Pilot Trial of QuitBet: A Digital Social Game That Pays You to Stop Smoking

Abstract

Contingency management is an effective treatment for cigarette smoking cessation but feasibility and acceptability concerns have been barriers to implementation. We conducted a pilot test of QuitBet, a commercial, digital (smartphone) social game for smoking cessation during which participants earned financial incentives for abstinence. QuitBet included a social feed for posting messages and entirely participant-funded incentives in the form of a deposit contract (the “bet”). QuitBet had a bet of $30 and lasted for 28 days. After a week to prepare for quitting, the quit day was Day 8. Between Day 9–28 (a 20-day period), participants earned back $1 of their $30 bet for each day of carbon monoxide (CO)-verified abstinence (≤6 ppm). Remaining bet money was pooled into a “grand prize” pot. Participants who were abstinent on at least 19 of the 20 days (1 “lapse” day allowed) were declared “winners” and split the grand prize pot equally. A game host posted a daily message containing evidence-based education about smoking cessation or a discussion topic. Recruitment goals were met. Among the players (N = 50 U.S. adults, 78% female), 17 (34%) were winners. Thirty-seven participants (74%) responded to a post-QuitBet survey, of whom 95% said they would be interested in playing another QuitBet and would recommend QuitBet to others. Overall, feedback was positive with some suggestions for improvement. In conclusion, a digital social game for smoking cessation with a deposit contract was feasible and acceptable. Next steps include conducting a randomized trial to establish efficacy and a sustainable business model.

Contingency management (CM) is an evidence-based intervention rooted in operant conditioning theory that involves using financial incentives to motivate and reinforce behavior change. CM for cigarette smoking cessation, which typically involves delivering financial incentives contingent upon biochemically verified abstinence (e.g., exhaled carbon monoxide breath testing), is among the most efficacious behavioral interventions for smoking cessation (Hartmann-Boyce et al., 2021; Notley et al., 2019).

However, traditional clinic-based CM procedures have been difficult to implement in “real world” community settings. There are concerns about relapse after incentives are discontinued, although relapse rates are similar to other treatments (Gupta, 2015; Hartzler et al., 2012; Petry, 2010). Other concerns pertain to feasibility and acceptability, including how to obtain funding for the incentives and the need for frequent clinic visits for testing of exhaled breath carbon monoxide (CO) to biochemically verify abstinence (Notley et al., 2019; Petry, 2010; Robertson et al., 2018; van den Brand et al., 2019). CO is elevated in individuals who use combustible tobacco products. Given that the recommended CO cut-off for abstinence (5–6 parts per million, ppm) is typically reached within 24 hr after the last cigarette (Benowitz et al., 2020), participants are typically tested at least daily. Other methods of biochemical verification of abstinence (e.g., urine or saliva cotinine testing) only need to be done weekly (Benowitz et al., 2020). However, these methods detect all forms of nicotine use and are not specific to smoking combustible products. Therefore, they cannot be used to verify abstinence from cigarettes if the participant is using nicotine replacement medications or other non-combustible products containing nicotine (e.g., e-cigarettes).

Digital Contingency Management

Delivering CM for smoking cessation via the Internet and/or smartphones (i.e., digital CM) is one promising strategy for increasing feasibility and acceptability. Dallery and colleagues pioneered the development of digital CM (Dallery & Raiff, 2011) that did not require any clinic visits. Participants were sent a professional-grade CO testing device (value $500+) and other equipment (e.g., web camera; Dallery et al., 2007, 2013) by mail to use for daily, remote (i.e., home-based) CO testing. Nevertheless, the cost of the CO testing device and the incentives and the logistics of shipping equipment back and forth limited real-world scalability. Bedfont recently introduced the iCO^™ Personal Smokerlyzer^®, the first portable, low-cost ($60) CO testing device (coVita, 2020). According to the iCO manual, the iCO has a range of 0–100 parts per million (ppm) with a sensitivity of 1 ppm and a suggested abstinence cut-off of 6 ppm. The iCO appears to have adequate reliability and validity relative to Bedfont’s professional-grade CO testing devices (Tuck et al., 2020; Wong et al., 2019). For example, Tuck et al. (2020) collected CO samples from the same group of participants at each of ten visits over a 3-week period using three different models of Bedfont CO monitors (30 samples per participant; ten each with the iCO and two different professional grade monitors). They reported no significant differences in diagnostic accuracy (i.e., prediction of smoking status based on urine cotinine testing) but the optimal cut-offs for determining smoking abstinence differed by monitor type. They recommended a cut-off of 6 ppm for the iCO, matching Bedfont’s recommendation. Personal CO testing devices like the iCO could facilitate community implementation of CM. Indeed, several CM studies have now been published in which participants used the iCO; abstinence cut-offs were 5–6 ppm (Dallery et al., 2021; Kendzor et al., 2020; Kurti et al., 2020).

