates approached 10%. Thus, for every 10 –20 smokers given 8 weeks rather than 2 weeks, at least one successfully quit who would otherwise have remained a smoker. This “Number Needed to Treat” in order to create a quitter is very favorable compared to other preventive, medical, and surgical interventions.24 The incremental cost effectiveness of additional NRT was comparable to other cessation interventions,25 and is better by orders of magnitude than the cost effectiveness of other commonly provided preventive and healthcare services.26,27 Paying an incremental $1384 –$3131 (Table 2) to create an additional quit provides unusual value for a healthcare
Table 2. Responder versus nonresponder demographics and tobacco history (% unless otherwise specified)
Mean (SD) age in years 18–29 30–49 50–64 ⱖ65 Women Hispanic Nonwhite ⱕhigh school education ⱖ20 years of smoking at registration One or more quit attempts prior to registration Mean (SD) cigarettes/day at registration Reported stage pre/contemplation at registration
Responder nⴝ411
Nonresponder nⴝ392
42.6 (11.4) 17.0 54.3 27.2 1.5 65.7 4.2 8.4 53.9 68.9 93.6 21.2 (11.1) 1.2
40.0 (11.8) 21.7 54.9 22.4 1.0 64.8 5.8 11.1 63.1 61.2 90.6 19.8 (10.0) 1.8
p value
Test statistic value
<0.01** 0.01*
⫺3.12 11.24
0.8* 0.3* 0.2* <0.01* 0.03* 0.1* 0.08** 0.5*
0.07 1.00 1.66 7.01 4.71 2.13 ⫺1.76 0.43
*Chi-square statistic **t-test statistic
August 2008
Am J Prev Med 2008;35(2)
107
Table 3. Six-month quit rates and cost Cell 1 Cell 2 2 weeks NRT % 8 weeks NRT % Absolute % difference OR (CI) p value for chi(#quit/total) (#quit/total) (number needed to treat) square statistic Adjusted intent-to-treat (nonworking phones excluded) 7-day point prevalence 15.6 (63/405) 21.9 (87/398) 30-day abstinence 14.3 (58/405) 19.6 (78/398) 90-day abstinence 10.4 (42/405) 15.8 (63/398) Cost effectiveness Cost per quit (30-day adjusted $1160 $1405 ITT) ICERa (30-day adjusted ITT) NA $2076 Cost per quit (responders⫺ITT) $564–1658 $738–2040 ICERa (respondersⴚITT) NA $1384–$3131
6.3 (16) 5.3 (19) 5.5 (18)
1.52 (1.06–2.17) p⬍0.02 1.45 (1.01–2.12) p⬍0.046 1.63 (1.07–2.47) p⬍0.03
NA
NA
NA NA NA
NA NA NA
ICER ⫽ incremental cost effectiveness ratio, or added cost per added quit for Cell 2 compared to Cell 1 ITT, intent to treat; NRT, nicotine replacement therapy a
intervention, given that quitting adds 3 to 10 years to life expectancy.28 The overall cost per quit for each free NRT intervention was similar to results achieved in Maine ($1344)29 and less than the free-patch program in Minnesota ($1934).30 The cost per quit for New York’s free NRT was reported as less than $500.15 However, the New York results are likely artificially low given that nonresponders were excluded from the analysis and successful quitting was defined as 7-day abstinence at 3–5month follow-up. The New York study comparing different durations of patch therapy provided to nonrandomized groups of smokers did not find significant differences in quit rates.15 These negative findings are likely related to population variations between the nonrandomized groups receiving different NRT regimens. The Oregon trial also provided more in-depth coaching, which may have increased effectiveness. There are important limitations to this study. Lack of a no-patch arm makes it impossible to determine with precision the incremental benefit of 2 weeks of NRT compared to no NRT. An earlier large randomized trial conducted through the ORQL found that the benefits of 8 weeks of NRT added to a two-call intervention with no NRT are substantial, with almost a doubling of the
quit rate at a year (10.7 vs 21.3%).23 However, how much of that benefit could be derived by providing only 2 weeks is unknown. Quit status outcomes were based on self-report without biochemical confirmation. However, others have found a strong correlation between self-reported abstinence and cotinine levels.31–33 Participants are unlikely to misrepresent their smoking status in low-demand settings such as this study, with a low proportion of self-reported quitters testing positive, equally distributed between control and treatment arms.32 The adjusted response rate of around 50% was lower than hoped for. Reconnecting with uninsured callers after a brief intervention proved to be more of a challenge than experienced in insured populations receiving more-intensive services.12,22 Future studies may benefit from larger incentives for survey completion and collection of more collateral contact information for aggressive follow-up tracing. These analyses underestimate cost effectiveness, since some of those not reached likely did quit. The primary outcome was 30-day adjusted intent-totreat abstinence. Some might argue that everyone randomized, including those with nonworking phones, should be analyzed to avoid a responder bias. This approach is appropriate for trials where participants
Table 4. Satisfaction, NRT use, and call completion (responders only)
6-month satisfaction (very and somewhat) Any patch use Got more patches elsewhere Mean # of weeks used Median (range) Mean # of calls completeda Mean minutes of total call contact time (SD)
Cell 1 2 weeks NRT
Cell 2 8 weeks NRT
90.3 (177/196) 86.5 (173/200) 39.3 (68/173) 4.3 2 (0–26) 1.6 22.3 (12.9)
96.7 (202/209) 94.3 (199/211) 16.2 (32/198) 6.3 6 (0–24) 2.0 28.1 (14.0)
p value
Test statistic
p⬍0.01* p⬍0.01* p⬍0.001* p⬍0.001**
6.78 7.21 25.2 ⫺3.46
p⬍0.001** p⬍0.001**
⫺8.71 8.75
a
Includes unscheduled “ad hoc” calls where participant calls in for additional support *Chi-square statistic **t-test statistic NRT, nicotine replacement therapy
108
American Journal of Preventive Medicine, Volume 35, Number 2
www.ajpm-online.net
are closely monitored and easily followed-up. However, in a mobile low-income population with limited study contact this approach is likely to markedly underestimate true cessation rates.34 Alternatively, focusing primarily on survey responders opens the issue of differential response bias. The rate of nonworking phones at follow-up was almost identical between cells, and there were no demographic or tobacco history differences between cells for those with nonworking phones, confirming that nonworking phone status was randomly distributed across cells. Required abstinence duration also varies widely in cessation trials. Drug trials often focus on 7-day point prevalence abstinence, while some cessation trials have required complete abstinence from the time of a quit attempt as their primary outcome.3 Thirty days was chosen as a compromise, because at 7 days, a quitter is still in the active phase of cessation, while continuous abstinence may miss differential treatment recycling effects.18 Because of heterogeneity in how quit rates are measured and reported, a survey respondent analysis and a more-conservative total sample intent-to-treat quit-rate analysis were also conducted. The effect size as measured by the OR (1.40 –1.63) remained similar and significant regardless of the method of analysis. Thus, a clear pattern of increased quit rates for 8 weeks of NRT is seen regardless of who is included and which quit measure is used. This was a real-world effectiveness trial of two different policies for NRT distribution. It was not an efficacy trial. The difference in quit rates