Article Text

Nicotine replacement therapy ‘gift cards’ for hospital inpatients who smoke: a prospective before-and-after controlled pilot evaluation
1. Kerri A Mullen1,
2. Kathryn L Walker1,
3. Shireen Noble1,
4. Gillian Pritchard1,
6. Natalie Martin1,
7. Andrew L Pipe1,2,
8. Robert D Reid1,2
1. 1Division of Cardiac Prevention & Rehabilitation, University of Ottawa Heart Institute, Ottawa, Ontario, Canada
2. 2Faculty of Medicine, University of Ottawa, Ottawa, Ontario, Canada
1. Correspondence to Dr Kerri A Mullen, Division of Prevention and Rehabilitation, University of Ottawa Heart Institute, Ottawa, ON K1Y 4W7, Canada; kmullen{at}ottawaheart.ca

## Abstract

Introduction A common barrier identified by individuals trying to quit smoking is the cost of cessation pharmacotherapies. The purpose of this evaluation was to: (1) Assess the feasibility of offering nicotine replacement therapy (NRT) ‘gift cards’ to hospitalised smokers for use posthospitalisation; and, (2) Estimate the effect of providing NRT gift cards on 6-month smoking abstinence.

### Randomisation and recruitment

Patients were eligible to be contacted for this evaluation if they had received an OMSC intervention while in hospital during either the control or QCI time periods. Random subsamples of eligible control and QCI participants were selected to be contacted 6 months following their hospitalisation. Randomisation was stratified by site to account for differences in size and population. Block randomisation was used for the two larger hospitals; participants were sorted into 2-week blocks based on their hospital discharge date. Using an online random number generator, half of the participants in each block were selected to be contacted. For the smaller hospital, the participants were sorted by discharge date, and sequentially assigned to either A or B. Using a coin toss, the participants assigned to group B were selected to be contacted. Those selected to be contacted were called no more than five times, leaving up to three voicemails. A total of 548 participants were randomly selected; 274 from each group.

### Variables, outcomes and data sources

Participants were contacted by telephone to complete a survey that collected sociodemographic (age, sex, education, income), smoking-related (cigarettes smoked per day, other smokers in the home) and health-related (history of depression and/or anxiety, alcohol use, cannabis use) variables.

Feasibility outcomes included: number of Quit Cards distributed to participating hospitals; number of Quit Cards distributed to patients (calculated as number of cards sent to participating hospital minus the number of unused cards returned at the end of the programme); number of Quit Cards redeemed by patients (reported on the STI Technologies platform); dollar amount used to purchase NRT (reported on the STI Technologies platform); type of NRT purchased (brand, type and amount reported on the STI Technologies platform); and, type and amount of NRT used (participant self-report). Using Likert Scales of 1 (not very) to 5 (extremely), all participants were asked to rate their satisfaction with the support they received, and QCI participants were asked to rate the extent to which they found the Quit Cards to be motivating, easy to use and helpful. Given QCI participants had different amounts of time to redeem their Quit Cards depending on when they received the card and the expiration date, we tracked the median number and range of days participants had to redeem their Quit Cards and assessed whether redemption rates, amount of NRT purchased and quit rates differed depending on how much time they had to use their card.

To estimate the effect of QCI on quitting, self-reported 7-day point prevalence smoking abstinence (‘Have you smoked any form of tobacco in the past 7 days?’) was gathered at 6 months. A random subsample of participants (representing approximately 10% of participants from each group) was asked to complete an expired carbon monoxide (CO) test. A CO reading of ≤4 ppm was considered confirmation of smoking abstinence. Abstinence rates were adjusted based on the observed misreporting rates.

To determine the perceived benefits and challenges of the intervention from hospital staff, a postprogramme survey (Appendix A: online supplemental material 1; pp. 2–13 (figures B–F, tables A,B)) was sent to staff involved in coordinating and/or delivering the QCI at their site. Each staff member surveyed had experienced offering the OMSC intervention to patients with and without Quit Cards.

