A pragmatic, randomized, controlled study evaluating the impact of access to smoking cessation pharmacotherapy coverage on the proportion of successful quitters in a Canadian population of smokers motivated to quit (ACCESSATION)

Study design and participants

This open-label RCT was conducted in 58 different sites across Canada (excepting Quebec, Alberta, and New Brunswick). Smokers were recruited by study site referrals and advertisement. Potential subjects were screened by a call centre to assess initial eligibility and drug insurance coverage. The investigator then confirmed eligibility (see Additional file 1).

Participants were adults (18–75 years) who smoked ≥10 cigarettes per day, were willing to set a quit date within 14 days following screening/randomization, had no period of abstinence >3 months in the past year, and had not attempted to quit smoking in the 30-day period before screening. Exclusion criteria were: unknown insurance status, access to unused pharmacotherapy or prescription, use of tobacco products other than cigarettes within the past month, and life-threatening illness (e.g., known or suspected cancer or other disease with a life expectancy of <1 year). Medical and psychiatric conditions were not exclusion criteria per se.

The study was conducted in compliance with the ethical principles originating in or derived from the Declaration of Helsinki and in compliance with all International Conference on Harmonization Good Clinical Practice Guidelines. Written informed consent for participation in the study was obtained from all participants. The final protocol, any amendments, and informed consent documentation were reviewed and approved by the Canadian Institutional Review Board (IRB) (IRB Services, Aurora, ON, Canada), or by local Canadian Ethics Committees (University of British Columbia Clinical Research Ethics Board, Vancouver, BC; Vancouver Island Health Authority Clinical Research Ethics Board, Victoria, BC; University of Manitoba Research Ethics, Winnipeg, MB; Winnipeg Clinic Ethics Committee, Winnipeg, MB; Research Ethics Board St. Joseph's Healthcare Hamilton, Hamilton, ON; University of Western Ontario Office of Research Ethics, London, ON).

Randomization and interventions

Eligible participants were first counseled about smoking cessation and then randomly and blindly assigned (1:1; centrally to prevent gaming) at the baseline visit to receive either full coverage for 26 weeks or no coverage. Randomization was performed by the investigators using blinded lots of computer-generated randomization codes from the study biostatistician. Randomization codes were mapped to SmartPayment™ cards (drug reimbursement cards), with a distinctive color linked to the study arm to which the subject was randomized. The SmartPayment™ cards were enclosed in sealed envelopes with the randomization codes printed on the envelopes. After unblinding, the investigator and participant agreed on the pharmacological and/or non-pharmacological smoking cessation method(s) to be used. Eligible pharmacotherapies were: varenicline; bupropion; and NRT patch and gum, prescribed according to the most recent version of the respective product monograph or equivalent (see Additional file 2). A prescription had to be issued by a physician for these pharmacotherapies to be reimbursed; NRTs obtained without a prescription were not eligible for coverage, but were considered medication for statistical analysis. Participants (irrespective of allocated group) were not obliged to use a pharmacotherapy in their quit attempt. A target quit date was set within 14 days of randomization. The use of a SmartPayment™ card permitted access to either full coverage of prescribed pharmacotherapy for 26 weeks (full coverage) or a reimbursement of $5.00 per pharmacy dispensation (no coverage). This facilitated tracking of prescriptions dispensed in both groups and provided an incentive for subjects in the no coverage group to present their card at each pharmacy visit. Participants received compensation for reasonable expenses for transportation or parking for study appointments, if needed.

Assessments and follow-up

Patient assessment, treatment recommendations, and prescriptions, as needed, were performed by physicians. Other study-related tasks were performed by trained personnel. Participants were screened at the baseline visit for demographic information, type of drug plan insurance, smoking history, Fagerström Test for Nicotine Dependence (FTND) score [13, 14], comorbid medical/psychiatric history including the Columbia Suicide Severity Rating Scale (C–SSRS [15]), and concomitant medications. Brief (10-minute) smoking cessation counseling was provided to all subjects at the screening visit and at each contact, per each site’s practice. Prior to study start, all participating sites were provided brief training in current clinical recommendations/guidelines [16]. Participants made four clinic visits (weeks 1, 26, 39, and 52), and received six telephone calls (weeks 2, 13, 30, 34, 44, and 48), during which smoking status, quit attempt(s)/methods, FTND score, and abstinence status (via self-report at clinic visits) were collected. Confirmatory urinary cotinine measurements were collected, but investigators and patients were blinded to the results to mimic real-world practice (where this is not widely used). Patients were not aware that physicians were blinded to the results.