between cells was likely created by interplay between pharmacologic drug effects and psychosocial and placebo effects. For example, it is not possible to determine with certainty how much of the improvement was due to a direct medication effect, and how much may have been due to requiring participants to complete the second call in order to receive a refill. Although almost all participants requested and received medication, there was a difference in reported actual patch use (86% in Cell 1, 94% in Cell 2). This initiation effect may have been mediated by increased motivation to use the larger quantity of free medication. In addition, the 2-week cell participants were actively encouraged to purchase more patches, and almost 40% did so. The spread between the reported amount of drug used in each group (4.3 weeks in Cell 1 vs 6.3 weeks in Cell 2) was smaller than would likely be seen in a pure efficacy trial. This may have dampened the difference in quit rates between Cells 1 and 2. Despite the improved effectiveness of 8 weeks NRT, specific policy objectives could justify providing 2 weeks rather than 8 weeks. For example, if a state provides 8 weeks of nicotine patches to all callers, it may unintentionally give private health plans a disincentive from providing coverage. Additionally, a state may wish to August 2008
increase reach and spur quit attempts by making NRT available, but not have funds to provide full treatment to all callers. An alternate NRT delivery model is gaining traction in a number of states, in which for those receiving NRT the quitline functions primarily as a fulfillment house, delivering little or no counseling. It is possible that under such a model less benefit might be derived from additional weeks of NRT, due to weaker medication adherence and lack of behavioral support. Four large randomized trials found that adding counseling improves medication outcomes.18,23,35,36 Future cost-effectiveness studies that include nopatch and 4-week cells, with sufficient sample sizes and aggressive follow-up, would be useful. Examination of 4-week offerings is particularly important in that the new CDC Best Practices for Comprehensive Tobacco Control Programs specifically proposes 4 weeks as an option for the uninsured.37 Examining the effects of requiring coaching contact in order to obtain initial NRT and refills, versus simply shipping all medication at once, is also a question of interest. In summary, these results provide strong support for providing 8 weeks of nicotine patches for uninsured callers to a state quitline. We would like to acknowledge the efforts of Nancy Clarke and Cathryn Cushing in the Tobacco Prevention Education Program, in the Oregon Department of Human Services, for their unstinting support of this study from conception through execution; to Amy Burke, Heidi Grossman, Anne Perez, and Ken Wassum at Free & Clear for their support in study execution; as well as Pam Brooks for her assistance in manuscript preparation. Funding for this trial was provided by the Oregon Department of Human Services. TM owns stock in Free & Clear, and JM and JF received consulting fees from the company. No other financial disclosures were reported by the authors of this paper.
References 1. McAfee TA. Quitlines: a tool for research and dissemination of evidencebased cessation practices. Am J Prev Med 2007;33(6S):S357–S367. 2. Anderson CM, Zhu SH. Tobacco quitlines: looking back and looking ahead. Tob Control 2007;16(1S):i81– 6. 3. Zhu SH, Anderson CM, Tedeschi GJ, et al. Evidence of real-world effectiveness of a telephone quitline for smokers. N Engl J Med 2002;347:1087–93. 4. Stead LF, Perera R, Lancaster T. Telephone counseling for smoking cessation. Cochrane Database Syst Rev 2006;3:CD002850. 5. Ossip-Klein DJ, McIntosh S. Quitlines in North America: evidence base and applications. Am J Med Sci 2003;326:201–5. 6. Lichtenstein E, Glasgow RE, Lando HA, Ossip-Klein DJ, Boles SM. Telephone counseling for smoking cessation: rationales and meta-analytic review of evidence. Health Educ Res 1996;11:243–57. 7. The Guide to Community Preventive Services: Tobacco Use Prevention and Control. Prev Med 2001;20(2S). 8. North American Quitline Consortium. Quitlines of North America and Europe 2006. Phoenix AZ: North American Quitline Consortium. Available at www.naquitline.org.
Am J Prev Med 2008;35(2)
109
9. An LC, Schillo BA, Kavanaugh A, Luxenberg MG, Joseph AM, McAfee T. Access to nicotine replacement therapy as part of a statewide tobacco telephone helpline. Am J Health Promot 2006;20:267–71. 10. Fix B, Carlin S, Celestino P, Hyland A, Higbee C, Cummings KM. Free NRT programs: a report on the reach, efficacy, and cost-effectiveness of NRT give away programs conducted in New York State. Roswell Park Cancer Institute, 2004. 11. Ringen K, Anderson N, McAfee T, Zbikowski SM, Fales D. Smoking cessation in a blue-collar population: results from an evidence-based pilot program. Am J Ind Med 2002;42:367–77. 12. Curry SJ, Grothaus LC, McAfee T, Pabiniak C. Utilization and cost effectiveness of smoking-cessation services under four insurance plans in a health maintenance organization. N Engl J Med 1998;339:673–9. 13. Fellows JL, Bush T, McAfee T, Dickerson J. Cost effectiveness of the Oregon quitline “free patch initiative.” Tob Control 2007;16(1S):i47–52. 14. Bauer JE, Carlin Mentor SM, Celestino PB, Hyland A, Cummings KM. Giving away free nicotine medications and a cigarette substitute (Better Quit) to promote calls to a quitline. J Public Health Manag Pract 2007;12:60 –7. 15. Cummings MK, Fix B, Celestino P, Carline-Menter S, O’Conner R, Hyland A. Reach, efficacy, and cost-effectiveness of free nicotine medication giveaway programs. J Public Health Manag Pract 2006;12:37– 43. 16. Warner D, Rigotti N, Land T, Rakovic AM, Keithly L. Reach, efficacy, and cost-effectiveness of using a state quitline to provide free nicotine patches to smokers in low-income communities. Poster presented at the 13th Annual Meeting of Society for Research on Nicotine or Tobacco; 2007 Feb 21–24; Austin, TX. 17. Cummings MK, Hyland A, Abrams SM, Fix B. How much free nicotine medication should we give smokers calling a quitline? A poster presented at the 13th Annual Meeting of Society for Research on Nicotine or Tobacco; 2007 Feb 21–24; Austin, TX. 18. Swan GE, McAfee T, Jack L, Javitz HS, Dacey S, Bergman K. Effectiveness of buproprion SR for smoking cessation in healthcare setting. Arch Intern Med 2003;163:2337– 44. 19. Javitz HS, Swan GE, Zbikowski SM, et al. Cost-effectiveness of different combinations of bupropion SR dose and behavioral treatment for smoking cessation: a societal perspective. Am J Manag Care 2004;10:217–26. 20. El-Bastawissi A, McAfee T, Zbikowski SM, et al. The uninsured and Medicaid Oregon tobacco user experience in a real world, phone based cessation programme. Tob Control 2003;12:45–51. 21. Bush T, McAfee T, Deprey M, Mahoney L, Fellows JL. The impact of a free nicotine patch starter kit on quit rates in a state quitline. Nicotine Tob Res 2008. In press. 22. Orleans CT, Schoenbach VJ, Wagner EH, et al. Self-help quit smoking interventions: effects of self-help materials, social support instructions, and telephone counseling. J Consult Clin Psychol 1991;59:439 – 48.