### Statistical methods

Feasibility outcomes and staff survey data were summarised using descriptive statistics. Binary logistic regression was used to assess smoking abstinence by group (control vs QCI) and in subgroup analyses. Our primary analysis of cessation compared the outcomes of survey responders only. A sensitivity analysis of cessation outcomes using intention-to-treat principles (the Russell standard)16 was also completed, whereby patients who had died, had a wrong number, had a language barrier, did not smoke cigarettes, had moved to long-term care or hospice, or for other reasons were deemed ineligible (eg, denied receiving the intervention) were removed from the analysis, and patients who did not answer or refused to complete the survey were assumed to be smoking. Several subgroups were identified by the evaluation team a priori to determine which variables would be important to consider in future trials. These included: sex, education, income, community size, cigarettes per day, living with other smokers, history of depression, history of anxiety, alcohol-use and cannabis-use. Little’s missing completely at random (MCAR) test was used. If data were found to be missing at random, multiple imputation would have been completed using the regression method. If missing data were found to be MCAR or missing not at random, only observed data would have been used.17 All assumptions of binary logistic regression were confirmed (linear relationship between the logit of the outcome variable and each selected predictor variable; no extreme values or outliers in continuous predictor variables; no high intercorrelations among the predictors). Analyses were carried out using IBM SPSS Statistics V.26.

### Bias

While this evaluation involved a real-world quality-improvement pilot programme using a before and after cohort design, efforts were made to reduce potential biases. To limit selection bias, random subsamples of control and QCI participants were selected to be contacted for this evaluation. To increase generalisability of the results, the evaluation took place at three locations—two major urban institutions and one small town hospital. To limit observer bias, outcome assessors were at arm’s length to the OMSC programme and were not informed of the evaluation’s objectives and hypotheses. To limit history bias and to avoid potential contamination/cross-over, the two cohorts were selected within a few months of each other and with the controls being recruited after expiration of the Quit Card programme. To limit response bias, smoking abstinence outcomes were adjusted using the CO test results.

### Power

Power was calculated a priori to determine the number of participants to randomly select. We randomly selected 548 participants (274 control, 274 QCI), assuming a response rate of 68% (372 completed surveys) would be achieved, based on the 6-month response rate observed in a previous programme study.15 Using logistic regression, with 186 in each group, we would have 80% power (two-sided test; α=0.05) to detect an OR of 1.84 for QCI versus control, assuming the QCI odds were similar to studies of NRT versus control.18

## Results

### Participants

We reached 73.1% of our 372 target, with 272 completed surveys (148 control; 124 QCI). Figure 1 displays the participant flow and reasons for exclusion. Results of Little’s MCAR test identified that missing data were MCAR (χ2=1.236, p=0.266), therefore, only observed data were used. Groups were similar in terms of baseline characteristics (table 1).

Figure 1

Participant flow.

Table 1

Participant characteristics

### Staff surveys

A total of 19 staff members involved in coordinating or delivering the smoking cessation interventions at their site completed the postprogramme staff survey. A full summary of the results can be found in online supplemental appendix A (pp. 2–13). The majority (58%) ranked their satisfaction with the programme as 10/10, with 84% ranking their satisfaction between 8 and 10. The majority (94%) felt that having Quit Cards to offer made it easier to intervene with patients who smoke. Seventy-eight per cent felt the Quit Card increased the motivation of staff to intervene with patients who smoke, and 100% felt that Quit Cards increased the motivation of patients to quit smoking. Nearly 80% (15/19) of those surveyed responded to the open-text question regarding the benefits of the programme. The most common themes surrounding the benefits of the Quit Card programme were: (1) Removing the barrier of cost of cessation pharmacotherapy for patients, particularly for lower income patients; (2) The card as an effective cessation tool for staff that facilitated their smoking cessation conversation with patients, (3) The card increasing motivation of patients to quit, (4) The card being convenient, flexible and providing continuity of care, and (5) Feelings of gratitude from patients towards staff. All staff surveyed responded to the open-text question regarding the challenges of the programme. Six (31.6%) responded that there were no challenges. The most common themes surrounding challenges were: (1) Pharmacies not initially knowing how to process the cards, charging dispensing fees and not always having enough NRT stock; and, (2) Expiry dates presenting a challenge and pressure for patients to redeem their cards quickly.

