Prize-Based Incentives for Smoking Cessation Among People With HIV: A Sequential Multiple Assignment Randomized Trial

Abstract

Introduction:

Contingency management (CM) is an incentive-based approach that has demonstrated efficacy for smoking cessation in various populations. There is an unmet need for feasible and effective smoking cessation interventions in people with HIV (PWH). The purpose of this study was to assess the efficacy of prize-based CM for smoking cessation in PWH using a Sequential Multiple Assignment Randomization Trial (SMART) design selected to tailor intervention intensity based on early treatment response.

Methods:

During phase I, 129 participants were randomly assigned to high-magnitude prize CM (HM-CM) or standard of care (SoC) for 4 weeks. Participants who did not reduce smoking were randomized in phase II to continued counseling with HM-CM plus monitoring support or only continued monitoring support for 8 weeks. Participants who reduced smoking were randomized to booster monitoring with low-magnitude CM or no additional care. Outcomes were biochemically verified smoking reduction and 7-day abstinence prevalence at posttreatment, 6-month, and 12-month follow-up.

Results:

Phase I responders (based on biochemical indicators of smoking reduction) were significantly less likely to return to smoking (during treatment and at 6 and 12 months) if they received low-magnitude incentives. Notably, initial exposure to CM versus SoC did not increase the rate of phase I response, and high-magnitude incentives later in treatment did not lead to greater smoking cessation for early-treatment nonresponders.

Conclusions:

Weekly CM sessions in the first 4 weeks of smoking cessation intervention did not perform significantly better than SoC. However, brief booster CM sessions aimed at maintaining early smoking cessation hold clinical promise and warrant further investigation.

Implications:

This represents the first trial to examine the use of CM for smoking cessation among PWH within the context of a SMART design.

Clinical Trial Registration:

NCT01965405 (https://clinicaltrials.gov/study/NCT01965405)

People with HIV (PWH) are disproportionately affected by tobacco smoking-related illnesses.¹ Nearly half of PWH receiving HIV treatment are daily smokers.² PWH who smoke experience a greater risk of pulmonary and cardiovascular diseases.^3,4 Smoking also exacerbates HIV disease including higher viral loads, lower CD4 counts, and lower quality of life; the latter is associated with findings of lower antiretroviral medication adherence among PWH who smoke tobacco.^3,4 It is vital that PWH receive effective smoking cessation resources, considering evidence that smoking abstinence is associated with decreased HIV symptom burden.⁵

PWH report numerous benefits of quitting smoking, such as increased T-cell count, lower infection, greater commitment to HIV care, and others.⁶ Furthermore, the number of studies examining the efficacy of cessation approaches for PWH has increased substantially in recent years.⁷ Pharmacotherapies such as nicotine replacement (NRT), bupropion, and varenicline are effective,⁸ and behavioral treatments incorporating motivational interviewing and cognitive approaches have also demonstrated some success.^9,10 Additionally, mobile health interventions, some of which incorporate online community support,¹¹ and cellphone-delivered counseling¹² are demonstrating effectiveness. These successes, however, need to be viewed within the context of ongoing challenges. For example, medication adherence is often low.¹³ Furthermore, not all trials demonstrate the effectiveness of evidence-based psychosocial interventions in comparison to usual care,^13,14 and cessation rates among PWH tend to be lower than among the general population.¹⁵

Given the mixed support for current interventions, there is a need for alternative effective stop-smoking treatments for PWH. One such intervention, contingency management (CM), is an incentive-based approach that uses tangible reinforcement to encourage individuals to complete a target behavior (eg, provide a biomarker test result consistent with smoking cessation). Cigarette smoking is maintained through operant conditioning via reinforcing effects of nicotine and related stimuli, and CM attempts to re-engineer the environment to help patients obtain positive consequences of abstinence while minimizing the positive consequences of tobacco use. By reinforcing an alternative behavior that is inconsistent with smoking, CM can counteract reinforcement from chronic smoking. Prior studies have demonstrated the efficacy of CM for treating numerous substance use disorders (eg, see meta-analysis¹⁶), and monetary CM has been shown to reduce smoking in a variety of populations.^17–19

In the present study, we examined the efficacy of a lower-frequency prize-based CM for smoking cessation among PWH. In prize CM, participants who complete a specified target behavior (eg, reduce their level of cotinine or expired carbon monoxide [CO]) can earn escalating numbers of prize draws of varying magnitude. Several studies support the efficacy of prize CM in people with substance use disorders.²⁰ An emerging literature examined prize CM for smoking.^18,21–23 Although some have examined the efficacy of CM for HIV pre-exposure prophylaxis²⁴ and for PWH who use substances,^25,26 no published studies have applied this intervention to smoking cessation among PWH (but we are aware of ongoing trials that will provide data in the near future²⁷).