110
23. Hollis JF, McAfee TA, Fellows JL, Zbikowski SM, Stark M, Riedlinger K. The effectiveness and cost effectiveness of telephone counseling and the nicotine patch in a state tobacco quitline. Tob Control 2007;16(1S):i53–9. 24. McQuay HJ, Moore AM. Using numerical results from systematic reviews in clinical practice. Ann Intern Med 1997;712–20. 25. Fiore MC, Bailey WC, Cohen SJ, et al. Treating tobacco use and dependence: clinical practice guidelines. Rockville MD: USDHHS, Department of Public Health Service, 2000. 26. Cromwell J, Bartosch WJ, Fiore MC, et al. Cost-effectiveness of the clinical practice recommendations in the AHCPR guideline for smoking cessation. Agency for Health Care Policy and Research. JAMA 1997;278:1759 – 66. 27. Warner KE. Cost-effectiveness of smoking cessation therapies: interpretation of the evidence and implications for coverage. Pharmacogenomics J 1997;11:538 – 49. 28. Doll R, Peto R, Boreham J, Sutherland I. Mortality in relation to smoking: 50 years’ observations on male British doctors. BMJ 2004;1519. 29. Swartz SH, Cowan TM, Klayman JE, Welton, MT, Leonard BA. Use and effectiveness of tobacco telephone counseling and nicotine therapy in Maine. Am J Prev Med 2005;288 –94. 30. An LC, Schillo BA, Kavanaugh AM, et al. Increased reach and effectiveness of a statewide tobacco quitline after the addition of access to free nicotine replacement therapy. Tob Control 2006;15:286 –93. 31. Patrick DL, Cheadle A, Thompson DC, Diehr P, Koesell T, Klinne S. The validity of self-reported smoking: a review and meta-analysis. Am J Public Health 1994;84:1086 –93. 32. Glasgow RE, Mullooly JP, Vogt TM, et al. Biochemical validation of smoking status: pros, cons, and data from four low-intensity intervention trials. Addict Behav 1993;18:511–27. 33. SRNT Subcommittee on Biochemical verification. Biochemical verification of tobacco use and cessation. Nicotine Tob Res 2002;4:149 –59. 34. Tomson T, Bjonstrom C, Gilljam H, Helgason AR. Are non-responders in a quitline evaluation more likely to be smokers? BMC Public Health 2005;5:52. 35. Macleod ZR, Charles MA, Arnaldi VC, Adams IM. Telephone counselling as an adjunct to nicotine patches in smoking cessation: a randomised controlled trial. Med J Aust 2003;179:349 –52. 36. Zhu S, Tedeschi GJ, Anderson CM, et al. Telephone counseling as adjuvant treatments for nicotine replacement therapy in a “real world” setting. Am J Prev Med 2000;31:357– 63. 37. Centers for Disease Control and Prevention. Best Practices for Comprehensive Tobacco Control Programs—2007. Atlanta: USDHHS, CDC, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, 2007.
American Journal of Preventive Medicine, Volume 35, Number 2
www.ajpm-online.net
A two-cell randomized trial comparing 2 weeks of NRT to 8 weeks.
Setting/ Uninsured callers to the Oregon Quit Line during a free-patch initiative from October 18, participants: 2004, to May 5, 2005, who were 18 years or older, smoked five or more cigarettes per day, did not have a medical contraindication to NRT use, and were interested in quitting in 30 days. Data were collected from April to November 2005, and analyzed in 2006 –2007. Intervention: Participants were eligible for two phone counseling sessions. 1154 participants were randomized to receive via the mail either 2 or 8 weeks of nicotine patches. Measures:
Primary outcome was self-reported complete abstinence from tobacco for 30 or more days at the 6-month phone survey. Secondary outcomes were 7-day point prevalence and 90-day abstinence, satisfaction, and patch use. ORs and CIs were computed. Cost per quit and incremental cost per additional quit were computed based on program costs.
Results:
Intent-to-treat 30-day abstinence was 14.3% in the 2-week group, and 19.6% in the 8-week group (OR 1.45 [CI⫽1.01, 2.12]). Average cost per quit was $1156 for 2 weeks and $1405 for 8 weeks, with an incremental cost effectiveness of $2068. Satisfaction increased from 90% to 97% with 8 weeks. Those receiving 8 weeks of NRT took more calls (2.0 vs 1.6) and used more patches (6.3 weeks vs 4.3 weeks), but were less likely to purchase patches (16.2% vs 39.3%).
Conclusions: Eight weeks of patches improved quit rates compared with 2 weeks, and was cost effective. (Am J Prev Med 2008;35(2):103–110) © 2008 American Journal of Preventive Medicine
Introduction
S
tate-level tobacco quitlines grew rapidly in the past decade, with over half a million people accessing evidence-based cessation treatment through them in 2005.1–3 A strong evidence base supports quitline efficacy and effectiveness, including multiple metaanalyses.4 –7 Nicotine replacement therapy (NRT) is being rapidly integrated into state tobacco quitlines in the U.S., Canada, Europe, and Australia/New Zealand. As of 2006, over 20 states reported some form of NRT distribution, with most either time-limited, populationlimited, or involving less-than-recommended durations of therapy.8 NRT has been introduced for two main From the Department of Health Services, School of Public Health, University of Washington (McAfee); Clinical and Behavioral Sciences, Free & Clear (McAfee, Bush, Deprey, Mahoney, Zbikowski); Group Health Center for Health Studies (McClure), Seattle, Washington; and the Center for Health Research, Kaiser Permanente Northwest (Fellows), Portland, Oregon Address correspondence and reprint requests to: Timothy A. McAfee, MD, MPH, Free & Clear, 999 Third Avenue, Suite 2100, Seattle WA 98104. E-mail: [email protected].
purposes: (1) as a promotional aid increasing calls to the quitline; and, (2) to improve quit rates. Several evaluations have established that offering free NRT will result in dramatic increases in calls to state-level,9,10 employer/union-based,11 and health plan12 quitlines. Using NRT as a means of promotion may be considerably more cost effective than the more traditional use of mass media.13 Because of limitations in state funding, a number of quitlines have been experimenting with courses of NRT that are considerably shorter than the manufacturer’s and U.S. Food and Drug Administration (FDA)’s recommendations (8 –12 weeks).14 Courses of 2 and 4 weeks are common, and even 1 week has been offered.15 The amount of NRT offered seems to be a minimal issue in terms of creating a promotional impact (Deprey M, et al., Free & Clear, unpublished finding).16 However, much less is known about the impact of shorter courses on quit rates. Cummings et al.17 compared different amounts of NRT and their effect on quit rates, and did not find a significant difference between 2 weeks and 8 weeks.