## Discussion

Our pilot evaluation found NRT gift cards (‘Quit Cards’) to be a feasible, simple to deliver and promising intervention for hospitalised patients who smoke. Uptake, delivery and satisfaction of the QCI was high among participating hospitals, and redemption and satisfaction rates were high among patients. Point prevalence abstinence at 6-month follow-up was higher for those who received the QCI, compared with controls. Not only was the increase statistically significant, but also higher than the minimal clinically important difference of 5% commonly used in smoking cessation trials. Hospitalisation has been shown to increase one’s motivation to quit smoking.19 All staff surveyed as part of this evaluation perceived that the Quit Card enhanced the motivation of patients to quit smoking.

Not using cannabis was an independent predictor of smoking abstinence in this evaluation. Recreational use of cannabis was legalised in Canada in October 2018. In 2017, prior to legalisation and within the same timeframe of this evaluation, 7.4% of Canadians ≥age 40 years reported cannabis use in the past year.20 This is much lower than the 18.4% prevalence observed among our sample of hospitalised patients who smoke tobacco. Participants who had not used cannabis in the past 6 months had higher odds of quitting smoking. A recent meta-analysis and narrative review of 20 studies (12 randomised controlled trials (RCTs) and 8 uncontrolled trials) found single and multisubstance interventions that addressed tobacco and/or cannabis showed weak evidence for an effect on either tobacco or cannabis cessation among co-users.21 Though dual-substance interventions targeting tobacco and cannabis appear to be feasible and acceptable, cannabis use will be an important variable to collect and assess in future tobacco-cessation trials, and more high-quality evidence is needed to determine what interventions may be most effective at helping co-users. Higher income (>$C75 000 per year) was associated with quitting among participants in both groups, although abstinence rates were an absolute 13.9% higher among lower-income Quit Card participants, compared with lower-income control participants. As this intervention was designed, in part, to remove the barrier of cost of cessation therapies, participant income may be an important consideration in the design of future trials. Our study adds to the growing evidence base examining the utility of financial incentives for smoking cessation in clinical and non-clinical populations. A systematic review and meta-analysis found that covering the cost of NRT increases the odds of: making a quit attempt (OR 1.11, 95% CI 1.04 to 1.17; four trials); using NRT (OR 1.79, 95% CI 1.54 to 2.09); and smoking abstinence (OR 1.77, 95% CI 1.37 to 2.28; six trials).13 Partial coverage also leads to greater quitting, compared with no coverage (OR 1.27, 95% CI 1.02 to 1.59; five trials). A systematic review examining the use of incentives (eg, cash payments, gift cards) aimed at prompting or reinforcing smoking cessation in non-clinical populations found that smokers who received incentives were more likely than controls to be abstinent at ≥6 months (OR 1.42, 95% CI 1.19 to 1.69; 17 trials).12 A recent RCT evaluating the effect of paying low-income hospitalised smokers for participating in smoking cessation counselling, using cessation pharmacotherapies, and for being smoke-free found a non-statistically significant, although minimally clinically important, difference of 10.7% in 6-month cessation rates favouring the incentive group, compared with controls (OR 2.56; 95% CI 0.84 to 7.83, p=0.10).22 A 2×2 factorial RCT found a 4.5% difference in intention-to-treat smoking abstinence at 6 months among patients discharged from hospital who received NRT patches compared with those who did not receive NRT patches, although the difference was not statistically significant (22.8% vs 18.3%, respectively; p=0.051).23 A three-site RCT found no difference in 6-month biochemically confirmed smoking abstinence between patients who received posthospitalisation automated telephone calls plus a 3-month supply of their cessation medication of choice (single form or combination of nicotine patch, nicotine gum, nicotine lozenge, bupropion or varenicline) and patients who received medication and counselling recommendations only (16.6% vs 15.5%, respectively; RR 1.07 (0.84–1.37)).24 As a patient incentive and promising medication-distribution mechanism, the Quit Card affords numerous advantages: the intervention is easy to deliver and takes minimal time to administer to participants; it requires little storage space; card redemption is tracked in real time, which facilitates distribution and financial management; the programme is only charged for the product purchased, potentially reducing waste and unused or expired product; and, participants can purchase the product at their own pharmacy, affording another opportunity for intervention with a health provider. This was a quasi-experimental evaluation of a pilot programme implemented under real-world conditions; however, several efforts were taken to minimise potential biases. An RCT is warranted to study the impact of the QCI using a more rigorous design. While we did randomly select participants from both the control and QCI groups, we did not reach our completed survey target. We estimated that we would reach 68% of participants randomly selected to complete the evaluation; however, this estimate was taken from clinical trial data and did not reflect the response rate observed in this real-world evaluation where patients were not necessarily expecting the call. Nonetheless, we intend to use this randomisation approach in future programme evaluations, where appropriate, and will apply oversampling to increase the likelihood of reaching the recruitment target. While both major urban and small-town hospital sites were used to conduct the evaluation, this multisite evaluation took place in only one region of Ontario, Canada. At the time of the evaluation, the Ontario provincial formulary provided coverage for varenicline and bupropion, but not NRT. This type of incentive programme might yield different results in jurisdictions that have more comprehensive smoking-cessation pharmacotherapy coverage. That said, given the simplicity of Quit Card distribution and management and the potential utility of the card as an incentive or motivational tool, future studies comparing such tools to other medication coverage and distribution mechanisms should be considered. The OMSC programme has previously implemented Quit Card programmes that covered all smoking-cessation pharmacotherapies (ie, NRT, varenicline and bupropion); they may be of further benefit in terms of individualising the intervention for patients. Nearly half of the participants in this evaluation were hospitalised with a cardiac condition. This may have contributed to the relatively high cessation rates observed in both groups. This evaluation only tested results among an inpatient population. The QCI may be useful in other fast-paced, acute care settings given the relative ease with which it can be delivered. ## Conclusion The NRT ‘gift card’ appears to be a feasible and effective smoking cessation tool that removes one of the main barriers to the use of evidence-based smoking-cessation pharmacotherapies, while potentially motivating both patients and health providers. Future studies should consider randomised controlled designs to further evaluate the effect of Quit Cards in a variety of populations and settings. ### What this paper adds • The cost of cessation medications is a common barrier identified by individuals wanting to stop smoking. Covering the cost of quit smoking pharmacotherapies, including nicotine replacement therapy (NRT) products, and using incentives to assist people in quitting smoking have been found to increase quit attempts and cessation rates. • The feasibility and effectiveness of distributing gift cards, valid only for the purchase of NRT, have not been evaluated. • This study found that distributing NRT gift cards worth$C300 to hospitalised patients who smoke was highly feasible and led to increased smoking abstinence at 6 months.