Treatment tailoring based on initial response is important for addressing individual-patient smoking-cessation needs.²⁸ People who are unable to achieve early abstinence likely need a higher level of care than those who achieve early smoking cessation. We used a Sequential Multiple Assignment Randomization Trial (SMART)²⁹ design, intended for building adaptive, time-varying interventions. As described later, participants were reassessed after an initial treatment phase to determine the appropriate level of subsequent care.

The present study aimed to assess the efficacy of adding a CM intervention to standard care (monitoring, support, and bupropion) for early treatment response (biochemically verified reduction in smoking at the end of 4 weeks) and smoking cessation (biochemically verified 7-day point prevalence at the end of treatment and 6- and 12-month post-baseline) among PWH. The treatment was administered within the context of a SMART trial designed to adjust treatment intensity based on early treatment responsiveness. We hypothesized that our CM intervention would increase the rates of smoking reduction at treatment week 4 (end of phase I) and increase the rates of smoking abstinence by the end of treatment for both phase I treatment responders and nonresponders.

Methods

Participants

Participants randomized were 129 daily cigarette smokers recruited from patients at a large, midwestern outpatient infectious diseases unit specializing in HIV care. Inclusion criteria were (1) age ≥18 years, (2) able to read and understand English, and (3) smoking ≥10 cigarettes/day. Participants were excluded if they (1) were actively suicidal or had uncontrolled manic or psychotic symptoms requiring immediate care, (2) were in recovery for gambling disorder, (3) had contraindications for bupropion treatment (eg, presence of epilepsy/seizure disorder, use of MAO inhibitors, presence of eating disorders, or very low weight), or (4) were already participating in other smoking cessation interventions. Participants who did not meet the inclusion/exclusion criteria were referred to other smoking cessation providers, to their insurance company for referral, or to the State of Michigan ICANQUIT helpline which provides telephone counseling, self-help resources, and smoking cessation medication at no cost.

Procedures and Design

All procedures were approved by the Wayne State University Institutional Review Board. Individuals who expressed interest in the study underwent a brief telephone screen to determine initial inclusion. Eligible participants were scheduled for an in-person intake at which time they provided written informed consent.

Figure 1 illustrates the overall SMART study design. Those who qualified provided baseline urine samples and were randomized to standard of care (SoC consisting of medication [bupropion], monitoring and support [MS]) treatment, or SoC plus an initial high-magnitude CM intervention to initiate abstinence (HM-CM) once weekly for 4 weeks. Treatment assignment was balanced by gender and average daily number of cigarettes smoked (<20/day vs. ≥20/day). Sequentially numbered randomization envelopes prepared by the study statistician using a stratified permuted-block design were opened by the study research staff or PI who were unaware of the content of each envelope. Participants were advised that the use of nicotine-containing products such as nicotine replacement or electronic cigarettes would result in a positive cotinine test and that the use of smoked cannabis products may result in a positive CO result.

Figure 1.

SMART design.

A second informed consent process was conducted before participants were rerandomized for treatment phase II (8 weeks) based on their phase I treatment response using the same randomization procedures. Treatment response was defined as a substantial reduction in the cotinine level (≥2 point reduction on a semiquantitative scale described later) or recent abstinence based on expired CO at their final visit (≤3 ppm).

Phase I treatment nonresponders (phase IIa) were randomized to (1) continued bupropion, counseling, and monitoring of smoking (MS) or (2) continued MS plus a high-magnitude prize CM procedure (MS+CM). Phase I treatment responders (phase IIb) were randomized to (1) no additional intervention (NAT) beyond 8 additional weeks of bupropion (consistent with SoC) or (2) continued bupropion, counseling and monitoring, and 8-week low-magnitude booster prize CM (LI+CM).