Am J Prev Med 2008;35(2) © 2008 American Journal of Preventive Medicine • Published by Elsevier Inc.
0749-3797/08/$–see front matter doi:10.1016/j.amepre.2008.04.017
103
However, the comparison groups were not randomized, had variable response rates, and were not comparable populations, with different groups varying in insurance status, geographic location, and recruitment method. Determining the impact of offering 2 weeks of NRT rather than 8 weeks on quit outcomes and satisfaction has financial, clinical, and theoretical importance. If additional NRT does not improve quit rates, providing shorter courses could free up additional financial resources to treat more people, provide more-intensive services such as proactive counseling calls, or provide increased promotion. For example, twice-a-day dosing of bupropion SR does not cost effectively increase 1-year quit rates compared to once-a-day, but proactive phone calls do when provided via a health plan quitline.18,19 However, if increased duration of NRT does improve quit rates, then policies developed assuming it does not may decrease an individual’s chances of succeeding. It is also expensive to drive calls to a state quitline; thus, providing less-optimal service may actually not be cost effective from a systems perspective. The uninsured use state quitlines disproportionately,20 and have fewer options for accessing evidence-based cessation support. Finally, if 2 weeks works as well as 8 weeks, the primary mechanism of action for NRT may be its mitigation of short-term withdrawal symptoms, and its role in long-term relapse prevention may be less important.15 The state of Oregon conducted a time-limited campaign in which all tobacco users in the state were eligible to receive free nicotine patches. The primary purpose of the overall campaign was to test the impact of the availability of free patches on call volumes to the Oregon Quit Line (ORQL). All residents were eligible for 2 weeks of patches during the campaign. This amount was chosen due to funding limitations and to avoid giving the health plans a disincentive from providing NRT support for their insured patients. Pre–post comparison in insured callers of quit outcomes,21 cost effectiveness,13 and promotional impact (Deprey M, et al., unpublished finding) are reported elsewhere. Because uninsured callers do not have any possibility of receiving additional NRT from a health plan, the state of Oregon was interested in determining what added benefit 8 weeks provided compared to 2 weeks. This paper presents results of a randomized trial for uninsured callers embedded in this larger initiative, comparing 2 weeks of NRT versus 8 weeks, with both groups getting two telephone counseling calls. Six-month quit rates, use patterns, satisfaction, and the cost effectiveness of additional NRT are reported.
Methods The study design was reviewed and approved on September 15, 2004, by the Western IRB and on September 24, 2004, by
104
the Department of Health Services, Public Health Division/ Multnomah County Public Health IRB. Data were collected from April to November 2005, and analyzed in 2006 –2007.
Eligibility Callers to the ORQL were eligible for this study if they called between October 18, 2004, and May 5, 2005, and were (1) uninsured, (2) 18 years or older, (3) spoke English, (4) had a working phone number, (5) smoked five or more cigarettes per day, (6) were interested in quitting within 30 days, and (7) provided informed consent over the telephone including consenting to randomization to receive different service offerings and to an additional studyrelated follow-up survey call. Eligibility also required callers to have an interest in using NRT and be medically appropriate for NRT based on FDA criteria.
Recruitment, Consent, and Randomization Availability of free nicotine patches was promoted via a press conference covered by most television, radio, and newspapers in the state, as well as through announcements to community and healthcare provider organizations. Word-of-mouth promotion was encouraged by inclusion of referral cards in materials sent to callers. Callers to the ORQL were screened initially by registration staff to determine eligibility and interest. Eligible, interested callers were transferred to a cessation counselor and formally consented. The consent process included informing potential participants that if they participated in the study, they would receive 2 weeks of NRT and a second counseling call (in addition to the standard service single call), and that some participants would be randomly assigned to receive up to 8 weeks of NRT. After consent, the counselor randomized the participant, using an application embedded in the counseling support database. Randomization occurred in blocks of 12 so that after every 12th person recruited the cells balanced.
Intervention Participants in both cells were eligible for two phone counseling sessions, and could also call in on an ad hoc basis. The first counseling session occurred immediately after randomization, and the second was scheduled for the week after their quit date. Counselors’ training included techniques to manage call length, but time limits were not set. Content of the phone counseling sessions was based on recommendations of the U.S. Public Health Service Clinical Practice Guideline. The two calls were an abbreviated version of a five-call program (Quit for Life™) found to be effective in multiple randomized trials.18,22,23 Counselors also provided education on the proper use of NRT, including side-effect management and ongoing use. Experienced telephone tobacco counselors provided the interventions, with computer-driven protocol support. A subset of calls was taped and reviewed with feedback to the counselors for quality assurance monitoring. All participants received identical printed support materials. Participants randomized to Cell 1 received a 2-week supply of patches. Individuals were mailed either 21 mg if the participant smoked more than ten cigarettes a day, or 14 mg if they smoked ten cigarettes a day or less, as recommended by the FDA. Participants were encouraged to save money not spent on cigarettes, in order to purchase additional patches
American Journal of Preventive Medicine, Volume 35, Number 2
www.ajpm-online.net
(a cut-out cardboard “piggy-bank” was included in the mailing). Participants randomized to Cell 2 were sent an initial 4-week supply of patches and were eligible to receive a refill including FDA-recommended lower doses for a second four weeks. Participants had to complete their second call, or call back, in order to receive a refill. Counseling and medication protocols are described in detail elsewhere.20,23
(OR) with confidence intervals for quit rates between Cells 1 and 2 were calculated. Reported patch use in each cell was tested for trend using the Cochran–Armitage test for trend, which tests for trend in binomial proportions (quit or not-quit) across levels of a single factor (patch use). Patch use was divided into 0 to ⬍1 week, 1–2 weeks, 3– 4 weeks, 5– 6 weeks, and 7⫹ weeks. The analysis was generated using SAS/STAT® Software 9.1.3.
Data Collection
Cost Effectiveness
Basic demographic information was collected during the routine registration process at the initial call. Participants in the study were called back 6 months after registration by an independent survey research firm, Marketing Strategies, Inc. Two weeks before the call window opened, a letter was sent reminding the participant of the upcoming call, and that they would receive $5 for completing the survey. For those who still had working phone numbers, 12 attempts were made over a 4-week window. The window opened 2 weeks before the due date of the call and closed 2 weeks after the due date. Responders were asked about their smoking status, quitline and additional service use, satisfaction with quitline services, experience with patches and other medications, and out-of-pocket expenses for additional cessation services and products.