Deposit Contracts

A promising strategy for reducing the costs of CM incentives is for participants to fund the incentives partly or entirely themselves via a deposit contract, in which each participant pays a “deposit” that can only be recouped by quitting smoking (Dallery et al., 2017; Gine et al., 2010; Paxton, 1980, 1981; Spring et al., 1978; White et al., 2013). Deposit contracts are rooted in the loss aversion principle, which posits that people are more motivated to avoid losses than to obtain equivalent gains (Kahneman & Tversky, 1979). To our knowledge, only two studies have directly compared the efficacy of CM with versus without a deposit contract. In one small study (Dallery et al., 2008), eight individuals participated in a brief (2-week) digital CM intervention for smoking cessation and were randomized to a $50 deposit contract or no deposit. In both groups, participants were instructed to submit CO tests twice per day and earned incentives on an escalating schedule for CO values that were below the specified cut-off. In the deposit group, participants could earn back their entire $50 deposit plus up to an additional $28.80 whereas in the no deposit group participants could earn a total of $78.80. Results indicated that there were no differences in abstinence rates between the two groups. In another study (Halpern et al., 2015), over 2,500 participants were randomized to one of four CM smoking cessation programs in which they could earn up to $800 in rewards for cotinine-confirmed abstinence or usual care (no CM); two of the four CM programs included a $150 deposit (net rewards of $650 in these programs). Abstinence was confirmed via cotinine testing at 14 days, 30 days, and 6 months after a target quit date. Whereas 90% of those who were randomized to a reward-only CM program chose to enroll in it, only 14% of those randomized to a deposit CM program chose to make the deposit and enroll. Among the sample who enrolled in their programs, abstinence rates were significantly higher in the deposit programs (68.2% vs. 23.3% at 30 days, 52.3% vs. 17.1% at 6 months, p’_s < .001). Even after adjusting for lower willingness to enroll in the deposit programs, the deposit programs had a higher abstinence rate than the reward-only programs. Taken together, findings from these two studies suggest that deposit contract approaches are promising, but more research is needed on how to balance effectiveness versus willingness to enroll, and that a $150 deposit may be unacceptably high.

Serious Games for Behavior Change

“Serious games” apply game elements (competition, prizes, social interaction) to health (Cugelman, 2013) to make behavior change engaging and fun. Interest is increasing rapidly in smartphone-based games for smoking cessation (e.g., Amante et al., 2020; Blok et al., 2019; Derksen et al., 2020; Edwards et al., 2018; Guo et al., 2020; Krebs et al., 2019; Marin-Gomez et al., 2019; Raiff et al., 2018; Rajani et al., 2019; Schlam & Baker, 2020; Scholten et al., 2019). WayBetter, Inc. offers smartphone games for behavior change directly to the public that include a deposit contract, which is called the “bet.” Their first game, DietBet for weight loss, has had over 1,000,000 paying “players,” and produces clinically meaningful outcomes (Leahey & Rosen, 2014). All WayBetter games are played in a group format and, in addition to the deposit contract, include a “social feed” on which players can post and respond to messages, photos, and videos. A game host leads and moderates the social feed, providing support, encouragement, education, and discussion topics. Games are offered in a range of durations and bet amounts. For example, the DietBet Kickstarter game lasts 4 weeks with a goal to lose 4% of one’s body weight; a typical bet amount in a Kickstarter is $30. All players who achieve the game’s goal are declared “winners” and equally split the pot of bets, minus an administrative fee (typically 15%–25% of the pot) which the company keeps to cover its costs, including paying the game hosts as well as transaction processing fees relating to credit card payments in and Paypal payments out (winners are typically paid out via Paypal). Thus, winners are refunded their entire bet plus receive extra money from the forfeited bets of players who did not meet the goal. WayBetter has established systems for objective verification of game outcomes involving submission of photos, videos, and/or data from wearable devices.

Present Study

In the present study, we conducted a pilot trial of QuitBet, a digital social game for smoking cessation modeled upon DietBet. QuitBet is innovative in that it is the first digital smoking cessation program to employ a self-sustaining, entirely participant-funded deposit contract model of CM and to combine this type of deposit contract with a game host and social feed intended to increase engagement and motivation for change, and to provide evidence-based education about smoking cessation. The goal of this pilot trial was to evaluate the feasibility and acceptability of QuitBet, with a focus on whether QuitBet had potential to be a successful commercial product that would balance fun (i.e., appealing, low participant burden) with effectiveness (i.e., for reducing and/or quitting smoking). The design of QuitBet was informed by previous CM studies from the academic literature as well as the business model that WayBetter has used successfully for its other games. Our pilot trial had several a priori hypotheses. First, we hypothesized that we could recruit 50 players within 1 month, which we believed would be an indication of feasibility and market interest. Second, we hypothesized that QuitBet players would be engaged and report high satisfaction, as evidenced by: (a) ≥75% of players submitting a CO test on ≥75% of game days, (b) ≥50% of players reporting that they would be interested in playing a second QuitBet, and (c) ≥75% of players reporting that they would recommend QuitBet to others. Our secondary outcome was preliminary effectiveness. We hypothesized that (a) ≥30% of players would be “winners,” (b) ≥50% of players would make a serious quit attempt (at least 24 hr of abstinence), and (c) ≥50% of players would reduce their average cigarettes smoked per day by at least 50%. We selected these benchmarks based on outcomes of other WayBetter games which have been commercially successful and results of past CM studies for smoking cessation.