Data were collected at baseline, 4 weeks (end of phase I), 12 weeks (end of phase II), 6 months, and 12 months post-baseline. Primary outcomes were treatment response (end of phase I only) and 7-day point prevalence abstinence at 12 weeks and follow-up visits. Seven-day point prevalence smoking abstinence was assumed if a participant self-reported no smoking in the 7 days before the assessment point using the Timeline Follow-Back and provided an expired CO ≤3 ppm or cotinine score lower than the most recent session. Participants were paid $35 for intake and $25 for each follow-up interview, which occurred at 4 weeks, 12 weeks, 6 months, and 12 months. Data collection occurred from January 2014 to October 2020 (end of grant). The full study protocol can be accessed by request to the first author.

Assessment

Demographics

At baseline, participants reported gender, age, marital status, race/ethnicity, education, employment status, insurance status, and annual income. The Structured Clinical Interview for DSM-IV³⁰ was used to collect information on psychopathology (for exclusion). Gambling disorder was assessed using the NORC DSM Screen for Gambling Problems (NODS³¹).

Smoking History

Information was collected at baseline on age of first smoking, family smoking history, and current number of cigarettes daily. The Fagerström Test for Nicotine Dependence questionnaire was used to assess tobacco use disorder.³² Additional smoking/tobacco-related measures were completed at each major assessment point. Timeline Follow-Back-Cigarette (TLFB-C) is a self-report, calendar-based measure of cigarette smoking frequency.³³ TLFB-C was completed at each follow-up and each treatment session. Expired CO levels were assessed at baseline using an EC50-MP Micro CO monitor (Bedfont) at each assessment point and every treatment session. CO levels ≤3 ppm are generally considered negative for smoking.³⁴ Cotinine is a metabolite of nicotine indicating cigarette smoking (and the use of other nicotine-containing products). Urine samples were collected at baseline, each visit throughout treatment, and each follow-up. Urine cotinine levels were analyzed using the Accutest NicAlert test-strip system (JANT Pharmacal Corporation, Encino, CA). Cotinine levels ≤100 ng/mL are usually considered negative for cigarette smoking.

Treatment Satisfaction

Participants rated how satisfied they were with their treatment experience at the end of phase I and phase II with scores ranging from 1 (very dissatisfied) to 7 (very satisfied).³⁵

Phase 1 Treatments

Standard-of-Care Treatment

SoC consisted of monitoring expired CO and urinary cotinine levels, brief counseling based on accepted SoCs for tobacco use disorder using the 5A/5Rs,³⁶ and evidence-based pharmacotherapy (bupropion). Consistent with SoC models, participants met with a study therapist once weekly for 4 weeks to provide samples, monitor medication compliance, and participate in 15-minute counseling sessions focused on smoking cessation. The clinician provided the participant with a smoking cessation self-help quit guide³⁷ and reviewed self-help materials with the participant. The self-help materials cover reasons for quitting smoking, preparing to quit, obtaining social support, skills building, use of medications, and preparing for return to use.

Participants were also provided extended-release bupropion, which has demonstrated efficacy in assisting smoking cessation.^38,39 We followed a flexible dosing procedure at 150 mg/d for the first 3 days, followed by a dose increase for most patients to 300 mg/d. Bupropion treatment continued for 12 weeks throughout phase I and phase IIa/b, with a 2-week taper (150 mg in week 11 and 75 mg in week 12 to stop at the end of week 12).

SoC Plus High-Magnitude Prize Contingency Management (HM-CM).

Participants in the HM-CM condition received SoC and earned once-weekly chances to win prizes when they met the criteria for early reduction/abstinence from cigarette smoking.^20,35

Participants received prize draws during their weekly session when their cotinine levels were lower than the most recent level or ≤2 on the 0–6 semiquantitative scale. If a participant began the study with a value of 6, for example, a prize draw was earned once the value reached 5 and again when the value reached 4. If a participant obtained a lower cotinine score but then returned to use, eligibility for prize draws was reset from the new level.