The cost of each intervention strategy, average cost per quit, and the cost effectiveness of counseling plus 8 weeks of NRT compared to counseling plus 2 weeks of NRT was estimated assuming a state-program perspective. The program perspective included state-paid intervention costs for telephone counseling, mailed self-help (quit kit) materials, and NRT. Quitting was defined as 30-day abstinence at 6 months assuming adjusted intent-to-treat (as described above). Cost per quit and incremental cost-effectiveness results were also estimated for responders only and for all sampled callers (even those without valid phone numbers). For the latter analysis, it was assumed all callers not reached were treatment failures. This provided a high and low range for intervention cost effectiveness. The incremental cost per additional quit of counseling with 8 weeks of NRT compared to counseling with 2 weeks of NRT was estimated by dividing the difference in intervention costs by the difference in quit rates. Quitline promotion costs during the initiative were the same for each strategy and were excluded from the cost-effectiveness analysis.
Outcome Measures and Analysis The primary outcome was self-reported complete abstinence from tobacco for 30 or more days at the 6-month survey. Six or more months is the North American Quitline Consortium’s recommended time interval for telephone quitline quit rate follow-up.8 The primary analysis used for costeffectiveness computation was based on an adjusted intent-totreat analysis where those randomized to receive treatment, regardless of whether they actually completed the two calls or used the patches, were included in the analysis. Those with working phone numbers who could not be contacted were included in the analysis, imputed conservatively to be smoking. However, it was decided a priori to exclude those who had nonworking phone numbers at 6-month follow-up from the primary analysis. This was done for three reasons: (1) study participants were uninsured, with a high rate of mobility and phone disconnects20; (2) phone disconnect rates are unlikely to be related to quit status; and (3) the study intervention involved only two sessions of brief phone-only contact conducted in a real-world environment with minimal opportunity to update contact information. Because there is wide variation in how quit status is reported in trials, 7-day point-prevalence and 90-day continuous abstinence at 6 months were also measured. In addition to the adjusted intent-to-treat quit-rate calculations, 7-, 30-, and 90-day quit rates were computed for survey respondents only, and for the total random sample (intent-to-treat analysis including those without working phone numbers imputed to be smokers). Satisfaction and patterns of NRT use were compared based on those responding to the survey. Thirty-day adjusted intentto-treat quit rates were compared between the two cells based on reported use or non-use of NRT. Chi-square two-tailed comparisons for statistical significance were made. Fisher’s exact test was used when a comparison included a cell with less than five counts. Odds ratios
August 2008
Results The overall response to the free-patch initiative is detailed elsewhere.13,21 Briefly, the initiative earned media attention, influenced six health plans to change their cessation benefits, and increased overall calls to the ORQL to greater than 10,000 calls in the first month (a 20-fold increase over baseline), with a continued doubling of baseline call volumes for 6 months after the public campaign of free NRT to all residents was concluded. There were 2934 uninsured callers registered with the quitline during the study period, of a total 10,596 registrants (27.7% compared to state uninsured prevalence of 17%); of those, 2092 were formally assessed for study eligibility by the program staff (Figure 1), and 1154 were randomized. The 6-month uninsured survey response rate was 49.4% for those offered 2 weeks of NRT (Cell 1) and 53.0% for those offered 8 weeks (Cell 2) (p⬍0.30), excluding those with nonworking phone numbers. Cell 1 and Cell 2 participants completing the survey did not differ significantly on demographics and tobacco history (Table 1). There was no difference between Cells 1 and 2 in the rate of nonworking phones (30% vs 31%, p⬍0.72). Those with nonworking phones also showed no significant differences for any baseline demographics and tobacco history between Cell 1 and Cell 2. There were significant differences across the cells beAm J Prev Med 2008;35(2)
105
Analysis
Follow-up
Allocation
Enrollmennt
tween those with working 7879 initially excluded and nonworking phones. 10,596 Oregon Quit Line 7662 insured Those with working registrants during study 217 uninsured enrollment window phones (both respond67 smoked <5 ers and nonresponders) cigarettes per day 59 non-English were more likely to be 625 not screeneda speaking older (41.3 vs 36.1, 91 not ready to quit ⫺7.16, p⬍0.001); women (65.3% vs 54.7%, 11.55, 2092 uninsured p⬍0.001); and white (90.2 938 not enrolled screened 300 declined enrollment vs 85.9, 4.61, p⬍0.03) 55 medical contrathan those with nonworkindication for NRT use ing phones. Having less 1154 randomized 583 recruited to a than a high school educaseparate study tion did not affect phone status. 578 received 2 weeks of NRT 576 received 8 weeks of NRT Baseline characteris576 NRT mailed 575 NRT mailed tics of responders and 2 NRT not mailed 1 NRT not mailed nonresponders were compared (Table 2). Responders were 2 576 total population ITT 578 total population ITT years older on average, more likely to be high school graduates, and 405 adjusted ITT 398 adjusted ITT more likely to have 200 responders 211 responders smoked more than 20 205 nonresponders 187 nonresponders years. 173 unreachable (no working 178 unreachable (no working Randomization to rephone) phone) ceive 8 weeks of NRT (Cell 2) increased the Figure 1. Participant flow diagram a Due to high call volumes during intense service delivery periods OR for being quit at 6 months for 30 days or ITT, intent to treat; NRT, nicotine replacement therapy more to 1.45 for adreported at follow-up actually using the patch were justed intent-to-treat. The OR for being quit for 7 days much more likely to quit for 30 or more days than those was 1.52, and for 90 days was 1.63 (Table 3). Including who did not use the patch. In Cell 1, 3/27 non–patch only responders in the analysis produced ORs of 1.43 to users (11%) versus 55/170 patch users (32%) were quit 1.59 (depending on quit definition), with absolute for 30 or more days at 6 months (p⬍0.03). In Cell 2, increases of 8% to 10%. Including those with nonwork1/12 non–patch users (8%) versus 77/197 patch users ing phones counted as smokers (classic intent-to-treat), (39%) were quit for 30 or more days (p⬍0.04). In both produced ORs ranging from 1.40 to 1.56. cells, quit rate increased with amount of patch used, Those randomized to receive 8 weeks were more likely to ranging in Cell 1 from 11% for those using zero to less be satisfied, and more likely to actually use patches (Table 4); than a week of patches to 57% for those using ⱖ7 weeks 66% of participants randomized to Cell 2 requested and (p⬍0.0001, Cochran–Armitage trend test, ⫺4.9146), were sent a second 4-week shipment of