Method

Participants

Participants (N = 50) were included if they were adult U.S. residents ≥21 years of age who smoked ≥10 cigarettes daily, had an Apple iPhone, and were willing to deposit $30. Individuals who self-reported that they were pregnant or planning to become pregnant within the next year were excluded. This study was approved by Advarra Institutional Review Board, Pro00035152 (QuitBet Phase I: A Digital Social Game that Pays You to Stop Smoking).

Procedure

Recruitment and Enrollment

Participants were recruited via messages sent to WayBetter’s customer databases (i.e., individuals who provided permission to receive e-mails from WayBetter), posts in WayBetter games (e.g., DietBet) and on WayBetter’s social media accounts, and paid social media advertising. All advertisements included a link to the QuitBet website, which described the study and game rules and displayed a link to an anonymous screening survey. Upon completing the screening survey, eligible applicants proceeded to an electronic consent form. They were asked to read the consent form, contact the study team if they had questions, and type their name and contact information into a box to indicate consent.

Measures

Baseline Survey.

Consenting participants were sent a link to a baseline survey (measures described below). Baseline survey questions included demographics (age, gender, race, ethnicity, marital status, educational attainment, employment status, and typical work shift (day, afternoon/evening, or overnight); WayBetter history (whether they had previously played DietBet, StepBet, and/or RunBet); smoking history [years smoked, cigarettes per day, other tobacco product use history including whether they had ever tried e-cigarettes or were currently using e-cigarettes, and nicotine dependence severity Fagerström Test for Cigarette Dependence (FTCD; Heatherton et al., 1991)], and quit history including number of past quit attempts and whether they had ever used behavioral and pharmacological treatments (checklist). Participants also reported the maximum amount that they would be willing to bet to play QuitBet and how much they were willing to pay for the iCO if they had to pay for it themselves. Uponcompleting the baselinesurvey, for which they received a $10 non-cash prize sent by mail(choice of QuitBet-branded popsocket, notebook, or charging bank), they were sent instructions for downloading the QuitBet app and depositing $30 via a credit card or PayPal. After paying their bet, they were sent an iCO via mail that arrived prior to their game start date.

Post-QuitBet Survey.

On the day after their game ended, participants were sent a link to a post-QuitBet survey and encouraged to complete it as soon as possible. We sent multiple reminders and continued accepting survey responses for several weeks. Participants received a $10 non-cash prize via mail or e-mail for completing this survey popsocket, notebook, or charging bank).

Participants reported their past week smoking behavior (number of days smoked and cigarettes per day) and other tobacco product use (including e-cigarettes) and whether they had used other smoking cessation treatments during QuitBet. The options for other smoking cessation treatments (“select all that apply”) were cold turkey, cut down on smoking before quitting, nicotine replacement therapy (a separate checkbox for each type: patch, gum, lozenge, inhaler, nasal spray), bupropion, varenicline, e-cigarettes, individual counseling, group counseling/classes, U.S. state quitline, other telephone counseling (not state quitline), hypnosis, acupuncture, self-help book, internet program or smartphone app (other than QuitBet), and other.

Participants also rated aspects of their experience and satisfaction with QuitBet on 5-point scales with 5 being the highest, including overall satisfaction, clarity of the game rules, helpfulness of hosts’ posts, supportiveness of hosts and other players, the contribution of QuitBet to their ability to quit smoking, satisfaction with their winnings (winners only), and concern that other players may have cheated. They also reported whether 28 days was a good length for QuitBet (5-point scale: 1 = strongly disagree to 5 = strongly agree) and how many days QuitBet should last, whether the CO cut-off of 6 ppm should be lower or higher or was just right, whether there were any days during QuitBet on which they smoked but their CO was ≤6 ppm (false negatives), whether a baseline positive CO test of ≥7 ppm should be required prior to playing (yes/no), whether they would be interested playing in another QuitBet game (yes/no), and whether they would recommend QuitBet to others (yes/no). Finally, they answered open-ended questions about technical issues they had with their iCO or other features of the QuitBet app and provided general comments.

Post-QuitBet Phone Interview.

Participants were invited to schedule a post-QuitBet phone interview, which occurred within 1 month after the end of the participant’s game. Participants were interviewed using a structured interview guide. They were asked to provide feedback about the enrollment process and their experience playing QuitBet and suggestions for additional features or improvements. Participants received a $30 non-cash prize for completing the interview (water bottle or movie theater gift card).

QuitBet Game

History of QuitBet

WayBetter has standard processes for designing new games, which include generating the game rules and playing test games using Facebook prior to investing in app development. Before conducting the present study, which was the first test of the QuitBet app, we designed a QuitBet prototype and hosted one QuitBet test game on Facebook in 2017 with 32 players (Bloom et al., 2019).

Social Gaming Features

The QuitBet app featured the results of the participant’s CO tests (visible only to the participant), a leaderboard that updated in real time and displayed the number of CO tests ≤6 ppm submitted by each other participant, and a social feed. A game host (Erika Litvin Bloom or Sandra J. Japuntich, each hosted one game) posted a daily message containing evidence-based information about smoking cessation (e.g., link to smokefree.gov) or a discussion topic (e.g., share strategies for coping with cravings). These posts were planned in advance and described in Table 1. The game hosts also served as moderators, responding to participants’ posts. We estimate that hosting duties took <15 min per day. A game referee (AP) provided technical support and answered questions about study procedures.

Table 1.