Participants earned draws from a prize bowl for each session they met the above criteria for smoking reduction/cessation. On the first day, participants could earn one draw, and the number of draws could escalate with each subsequent reduced cotinine test. Thus, by the fourth week of the program, participants could draw four times from the prize bowl for a maximum of 10 draws across 4 weeks. Prize-draw resets occurred for return to smoking or unexcused missed sessions. Phase I prize bowl was filled with 50 slips of paper, with 60% (30) resulting in a “large” prize ($20), 30% (15) resulting in a “super” prize (worth $50), and 10% (5) resulting in a “jumbo” prize (worth $100). A 100% prize bowl was used because of the lower frequency of reinforcement. Over 4 weeks of treatment, participants could earn up to approximately (maximum values were based on a variable schedule) $370 in reinforcement if they abstained throughout phase I.

Phase IIa (Nonresponder) Treatments

Nonresponders continued to receive medical monitoring of bupropion treatment for 8 additional treatment weeks, with additional conditions as noted later.

Counseling and Monitoring of Smoking.

Participants randomized to the MS condition met with a study counselor twice weekly to review progress toward smoking cessation. Participants provided breath samples for CO testing and urine samples for cotinine testing weekly. Counselors provided ongoing support and encouragement and feedback on results of weekly testing.

Counseling and Monitoring of Smoking Plus Contingency Management (MS+CM).

Participants randomized to the MS+CM condition received the same treatment as the MS condition, plus prize-CM for smoking reduction/abstinence twice weekly for 8 weeks. The prize-drawing method was similar to phase I, but drawings were scaled to be more frequent than phase I.

Participants in the MS+CM group who met the criteria for recent reduction/abstinence received one draw on day 1 of treatment, and the number of draws could escalate with each consecutive negative test, up to 10 maximum draws on a given day. If the participant presented with a cotinine level exceeding the cutoff for reinforcement, refused to provide a sample, or did not attend 1 day (other than excused absences), his/her number of draws was reset to one draw for the next negative sample provided. Resets could be reversed (i.e., number of prizes restored to the highest level achieved before the reset) once the participant achieved three consecutive negative samples. Furthermore, for every third consecutive test that met the reinforcement criteria, the participant received five bonus draws. Phase II prize fishbowl consisted of 50% of slips (250) not resulting in a prize (“good job”), 42.6% “small” prizes (213; worth about $2), 7% “large” prizes (35; worth $20), and 0.4% “jumbo” prize (2; worth $100). CM participants could earn up to 115 draws plus 25 bonus draws over the 8-week treatment; participants could earn up to approximately $371 in prizes (maximum values were based on a variable schedule).

Phase IIb (Responder) Treatments

Responders (like nonresponders) continued to receive medical monitoring of bupropion treatment for 8 more treatment weeks, with additional conditions as noted later.

No Additional Treatment

Phase I treatment responders who were assigned to the no additional treatment (NAT) condition received no additional counseling or monitoring for smoking following phase I other than outcome assessments.

Counseling and Monitoring of Smoking Plus Low-Intensity CM (LI±CM).

Phase I responders assigned to the LI+CM condition met with a counselor for up to six occasions, weekly for 4 weeks and then biweekly for 4 weeks. Additionally, these participants had the opportunity to draw prizes for negative urine cotinine readings using the same prize urn as in the MS+CM condition. Participants started with 5 draws, and these draws escalated to a maximum of 10 for subsequent negative cotinine tests. LI+CM participants could earn up to 45 draws plus 5 bonus draws over the 8-week treatment period; participants could earn up to approximately $133 in prizes (maximum values were based on a variable schedule).

Data Analysis

Preliminary analyses were conducted to examine demographic and smoking differences by group using chi-square and t tests. All analyses were intent-to-treat, and those who did not attend their assessment were assumed to be nonabstinent. The primary hypothesis of testing the efficacy of HM-CM and SoC compared to SoC was examined using chi-square analysis. The outcomes for phases IIa and IIb were tested using logistic regression. The follow-up visits were analyzed using logistic regression. A priori power analysis (two-sided test with 80% power and alpha of 0.050) assuming medium–large effect sizes revealed needed sample sizes of N = 78 for phase I and N = 94 for each of the phase II treatments.

t Tests were used to examine mean differences in treatment satisfaction by group for phases I, IIa, and IIb. The total costs of prizes earned per day of smoking reduction were summarized. Raw CO and cotinine scores at each assessment point are presented in Supplementary Figures.