NRT. Survey reand in Cell 2 ranging from 12% to 48% (p⬍0.0001, sponders received the second shipment at a higher rate than ⫺3.7532). nonresponders (72% vs 58%, p⬍0.03, chi-square 8.93). Intervention costs per participant were $165.82 for Among responders, those who got the second shipment counseling and 2 weeks of NRT (Cell 1) and $275.40 were much more likely to be quit for 30 days or more at 6 for counseling and 8 weeks of NRT (Cell 2). These costs months than responders who did not (46% vs 15%, adjusted reflect actual NRT delivery. Assuming 30-day abstiintent-to-treat, p⬍0.0001, chi-square 17.1). nence at 6 months (based on adjusted intent-to-treat), Those randomized to receive only 2 weeks of NRT estimated average costs per quit were estimated at were almost 2.5 times more likely to purchase more $1156 for Cell 1 and $1405 for Cell 2. Each additional NRT (Table 4). quit for Cell 2 cost $2068 compared to Cell 1. For Although NRT use was not a precondition for particresponders only, mean costs per quit were $564 for Cell ipation, only three participants were not sent NRT (two 1 and $738 for Cell 2. Incremental cost effectiveness for in Cell 1, one in Cell 2), due to lack of interest. They Cell 2 was $1384 compared to Cell 1. When all nonrewere included in all analyses. Those in both cells who 106
American Journal of Preventive Medicine, Volume 35, Number 2
www.ajpm-online.net
Table 1. Demographic characteristics of responders by cell (% unless otherwise specified)
Mean (SD) age in years 18–29 30–49 50–64 ⱖ65 Women Hispanic Nonwhite Marrieda Employed full- or part-timea ⱕhigh school education Unable to worka,b Mean (SD) cigarettes/day at registrationc Report prior use of patches
Cell 1 2 weeks NRT nⴝ200 (49.4% response rate)
Cell 2 8 weeks NRT nⴝ211 (53.0% response rate)
41.7 (11.3) 18.5 56.0 25.0 0.5 66.5 3.6 10.2 35.0 67.5 54.3 43.1 20.9 (10.2) 41.2
43.5 (11.5) 15.6 52.6 29.4 2.4 64.9 4.8 6.8 42.9 64.8 53.6 42.5 21.6 (11.9) 49.5
p value
Test statistic
0.1** 0.3***
⫺1.68 0.0004
0.7* 0.6* 0.2* 0.1* 0.6* 0.9* 0.9* 0.5** 0.09*
0.11 0.36 1.50 2.66 0.34 0.02 0.005 ⫺0.68 2.87
a
Data obtained from 6-month survey, not available from registration data Among those not employed Among those who reported smoking at least 1 cigarette per day *Chi-square statistic **t-test statistic ***Fisher’s exact test (small n in cells) NRT, nicotine replacement therapy
b c
spondents were assumed to be treatment failures, mean cost per quit was $1658 for Cell 1 and $2040 for Cell 2, and an incremental cost effectiveness of $3131 for Cell 2 compared to Cell 1.
Discussion An improved quit rate was found in uninsured callers to a state quitline randomized to receive 8 weeks of NRT compared to 2 weeks. Adjusted intent-to-treat 30-day quit rates at 6 months showed an absolute increase in quit rate of 5.0%, and a 1.45 OR compared to 2 weeks. The absolute increase was sustained for 90-day abstinence, and the OR increased to 1.63. Looking only at
survey responders, the absolute difference in quit rates approached 10%. Thus, for every 10 –20 smokers given 8 weeks rather than 2 weeks, at least one successfully quit who would otherwise have remained a smoker. This “Number Needed to Treat” in order to create a quitter is very favorable compared to other preventive, medical, and surgical interventions.24 The incremental cost effectiveness of additional NRT was comparable to other cessation interventions,25 and is better by orders of magnitude than the cost effectiveness of other commonly provided preventive and healthcare services.26,27 Paying an incremental $1384 –$3131 (Table 2) to create an additional quit provides unusual value for a healthcare
Table 2. Responder versus nonresponder demographics and tobacco history (% unless otherwise specified)
Mean (SD) age in years 18–29 30–49 50–64 ⱖ65 Women Hispanic Nonwhite ⱕhigh school education ⱖ20 years of smoking at registration One or more quit attempts prior to registration Mean (SD) cigarettes/day at registration Reported stage pre/contemplation at registration
Responder nⴝ411
Nonresponder nⴝ392
42.6 (11.4) 17.0 54.3 27.2 1.5 65.7 4.2 8.4 53.9 68.9 93.6 21.2 (11.1) 1.2
40.0 (11.8) 21.7 54.9 22.4 1.0 64.8 5.8 11.1 63.1 61.2 90.6 19.8 (10.0) 1.8
p value
Test statistic value
<0.01** 0.01*
⫺3.12 11.24
0.8* 0.3* 0.2* <0.01* 0.03* 0.1* 0.08** 0.5*
0.07 1.00 1.66 7.01 4.71 2.13 ⫺1.76 0.43
*Chi-square statistic **t-test statistic
August 2008
Am J Prev Med 2008;35(2)
107
Table 3. Six-month quit rates and cost Cell 1 Cell 2 2 weeks NRT % 8 weeks NRT % Absolute % difference OR (CI) p value for chi(#quit/total) (#quit/total) (number needed to treat) square statistic Adjusted intent-to-treat (nonworking phones excluded) 7-day point prevalence 15.6 (63/405) 21.9 (87/398) 30-day abstinence 14.3 (58/405) 19.6 (78/398) 90-day abstinence 10.4 (42/405) 15.8 (63/398) Cost effectiveness Cost per quit (30-day adjusted $1160 $1405 ITT) ICERa (30-day adjusted ITT) NA $2076 Cost per quit (responders⫺ITT) $564–1658 $738–2040 ICERa (respondersⴚITT) NA $1384–$3131
6.3 (16) 5.3 (19) 5.5 (18)
1.52 (1.06–2.17) p⬍0.02 1.45 (1.01–2.12) p⬍0.046 1.63 (1.07–2.47) p⬍0.03
NA
NA
NA NA NA
NA NA NA
ICER ⫽ incremental cost effectiveness ratio, or added cost per added quit for Cell 2 compared to Cell 1 ITT, intent to treat; NRT, nicotine replacement therapy a
intervention, given that quitting adds 3 to 10 years to life expectancy.28 The overall cost per quit for each free NRT intervention was similar to results achieved in Maine ($1344)29 and less than the free-patch program in Minnesota ($1934).30 The cost per quit for New York’s free NRT was reported as less than $500.15 However, the New York results are likely artificially low given that nonresponders were excluded from the analysis and successful quitting was defined as 7-day abstinence at 3–5month follow-up. The New York study comparing different durations of patch therapy provided to nonrandomized groups of smokers did not find significant differences in quit rates.15 These negative findings are likely related to population variations between the nonrandomized groups receiving different NRT regimens. The Oregon trial also provided more in-depth coaching, which may have increased effectiveness. There are important limitations to this study. Lack of a no-patch arm makes it impossible to determine with precision the incremental benefit of 2 weeks of NRT compared to no NRT. An earlier large randomized trial conducted through the ORQL found that the benefits of 8 weeks of NRT added to a two-call intervention with no NRT are substantial, with almost a doubling of the
quit rate at a year (10.7 vs 21.3%).23 However, how much of that benefit could be derived by providing only 2 weeks is unknown. Quit status outcomes were based on self-report without biochemical confirmation. However, others have found a strong correlation between self-reported abstinence and cotinine levels.31–33 Participants are unlikely to misrepresent their smoking status in low-demand settings such as this study, with a low proportion of self-reported quitters testing positive, equally distributed between control and treatment arms.32 The adjusted response rate of around 50% was lower than hoped for. Reconnecting with uninsured callers after a brief intervention proved to be more of a challenge than experienced in insured populations receiving more-intensive services.12,22 Future studies may benefit from larger incentives for survey completion and collection of more collateral contact information for aggressive follow-up tracing. These analyses underestimate cost effectiveness, since some of those not reached likely did quit. The primary outcome was 30-day adjusted intent-totreat abstinence. Some might argue that everyone randomized, including those with nonworking phones, should be analyzed to avoid a responder bias. This approach is appropriate for trials where participants