Safety evaluations included adverse event (AE) monitoring, physical examinations, and C–SSRS [15]. All observed or volunteered AEs, regardless of smoking cessation method or suspected causal relationship, were to be reported. The investigator was required to pursue information to both determine the outcome of the AE and assess whether it met criteria for a serious AE. Events were recorded up to and including 28 days after the last administration of one of the smoking cessation methods or week 26 (whichever was longer).

Effectiveness evaluations

The primary endpoint was self-reported 7-day point prevalence of abstinence (PPA) at week 26 (end of coverage eligibility), defined as abstinence from smoking (not a single puff) for at least the preceding 7 days [17]. This pragmatic endpoint was chosen because in real-world settings there is no evidence of significant difference between estimates of smoking prevalence based on self-report versus urinary cotinine measure [18]. Whereas the primary endpoint was more pragmatic, the key secondary endpoints (self-reported 7-day PPA at week 26 and continuous abstinence rates [CAR] at weeks 26–52, confirmed by urine cotinine) were more explanatory. Other secondary endpoints were self-reported 7-day PPA at weeks 2 and 13, self-reported CAR (weeks 26–39), total number of quit attempts from randomization through week 26, and use of smoking cessation pharmacotherapies during quit attempts.

Statistical analysis

Based on previous clinical studies of varenicline [19, 20], market research, and relevant literature [21–24], quit rates in the no coverage versus full coverage group were assumed to be 15% versus 21%, respectively. A sample size of at least 676 participants per group was estimated to provide 80% power to detect a conservative difference of 6% between groups. Some consider a 2% difference clinically significant because of the estimated 50% risk reduction in smoking-related mortality associated with cessation [25]. Outcome measures were assessed using an intent-to-treat (ITT) analysis that included all randomized participants. Individuals who discontinued the study or were lost to follow-up were counted as smokers from the time of discontinuation. For calculation of CAR, participants who reported no smoking or no use of nicotine since the last contact and had a negative urine cotinine test were counted as non-smokers. Participants with missing urine cotinine test results were considered smokers.

Endpoints were analyzed using logistic regression models that included randomized group effect (full coverage versus non-coverage) as the independent variable and investigation centre as a covariate. In addition, the interaction between randomized group and centre was tested by expanding the logistic regression model to add the interaction term. In testing for the primary and key secondary endpoints, the Type-I overall error rate of 0.05 was preserved by using statistical methods that adjusted for multiple tests.

Safety data were summarized for all randomized subjects. The incidence of AEs reported during the study was summarized by randomized arm (coverage; non-coverage) and pharmacotherapy (users; non-users) for descriptive rather than comparative purposes.

Baseline characteristics

Eligible participants were recruited from March 3, 2009 to September 20, 2010. Participants attended clinic visits at the time of randomization and at regular intervals for 52 weeks. Using postal codes as a proxy, participating sites were identified as being both in urban (54) and rural (4) locations. Participants were recruited by eight specialists (n = 115; 8.3%) and 50 generalists (n = 1265; 91.7%) and were referred from clinical practice (n = 1289; 93.4%) or advertisement (n = 91; 6.6%). Participant disposition is summarized in Figure 1. Baseline demographics, smoking history, and level of nicotine dependence were well balanced (Table 1). Interactions for randomization group by centre were not statistically significant (p > 0.4), supporting the consistency of using combined results from different centres. The primary analysis was ITT and involved all patients who were randomly assigned (696 subjects in the full coverage group and 684 subjects in the no coverage group). Participants were analysed by original group assignment.

Effectiveness of coverage of pharmacotherapy

During the 26-week coverage eligibility period, a higher proportion of participants in the full coverage versus no coverage group made at least one quit attempt (86.9% versus 70.0%, respectively; p < 0.001), but the number of quit attempts were not statistically different across groups (mean 3.1 versus 2.6, respectively; p = 0.103), and more subjects were dispensed at least one pharmacotherapy (98.0% versus 63.6%, respectively; p < 0.001). The median number of daily doses dispensed per participant was higher in the full versus no coverage group, except for bupropion (70 versus 42, respectively; p < 0.001, for varenicline; 56 versus 18.5; p < 0.001, respectively, for NRTs; and 39 versus 34; p = 0.454, respectively, for bupropion).

In the full coverage group, varenicline was dispensed at least once for 558 participants (80.2%), nicotine patches/gum for 201 participants (28.9%), and bupropion for 60 (8.6%) participants. The corresponding figures for the no coverage group were 297 (43.4%), 126 (18.4%), and 58 (8.5%). In the full coverage group, 18.4% of participants used at least two distinct pharmacotherapies, whereas only 6.3% did so in the no coverage group. Non-pharmacological methods were used by 5.7% and 9.9%, respectively.

Adverse events