QuitBet Game Host Daily Post Topics

Day	Host Post Topic
1	Host welcome, link to iCO website, summary of game rules.
2	Discussion topic: share reasons for quitting.
3	Links to websites with smoking cessation resources (e.g., smokefree.gov)
4	Reminder that medication use is allowed, suggestion to call quitline (1–800-QUIT-NOW).
5	“Funny Friday” - post a meme, gif, or emoji that describes how you are feeling about quit day.
6	Tips to prepare for quit day (e.g., get rid of all cigarettes)
7	Reminder about game rules and CO test procedures.
8	Quit day discussion topic: share how you are coping with cravings, links to websites about managing nicotine withdrawal.
9	Congrats to players who earned $1 today, encouragement to keep trying if still smoking.
10	Link to website with suggestions for dealing with cravings.
11	Link to website with tips for staying smoke-free during holidays.
12	“Funny Friday” - post a meme or gif that describes your experience with quitting so far, reminder that you can still earn $1 per day for CO tests of ≤6ppm even if not eligible to be winner.
13	Discussion topic: share tips for coping with weekend cravings.
14	Reminder about game rules, encouragement to keep trying to quit
15	Link to website about quitting on Mondays, the most popular day to quit.
16	Discussion topic: describe benefits of quitting you are experiencing, link to website with timeline of health benefits of quitting.
17	Discussion topic: money saved so far from not smoking.
18	Discussion topic: how you will reward yourself for quitting.
19	“Funny Friday” - share a song title or lyrics that describe how are you “breaking up” with cigarettes.
20	Discussion topic: share tips for coping with cravings on the weekend or getting back on track after a lapse, reminder about $1/day incentive for CO tests ≥6ppm
21	Link to website with tips about staying smoke-free for the long-term.
22	Reminder about game rules and procedures, link to website about getting back on track after a lapse.
23	Discussion topic: describe a time when you were tempted to smoke but didn’t and how you stayed smoke-free.
24	Discussion topic: other helpful apps or websites for quitting smoking.
25	Reminder that game is ending soon, encouragement to complete post-QuitBet survey and phone interview.
26	“Funny Friday” - post a meme or gif that describes how you’re feeling about quitting or overall experience since QuitBet started.
27	Reminder that game ends tomorrow; discussion topic: tell us how you will stay smoke-free after QuitBet ends or if still smoking, your future plans; link to website about continuing to try to quit.
28	Congrats to winners, reminder that they will receive link to post-QuitBet survey and encouragement to schedule interview.

Deposit Contract and Breath Carbon Monoxide (CO) Testing Procedure

Clinical practice guidelines for smoking cessation recommend selecting a quit date (Fiore et al., 2008). Therefore, most smoking cessation programs include a “pre-quit” phase during which participants receive education that is intended to prepare them for a future specified quit date. QuitBet lasted 28 days (4 weeks), for which the 1st week was a “warm-up” period to prepare for quitting and the official quit day was Day 8 (i.e., the goal of the game was to quit smoking for the last 21 days of the game). A 28-day duration was chosen for several reasons. First, this duration is consistent with other WayBetter games as well as with other digital CM studies in the academic literature, which have typically lasted about 4 weeks (e.g., Alessi et al., 2017; Dallery et al., 2017; Kendzor et al., 2020). Second, this duration is supported by research indicating that the majority of smoking relapse occurs within 1 week after the quit date and that maintaining abstinence for at least 2 weeks is predictive of long-term abstinence (Kenford et al., 1994; Romanowich & Lamb, 2010). Third, we believed that a 28-day duration would support feasibility of recruitment because we could keep the deposit amount lower ($30) than recent past CM studies for smoking cessation which had deposits of $50–$150 (Dallery et al., 2017; Halpern et al., 2015) and consistent with other WayBetter games. Finally, if QuitBet were brought to market, it would be possible to play multiple consecutive games to extend game play as long as needed or desired to achieve long-term abstinence. WayBetter has determined that between 35%–50% of DietBet and StepBet players play at least two consecutive games, defined as playing in two games simultaneously with overlapping date ranges (this is allowed) or initiating a second game within 30 days of the end of the first game. We envisioned ultimately offering two versions of QuitBet, as has been done with DietBet: QuitBet for individuals who are smoking at the initiation of the game and QuitBet “Maintainer” for relapse prevention for individuals who quit during a previous QuitBet.

CO testing was integrated into the QuitBet app. To initiate a CO test, participants pressed a button in the app, which led them through the process of connecting their iCO device and taking the test. As they exhaled into the iCO device, the QuitBet app automatically recorded a video of their face to verify their identity. Upon completion of the test, a “submit” button appeared to submit the video and the test result to WayBetter. The app only permitted participants to take one CO test per day, between the hours of noon and 11:59 p.m. local time. The CO testing window began at noon to prevent players from testing immediately upon waking in the morning, when CO was likely to be lowest (i.e., if they abstained overnight while sleeping). The game rules stated that a QuitBet referee (AP) would review CO test submissions and that a submission could be rejected if “anything seems off” and they could be asked to re-take the test. AP verified all CO test results (the number in ppm) and viewed videos in cases in which participants experienced an error in the submission process and had to submit their video via an alternative method (see “Technical Issues” in Results below).