Table 4. Satisfaction, NRT use, and call completion (responders only)
6-month satisfaction (very and somewhat) Any patch use Got more patches elsewhere Mean # of weeks used Median (range) Mean # of calls completeda Mean minutes of total call contact time (SD)
Cell 1 2 weeks NRT
Cell 2 8 weeks NRT
90.3 (177/196) 86.5 (173/200) 39.3 (68/173) 4.3 2 (0–26) 1.6 22.3 (12.9)
96.7 (202/209) 94.3 (199/211) 16.2 (32/198) 6.3 6 (0–24) 2.0 28.1 (14.0)
p value
Test statistic
p⬍0.01* p⬍0.01* p⬍0.001* p⬍0.001**
6.78 7.21 25.2 ⫺3.46
p⬍0.001** p⬍0.001**
⫺8.71 8.75
a
Includes unscheduled “ad hoc” calls where participant calls in for additional support *Chi-square statistic **t-test statistic NRT, nicotine replacement therapy
108
American Journal of Preventive Medicine, Volume 35, Number 2
www.ajpm-online.net
are closely monitored and easily followed-up. However, in a mobile low-income population with limited study contact this approach is likely to markedly underestimate true cessation rates.34 Alternatively, focusing primarily on survey responders opens the issue of differential response bias. The rate of nonworking phones at follow-up was almost identical between cells, and there were no demographic or tobacco history differences between cells for those with nonworking phones, confirming that nonworking phone status was randomly distributed across cells. Required abstinence duration also varies widely in cessation trials. Drug trials often focus on 7-day point prevalence abstinence, while some cessation trials have required complete abstinence from the time of a quit attempt as their primary outcome.3 Thirty days was chosen as a compromise, because at 7 days, a quitter is still in the active phase of cessation, while continuous abstinence may miss differential treatment recycling effects.18 Because of heterogeneity in how quit rates are measured and reported, a survey respondent analysis and a more-conservative total sample intent-to-treat quit-rate analysis were also conducted. The effect size as measured by the OR (1.40 –1.63) remained similar and significant regardless of the method of analysis. Thus, a clear pattern of increased quit rates for 8 weeks of NRT is seen regardless of who is included and which quit measure is used. This was a real-world effectiveness trial of two different policies for NRT distribution. It was not an efficacy trial. The difference in quit rates between cells was likely created by interplay between pharmacologic drug effects and psychosocial and placebo effects. For example, it is not possible to determine with certainty how much of the improvement was due to a direct medication effect, and how much may have been due to requiring participants to complete the second call in order to receive a refill. Although almost all participants requested and received medication, there was a difference in reported actual patch use (86% in Cell 1, 94% in Cell 2). This initiation effect may have been mediated by increased motivation to use the larger quantity of free medication. In addition, the 2-week cell participants were actively encouraged to purchase more patches, and almost 40% did so. The spread between the reported amount of drug used in each group (4.3 weeks in Cell 1 vs 6.3 weeks in Cell 2) was smaller than would likely be seen in a pure efficacy trial. This may have dampened the difference in quit rates between Cells 1 and 2. Despite the improved effectiveness of 8 weeks NRT, specific policy objectives could justify providing 2 weeks rather than 8 weeks. For example, if a state provides 8 weeks of nicotine patches to all callers, it may unintentionally give private health plans a disincentive from providing coverage. Additionally, a state may wish to August 2008
increase reach and spur quit attempts by making NRT available, but not have funds to provide full treatment to all callers. An alternate NRT delivery model is gaining traction in a number of states, in which for those receiving NRT the quitline functions primarily as a fulfillment house, delivering little or no counseling. It is possible that under such a model less benefit might be derived from additional weeks of NRT, due to weaker medication adherence and lack of behavioral support. Four large randomized trials found that adding counseling improves medication outcomes.18,23,35,36 Future cost-effectiveness studies that include nopatch and 4-week cells, with sufficient sample sizes and aggressive follow-up, would be useful. Examination of 4-week offerings is particularly important in that the new CDC Best Practices for Comprehensive Tobacco Control Programs specifically proposes 4 weeks as an option for the uninsured.37 Examining the effects of requiring coaching contact in order to obtain initial NRT and refills, versus simply shipping all medication at once, is also a question of interest. In summary, these results provide strong support for providing 8 weeks of nicotine patches for uninsured callers to a state quitline. We would like to acknowledge the efforts of Nancy Clarke and Cathryn Cushing in the Tobacco Prevention Education Program, in the Oregon Department of Human Services, for their unstinting support of this study from conception through execution; to Amy Burke, Heidi Grossman, Anne Perez, and Ken Wassum at Free & Clear for their support in study execution; as well as Pam Brooks for her assistance in manuscript preparation. Funding for this trial was provided by the Oregon Department of Human Services. TM owns stock in Free & Clear, and JM and JF received consulting fees from the company. No other financial disclosures were reported by the authors of this paper.
References 1. McAfee TA. Quitlines: a tool for research and dissemination of evidencebased cessation practices. Am J Prev Med 2007;33(6S):S357–S367. 2. Anderson CM, Zhu SH. Tobacco quitlines: looking back and looking ahead. Tob Control 2007;16(1S):i81– 6. 3. Zhu SH, Anderson CM, Tedeschi GJ, et al. Evidence of real-world effectiveness of a telephone quitline for smokers. N Engl J Med 2002;347:1087–93. 4. Stead LF, Perera R, Lancaster T. Telephone counseling for smoking cessation. Cochrane Database Syst Rev 2006;3:CD002850. 5. Ossip-Klein DJ, McIntosh S. Quitlines in North America: evidence base and applications. Am J Med Sci 2003;326:201–5. 6. Lichtenstein E, Glasgow RE, Lando HA, Ossip-Klein DJ, Boles SM. Telephone counseling for smoking cessation: rationales and meta-analytic review of evidence. Health Educ Res 1996;11:243–57. 7. The Guide to Community Preventive Services: Tobacco Use Prevention and Control. Prev Med 2001;20(2S). 8. North American Quitline Consortium. Quitlines of North America and Europe 2006. Phoenix AZ: North American Quitline Consortium. Available at www.naquitline.org.