• The NRT gift card intervention led to high satisfaction and was perceived to enhance motivation among patients and the hospital staff who were delivering the intervention.

## Data availability statement

Data are available upon reasonable request.

## Ethics statements

### Ethics approval

This project was reviewed by the Ottawa Health Science Network Research Ethics Board (OHSN-REB) and approved as a quality improvement program evaluation. After initial review of our pilot study proposal by the OHSN-REB, it was deemed that this project fell within the context of quality initiative, quality improvement, quality assurance, and/or program evaluation. Consequently, they determined that the study was not ‘human subject research’; therefore, full review by the OHSN-REB was not required.

## Acknowledgments

The authors thank the organisations that participated in the Quit Card pilot programme, including the sites and health providers that distributed the cards, the pharmacies that processed the cards for redemption, and the team at STI Technologies Inc. The authors also thank the following OMSC staff, involved in programme delivery: Manu Sandhu, Sarah Mackler, Pamela Heise, Priscilla Belanger, Jo-Anne Gagnier, Laura Hobler, and Marie-Lyne Do Couto. We thank the evaluation assistants, Emilie Langevin, Brenna Czajkowski, Ines Lopez-Espinosa, Shainuka Kannathas, Tosi Nnaji and Jonathon Cecilon. The authors also thank the smoking cessation nurses and counsellors, Marta Klepaczek, Anne Marie LaRue, Donna Pittman, Donna Taylor, Marie France Walter, Cindy Knight-Vigneron, Lauren Johannesen, Tulsi Jose, and Rebecca Nadler.

• ## Supplementary Data

This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.