During the 1st week (Day 1–7 pre-quit “warm up” week), game hosts provided participants with education about smoking cessation (see Table 1). Players were also encouraged to practice taking CO tests but these tests did not count toward winning. Any participant who requested to withdraw during the “warm up” week was given a refund of their $30 bet upon receipt of their iCO, with return shipping paid for by the study. Day 8 was the quit day; bets were non-refundable beginning on Day 8. Starting on Day 9 and continuing through Day 28 (a 20-day period), participants earned back $1 of their $30 bet per day for each day that their CO test result was ≤6 ppm. On Day 28, all remaining bet money was pooled into a “grand prize” pot. Participants whose CO result was ≤6 ppm on at least 19 of the 20 days (i.e., one “lapse” day allowed) were declared winners and split the pot equally (for this study, WayBetter did not retain any administrative cut). Participants were not asked to self-report their smoking behavior in the QuitBet app; all incentives were contingent only upon their CO test results.

Data Analysis

Analysis of survey (baseline and post-QuitBet) and game (i.e., CO test results) data was primarily descriptive and was conducted using SPSS v. 24. This pilot trial included two games (details below). We conducted all analyses for Game 1 and Game 2 players separately but there were very few significant differences; therefore, we present the results for both games combined but note the instances in which there were differences between games. Interview recordings and notes were reviewed to determine themes.

Results

Participant Characteristics

We met our recruitment goal. During a 30-day period between October–November 2019, the screening survey was completed 687 times, of which 235 were eligible, 106 provided consent, 77 completed the baseline survey, and 50 of these 77 paid their $30 bet and comprise the final sample.

The 50 participants were from 26 U.S. states. The first 23 players to enroll were placed into one game and the remaining 27 players were placed into a second game. The start dates for the two games were staggeredby 1week to avoid a long delay between recruitment and the game start date for those recruited earlier. Three of the 50 participants (all from Game 2) requested a refund of their $30 bet before Day 8 (i.e., the period during which players were permitted to request a refund, contingent upon returning their iCO to WayBetter; beginning on Day 8, the bet was non-refundable). Upon being sent instructions for how to return their iCO, only one of these three decided to return their iCO and received a refund; the other two did not return their iCO and therefore did not receive a refund. None of these three participants submitted any CO tests during the period that counted toward winning (Day 9–28) nor did any of them complete the post-QuitBet survey or interview but we nevertheless included them in analyses whenever possible (i.e., intent-to-treat analysis) as indicated below.

Baseline participant characteristics are shown in Table 2. Most (90%) had previously played a WayBetter game (DietBet, StepBet, and/or RunBet). Additionally, 4 of the 50 participants had participated in the 2017 QuitBet test game. Participants were willing to bet a mean of $47.80 (SD = $23.15, median = $40, range $30–$100), while 42% said they would pay an additional $60 for the iCO, 24% were not willing to pay for the iCO, and the remainder were willing to pay between $10–$40.

Table 2.

Baseline Participant Characteristics (N = 50)

Variable	Mean (SD) or N (%)
Age (years)*	40.4 (8.0)
Gender (female)	38 (76%)
Race (White)	47 (94%)
Ethnicity (Hispanic)	2 (4%)
Married	29 (58%)
Employed full-time	42 (84%)
Works day shift	42 (84%)
Bachelor’s degree or higher	24 (48%)
Income below $75,000	21 (42%)
Years smoked*	20.4 (9.4)
Cigarettes per day	16.1 (6.4)
FTCD	3.8 (1.4)
Lives with other smoker(s)	21 (42%)
Tried e-cigarettes in past	35 (70%)
Currently using e-cigarettes	1 (2%)
Previous quit attempt(s)	47 (94%)
Treatment History
NRT	31 (62%)
Varenicline	22 (44%)
Bupropion*	16 (32%)
Counseling	13 (26%)
Internet/app	15 (30%)
Previous WayBetter player	45 (90%)

^*Significant difference between Game 1 and Game 2 players (p < .05). Game 1 players were older, had smoked for more years, and were more likely have used bupropion in the past.

QuitBet Game Outcomes

Participation

Among the 50 participants, 82% submitted at least one CO test (Day 1–28), 58% submitted at least one CO test during the period that counted toward winning (Day 9–28), and 38% submitted a CO test, regardless of result (≤6 ppm or >6 ppm), on at least 75% of game days that counted toward winning (i.e., at least 15 CO tests). Also, 38 (76%) of participants posted at least once on the social feed.

Abstinence and Winners

In total, participants submitted 421 CO tests between Day 9 and 28, of which only nine tests were ≥7 ppm (“positive”). Figure 1 displays the number of negative (≤6 ppm) tests submitted by players; the distribution is bi-modal; most players submitted 0, 19, or 20 negative tests. Twenty-six participants (52%) submitted CO tests with a value of ≤6 ppm on at least two consecutive days between Day 9–28, suggesting that they made a serious quit attempt during which they abstained from smoking for at least 24 hr. Seventeen (34%; 9 of 23 players in game 1 and 8 of 27 players in game 2) submitted a CO test with a value of ≤6 ppm on at least 19 of the 20 days between Day 9–28 and were declared winners, of whom 8 (Five winners in Game 1 and 3 winners in Game 2; 47% of all winners) had the allowed “lapse” day. Game 1 winners received $74.22 each and Game 2 winners received $104.71 each.