Am J Prev Med 2008;35(2)
109
9. An LC, Schillo BA, Kavanaugh A, Luxenberg MG, Joseph AM, McAfee T. Access to nicotine replacement therapy as part of a statewide tobacco telephone helpline. Am J Health Promot 2006;20:267–71. 10. Fix B, Carlin S, Celestino P, Hyland A, Higbee C, Cummings KM. Free NRT programs: a report on the reach, efficacy, and cost-effectiveness of NRT give away programs conducted in New York State. Roswell Park Cancer Institute, 2004. 11. Ringen K, Anderson N, McAfee T, Zbikowski SM, Fales D. Smoking cessation in a blue-collar population: results from an evidence-based pilot program. Am J Ind Med 2002;42:367–77. 12. Curry SJ, Grothaus LC, McAfee T, Pabiniak C. Utilization and cost effectiveness of smoking-cessation services under four insurance plans in a health maintenance organization. N Engl J Med 1998;339:673–9. 13. Fellows JL, Bush T, McAfee T, Dickerson J. Cost effectiveness of the Oregon quitline “free patch initiative.” Tob Control 2007;16(1S):i47–52. 14. Bauer JE, Carlin Mentor SM, Celestino PB, Hyland A, Cummings KM. Giving away free nicotine medications and a cigarette substitute (Better Quit) to promote calls to a quitline. J Public Health Manag Pract 2007;12:60 –7. 15. Cummings MK, Fix B, Celestino P, Carline-Menter S, O’Conner R, Hyland A. Reach, efficacy, and cost-effectiveness of free nicotine medication giveaway programs. J Public Health Manag Pract 2006;12:37– 43. 16. Warner D, Rigotti N, Land T, Rakovic AM, Keithly L. Reach, efficacy, and cost-effectiveness of using a state quitline to provide free nicotine patches to smokers in low-income communities. Poster presented at the 13th Annual Meeting of Society for Research on Nicotine or Tobacco; 2007 Feb 21–24; Austin, TX. 17. Cummings MK, Hyland A, Abrams SM, Fix B. How much free nicotine medication should we give smokers calling a quitline? A poster presented at the 13th Annual Meeting of Society for Research on Nicotine or Tobacco; 2007 Feb 21–24; Austin, TX. 18. Swan GE, McAfee T, Jack L, Javitz HS, Dacey S, Bergman K. Effectiveness of buproprion SR for smoking cessation in healthcare setting. Arch Intern Med 2003;163:2337– 44. 19. Javitz HS, Swan GE, Zbikowski SM, et al. Cost-effectiveness of different combinations of bupropion SR dose and behavioral treatment for smoking cessation: a societal perspective. Am J Manag Care 2004;10:217–26. 20. El-Bastawissi A, McAfee T, Zbikowski SM, et al. The uninsured and Medicaid Oregon tobacco user experience in a real world, phone based cessation programme. Tob Control 2003;12:45–51. 21. Bush T, McAfee T, Deprey M, Mahoney L, Fellows JL. The impact of a free nicotine patch starter kit on quit rates in a state quitline. Nicotine Tob Res 2008. In press. 22. Orleans CT, Schoenbach VJ, Wagner EH, et al. Self-help quit smoking interventions: effects of self-help materials, social support instructions, and telephone counseling. J Consult Clin Psychol 1991;59:439 – 48.
110
23. Hollis JF, McAfee TA, Fellows JL, Zbikowski SM, Stark M, Riedlinger K. The effectiveness and cost effectiveness of telephone counseling and the nicotine patch in a state tobacco quitline. Tob Control 2007;16(1S):i53–9. 24. McQuay HJ, Moore AM. Using numerical results from systematic reviews in clinical practice. Ann Intern Med 1997;712–20. 25. Fiore MC, Bailey WC, Cohen SJ, et al. Treating tobacco use and dependence: clinical practice guidelines. Rockville MD: USDHHS, Department of Public Health Service, 2000. 26. Cromwell J, Bartosch WJ, Fiore MC, et al. Cost-effectiveness of the clinical practice recommendations in the AHCPR guideline for smoking cessation. Agency for Health Care Policy and Research. JAMA 1997;278:1759 – 66. 27. Warner KE. Cost-effectiveness of smoking cessation therapies: interpretation of the evidence and implications for coverage. Pharmacogenomics J 1997;11:538 – 49. 28. Doll R, Peto R, Boreham J, Sutherland I. Mortality in relation to smoking: 50 years’ observations on male British doctors. BMJ 2004;1519. 29. Swartz SH, Cowan TM, Klayman JE, Welton, MT, Leonard BA. Use and effectiveness of tobacco telephone counseling and nicotine therapy in Maine. Am J Prev Med 2005;288 –94. 30. An LC, Schillo BA, Kavanaugh AM, et al. Increased reach and effectiveness of a statewide tobacco quitline after the addition of access to free nicotine replacement therapy. Tob Control 2006;15:286 –93. 31. Patrick DL, Cheadle A, Thompson DC, Diehr P, Koesell T, Klinne S. The validity of self-reported smoking: a review and meta-analysis. Am J Public Health 1994;84:1086 –93. 32. Glasgow RE, Mullooly JP, Vogt TM, et al. Biochemical validation of smoking status: pros, cons, and data from four low-intensity intervention trials. Addict Behav 1993;18:511–27. 33. SRNT Subcommittee on Biochemical verification. Biochemical verification of tobacco use and cessation. Nicotine Tob Res 2002;4:149 –59. 34. Tomson T, Bjonstrom C, Gilljam H, Helgason AR. Are non-responders in a quitline evaluation more likely to be smokers? BMC Public Health 2005;5:52. 35. Macleod ZR, Charles MA, Arnaldi VC, Adams IM. Telephone counselling as an adjunct to nicotine patches in smoking cessation: a randomised controlled trial. Med J Aust 2003;179:349 –52. 36. Zhu S, Tedeschi GJ, Anderson CM, et al. Telephone counseling as adjuvant treatments for nicotine replacement therapy in a “real world” setting. Am J Prev Med 2000;31:357– 63. 37. Centers for Disease Control and Prevention. Best Practices for Comprehensive Tobacco Control Programs—2007. Atlanta: USDHHS, CDC, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, 2007.
American Journal of Preventive Medicine, Volume 35, Number 2
www.ajpm-online.net