Figure 1.

Number of CO Tests ≤6 ppm Submitted Between Day 9–28 (the 20-Day Period That Counted Toward Winning)

Note. Players who submitted 19 or 20 tests ≤6 ppm were winners.

Post-QuitBet Survey and Interview

Thirty-seven participants (74%; 17 winners, 20 non-winners) completed the post-QuitBet survey and 24 (48%; 15 winners, nine non-winners) completed the interview. We conducted intent-to-treat analyses for all smoking behavior outcomes (i.e., we included all 50 participants and used baseline values for non-respondents, assuming no change in their smoking behavior), whereas other analyses are restricted to respondents only.

Smoking Behavior and Other Smoking Cessation Treatment

Thirteen of the 50 participants (26%) reported past-week abstinence and 25 (50%; includes the 13 who reported abstinence plus 12 additional participants) reported smoking at least 50% fewer cigarettes during the past week compared to their pre-QuitBet baseline. Mean cigarettes per day (assuming no change for non-respondents) for the past week at post-QuitBet was 8.5 (SD = 7.9), compared to 16.1 (SD = 6.4) typical daily cigarettes at baseline. During QuitBet, 20 participants (40%; 10 winners, 10 non-winners) used at least one FDA-approved cessation medication, Eight (16%; six winners, two non-winners) used at least one behavioral treatment, and three participants (6%; all winners) used e-cigarettes. Compared to Game 2 participants, Game 1 participants were significantly more likely to report having used at least one medication (57% vs. 26%, X = 4.84, p = .03) and at least one behavioral treatment (30% vs. 4%, X = 6.60, p = .01).

QuitBet Experience and Satisfaction

Thirty-five (95% of respondents) participants reported that they would be interested in playing another QuitBet game and the same proportion said they would recommend QuitBet to others. Game ratings are shown in Table 3. There were no significant differences in these ratings between winners and non-winners except that winners’ mean rating was higher for the contribution of QuitBet to their ability to quit (p < .001). All respondents who disagreed that 28 days was a good length for QuitBet suggested that QuitBet be longer (M = 54.6 days, SD = 21.5, range 30–90 days). Most respondents (78%) thought a CO test of ≥7 ppm should be required prior to playing QuitBet, with Game 1 participants significantly less likely to endorse this than Game 1 participants (95% vs. 65%, Fisher’s exact test p = .04). Additionally, 73% of respondents thought the CO cut-off was “just right” and only one participant thought it should be higher.

Table 3.

QuitBet Ratings (1–5 scale with 5 being highest/best)

Item	Mean (SD)
Overall satisfaction	3.4 (1.2)
Clarity of rules	3.8 (1.0)
Helpfulness of hosts’ posts	3.6 (1.0)
Hosts’ supportiveness	4.1 (0.7)
Players’ supportiveness	3.2 (0.8)
QuitBet contribution to ability to quit	3.2 (1.4)
Satisfaction with winnings (winners only)	4.8 (1.3)
Concern about cheating	1.4 (0.9)
28 days is a good length	3.1 (1.4)

“False Negative” and “False Positive” CO Tests

Twelve respondents (32%; seven winners, five non-winners) reported at least one “false negative” Among these 12, nine had between 1 and 5 “false negative” days and reported smoking a maximum of 1–4 cigarettes on a false negative day. There were three outliers (one winner, two non-winners) who either had more false negative days and/or reported smoking a maximum of more than 4 cigarettes on a false negative day. During the game, only one participant reported a false positive (CO of 7 ppm). This participant was permitted to re-test using the free iCO companion app, achieved a CO of ≤6 ppm, and received the $1 incentive for that day.

Technical Issues

Eighteen respondents (49%; 11 winners, seven non-winners) reported experiencing a technical difficulty with CO testing but most were easily resolved and no participants reported that they were prevented from winning because of a technical issue. Most issues were related to establishing the connection between the iCO and their iPhone or submitting CO test videos. After taking a CO test, it could take up to 5 min to submit the video, during which time they had to keep the QuitBet app open and their iPhone “awake” (screen on). Sometimes, they had to make multiple submission attempts. In a small minority of cases (20 tests from seven participants out of a total of 592 tests submitted), participants were unable to submit a CO test via the QuitBet app and were permitted to use an alternative submission method (e.g., screenshot from the QuitBet app or the free iCO companion app instead of a video).

Interview Themes

Overall, participant feedback was positive, consistent with responses to the post-QuitBet survey, and patterns of responses did not seem to differ by winner status. About one third of participants said they didn’t post on or weren’t interested in the social feed for various reasons (e.g., only interested in the monetary incentives, they do not post on social media in general). Nevertheless, some of these participants said that they did read and benefit from the host’s posts. Other participants indicated that the social feed was an important part of their experience and they enjoyed participating actively and reading posts by the hosts and other players. Participants were divided as to whether they clicked on links posted by hosts. Participants who didn’t click the links offered various reasons why. For example, two players indicated that they did not click on the links because they were doing well with quitting and didn’t need or already were familiar with the information. One of these players said that if they were struggling they probably would have clicked on them. Another said that people don’t want to go through the additional step of clicking a link to obtain the information. Players offered a variety of suggestions for increasing engagement (encouraging posting on the social feed and submitting CO tests), such as special recognition for posting (e.g., host names a “most valuable player” of the week) or a financial incentive (e.g., entry into a raffle for submitting CO tests regardless of result). Consistent with the post-QuitBet survey data, more than half of the players suggested that the “warm-up” week be shortened or eliminated and/or the game be longer. Other suggestions included adding notification functions that would remind them to take CO tests and alert them to new posts or replies, and incentives (monetary and/or non-monetary) for submitting CO tests independent of smoking behavior.

Discussion

We pilot tested a digital social game for smoking cessation (QuitBet) that included a deposit contract and other social gaming features. The primary outcomes were feasibility, acceptability, and preliminary effectiveness. We met our goal of recruiting 50 participants within 1 month. Most participants (90%) had played WayBetter games previously. There were 235 completed screening surveys that met eligibility criteria; however, because the screening survey was anonymous and could be completed an unlimited number of times by the same individual, we cannot determine how many unique individuals were eligible. Nevertheless, among consenting participants who completed the baseline survey, 65% paid their bet, which is similar to a recent study in which 71% of those otherwise eligible paid the $50 deposit (Dallery et al., 2017). In another study, only 14% of those offered a CM program with a $150 deposit and the opportunity to earn back this $150 deposit plus $650 in additional incentives paid the deposit and enrolled versus 90% offered a CM program without a deposit but with similar maximum earning potential ($800) enrolled, although the deposit programs produced higher abstinence rates than the programs without deposits even after adjusting for decreased willingness to enroll (Halpern et al., 2015). Smaller deposits (e.g., $50 or less) will still have some degree of self-selection bias, but may better balance willingness to enroll with abstinence rates (Jarvis & Dallery, 2017). We also believe retention was acceptable, with 74% completing the post-QuitBet survey, especially given the timing of this survey (mid-to-late December, overlapping with winter holidays). Indeed, it has long been observed that retention rates for digital studies are significantly lower than traditional clinic-based RCTs (Eysenbach, 2005; Wiseman et al., 2019).

Nearly all participants (95%) who responded to the post-QuitBet survey expressed interest in playing a future QuitBet game and the same proportion said they would recommend QuitBet to others. Engagement was high at the beginning of the game. However, few players continued submitting CO tests after becoming ineligible to win, despite the opportunity to continue to earn $1 per day for each abstinent day. We assume they stopped submitting CO tests if they were smoking because there was no financial incentive to submit a positive CO test. Offering additional incentives for participation independent of CO test value (i.e., regardless of whether the player is smoking or abstinent) may help to sustain engagement after players become ineligible to be winners and promote continued attempts to quit.

QuitBet showed some evidence of preliminary effectiveness. Seventeen (34%) of participants were winners and 26% self-reported past-week abstinence on the post-QuitBet survey, which was completed within 3 weeks after the last day of the game. It is difficult to compare abstinence rates among CM studies because study designs and populations vary considerably (Notley et al., 2019). Also, we acknowledge that seven of the 17 winners admitted to having at least one “false negative” day and that therefore the true win rate (i.e., continuous abstinence for 20 days with the exception of 1 lapse day) may be lower (i.e., as low as 20% if we only count the 10 winners who denied false negatives). Nevertheless, we believe that our data indicate that participants reduced their smoking compared to pre-QuitBet. The maximum incentive potential (^~$50–75 net profit) was low compared to other traditional CM studies ($100s–$1000s) (Breen et al., 2020). However, previous research has shown that maximum incentive potential is not associated with quit rates with CM nor are CM outcomes moderated by participants’ income levels (Breen et al., 2020; López-Núñez et al., 2017; Notley et al., 2019). Additionally, QuitBet seemed to promote serious (i.e., 24-hr) quit attempts, with 52% of players submitting a CO test of ≤6 ppm on at least two consecutive days.

Players were mixed on whether the social feed added to their game playing experience and perceived that activity on the social feed was low, despite 76% of players posting on the game board at least once. It could be that 23–27 players per game were too few to sustain a high level of activity. Many WayBetter games have hundreds or thousands of players. Also, the preliminary version of the QuitBet app tested in this pilot trial lacked notification functions to alert participants to new posts and remind them to take CO tests. This may have limited activity on the social feed as well as lowered CO test submission rates. Game hosts encouraged players to submit CO tests, but players who did not open the app would not have seen these posts.

Many players wanted QuitBet to last longer than 28 days. If commercially available, users could play a series of consecutive 28-day games (i.e., as many as they want), which would extend the duration of game play indefinitely, addressing concerns about relapse when incentives are discontinued in CM programs (Petry, 2010). As previously noted, WayBetter has determined that 35%–50% of players in DietBet and StepBet have played two consecutive games.

Limitations

This was a small feasibility study that lacked a control group and was not powered to evaluate efficacy for smoking cessation. We also acknowledge many other limitations. First, given that the screening survey was available at a public link and was anonymous, the same individual could complete it an unlimited number of times and therefore we cannot determine how many unique individuals are represented in the 687 total surveys and 235 eligible surveys and thus cannot calculate the proportion of eligible individuals who enrolled.

Another limitation is that the participants were not representative of the general population of people who smoke in the U.S. (Hughes & Callas, 2010). Most participants were white, of relatively high socioeconomic status, and already familiar with WayBetter. Nearly a quarter of players were unwilling to spend $60 to purchase the CO testing device that was required to play the game and we expect that this proportion may be even higher among individuals with lower income. If QuitBet were brought to market, health insurance plans, employers, and/or governments could subsidize all or part of the device cost.

There were limitations to the CO testing schedule (once daily) and verification processes (i.e., not all videos were reviewed). About one-third of post-QuitBet survey respondents (seven winners and five non-winners) reported recording a CO value of ≤6 ppm on a day when they smoked at least one cigarette (a false negative). While requiring more frequent CO testing (e.g., twice daily testing) and/or specifying a lower CO cutoff (e.g., as low as 4 ppm; Dallery et al., 2017) may have prevented this smoking, such requirements would have increased burden and the risk of false positives, such that participants would lose money even if abstinent. We believe this is an acceptable trade-off in this real world context. Also, previous studies with lower cut-offs used professional-grade CO monitors for which there is data to support lower cut-offs; the recommended cut-off for the iCO is 6 ppm (Tuck et al., 2020). Furthermore, based on our data, we recommend that self-reported false negatives should be routinely collected and reported in CM studies. Currently, CM studies do not typically collect or report this data. One recent digital CM study did report information about false negatives that is roughly consistent with our data: Four of nine participants whose CO data suggested no smoking during the past week self-reported low levels of smoking during that period (Dallery et al., 2021). Related to this false negative issue, it is possible that our week “warm-up” period, during which CO tests did not count toward winning, could have inadvertently provided an opportunity for participants to practice “gaming the system” (i.e., determining how much they could smoke while keeping their CO under 7 ppm). However, we believe the relatively low win rate of 34% indicates that it is very unlikely that this behavior or any outright cheating (e.g., having a non-smoking friend take CO tests) occurred. Given the 1-day lapse allowance and the rate of false negative CO tests, we believe that our win rate highlights how difficult it is for people who smoke daily to significantly reduce and quit smoking. Related to this limitation of CO testing, in the post-QuitBet survey we only asked about smoking behavior for the past week, which may or may not have included game days depending on when the survey was completed; we did not ask about total cigarettes smoked during the entire 28-day game.

Another limitation is that we were unable to conduct detailed analyses of the social feed data, because the version of the QuitBet app used in this study did not track engagement metrics such as number of logins, time spent logged in, number of posts per user, etc. Therefore, we cannot determine the contribution of social feed participation to outcomes.

Finally, a significant proportion of winners reported that they used other treatments (medication and/or behavioral) during QuitBet. Given our small sample size, we cannot compare outcomes for players who did versus did not use other treatments nor can we definitively attribute success to QuitBet among those who did use other treatments. QuitBet is a “real world” program that was designed to be used as a standalone program or in combination with other treatments. Game hosts encouraged players to use other treatments and provided evidence-based information about medications and quitline counseling.

Future Directions

We recently received funding to conduct a larger randomized controlled trial (RCT) of QuitBet in which we plan to address many of the limitations of this pilot trial. We will have a larger advertising budget to recruit more players who do not have previous familiarity with WayBetter and help us determine how broad the market is for QuitBet. Also, the QuitBet app will be available for both iPhone and Android. As of December 2020, Android accounted for 48.8% of U.S. smartphone subscribers versus 50.7% for iPhone (Comscore MobiLens, 2021). Android users have a lower median household income than iPhone users (Comscore, 2014) and Android is significantly more popular than iOS among Black individuals (Smith, 2013). Therefore, including Android should facilitate recruitment of a more diverse sample with regard to race and income, which will be critically important to establish generalizability, acceptability, and efficacy. We will use an updated Bluetooth version of the iCO (called the iCOquit) that may reduce technical difficulties. Finally, a primary goal of the trial will be to specifically examine the contribution of the social feed to game outcomes; half of the participants will receive a version of QuitBet that does not include the social feed (deposit contract only). We plan to have more players per game, add notification functions, and program the app to track various engagement metrics including more detailed data about social feed participation. We will also ensure that a random proportion of CO test videos are reviewed by game referees and encourage players to share CO test results (video and/or screenshot) in the social feed for additional “crowdsourced” verification. Finally, we will examine whether longer-term abstinence can be achieved through playing a series of consecutive QuitBet games and whether outcomes differ for players who are more engaged in the social feed and who report use of other smoking cessation treatments in addition to QuitBet. Participants who were winners in a previous QuitBet game will play a QuitBet “Maintainer” relapse prevention game for their next game.

Conclusion

In conclusion, using a deposit contract as part of a digital social game for smoking cessation was feasible and acceptable. We have presented preliminary evidence that this game can help induce quit attempts and reduce cigarettes smoked per day and is potentially more scalable and sustainable than previous CM programs. Digital social games are a novel cessation strategy focused on providing support for health behavior change in a fun and supportive environment.

Public Significance Statement.

This study suggests that a commercial, digital smoking cessation “game” administered via a smartphone application in which the participants earned money for quitting smoking was feasible and well-liked. However, a larger study is needed to determine if this game is effective for helping people quit smoking and whether a sustainable business model can be established.