Results

Participant Characteristics

Consistent with the overall SMART study design (Figure 1), three CONSORT diagrams depict participant flow through the study phases. Figure 2 shows that the follow-up assessment rate was 91.5% (118/129) at the end of phase I. For treatment nonresponders who continued into phase II (Figure 3A), assessment rates exceeded 80%: 82.9% (63/76) for post-phase II, 85.5% (65/76) for 6 months, and 80.3% (61/76) for the 12-month follow-up. For treatment responders who continued into phase II (Figure 3B), assessment rates also exceeded 80%: 88.1% (37/42) for post-phase II, 90.5% (38/42) for 6 month, and 88.1% (37/42) for the 12-month follow-up. CM and non-CM treatment conditions did not significantly differ on demographic or baseline smoking-related variables during any phase of treatment (Tables 1 and 2).

Figure 2.

Phase I enrollment (initial treatment).

Figure 3.

(A) Phase IIa enrollment (phase I treatment nonresponders).. (B) Phase IIb enrollment (phase I treatment responders).

Table 1.

Demographic and Smoking Variables for Standard of Care (SoC) and High Magnitude Contingency Management (HM-CM) Treatment Assignments at Study Phase I

Variable	Phase 1 Treatment		Value
	SoC (n = 65)	HM-CM (n = 64)
Age (years)	47.9 (10.6)	49.3 (9.1)	t(127) = −0.77
Gender
Male	45 (69.2%)	43 (68.3%)	χ2(1) = 0.01
Female	20 (30.8%)	20 (31.7%)
Race (African American/Black)	60 (92.3%)	61 (95.3%)	χ2(1) = 0.50
Married	7 (10.8%)	5 (7.8%)	χ2(1) = 0.33
Education (years)	12.2 (1.8)	12.1 (1.8)	t(127) = 0.54
Employed	23 (35.4%)	19 (29.7%)	χ2(1) = 0.48
Income (Median/IQ range)	$8309 (7449)	$8796 (3861)	U = 2039.00
Fagerstrom Score	5.1 (1.8)	5.6 (1.8)	t(127) = −1.70
Age first regular smoking (years)	16.5 (4.4)	18.0 (4.6)	t(127) = −1.87
Cigarettes/day (#)	15.0 (6.9)	15.0 (7.0)	t(127) = −0.05
Baseline carbon monoxide (ppm)	14.7 (7.9)	15.0 (7.0)	t(127) = −1.38

No comparisons were statistically significant at p ≤ .05 (two-tailed). All p values >.05.

Table 2.

Demographic and Smoking Variables for Standard of Care (SoC) and Contingency Management (HM-CM) Treatment Assignments for Study Phase II Treatment Nonresponders and Treatment Responders

Variable	Phase IIa (Nonresponders)		Value	Phase IIb (Responders)		Value
	Non-CM(n = 38)	CM (n = 38		Non-CM(n = 22)	CM (n = 20)
Age	47.4 (9.9)	48.4 (10.0)	t(74) = −0.45	51.6 (9.5)	45.8 (10.9)	t(40) = 1.86
Gender			χ2(1) = 0.26			χ2(1) = 0.01
Male	28 (73.7)	26 (68.4)		14 (63.6)	13 (65.0)
Female	10 (26.3)	12 (31.6)		8 (36.4)	7 (35.0)
Race (African American)	35 (92.1)	36 (94.7)	χ2(1) = 0.21	21 (95.5)	18 (90.0)	χ2(2) = 0.47
Married	5 (13.2)	3 (7.9)	**	0 (0)	2 (10.0)	**
Education	12.4 (1.7)	11.8 (1.8)	t(74) = 1.42	12.5 (2.4)	12.1 (1.7)	t(40) = 0.69
Employed	12 (31.6)	15 (39.5)	χ2(1) = 0.52	5 (22.7)	6 (30.0)	χ2(2) = 0.29
Income (median/IQ range)	$8796 (7362)	$8910 (7559)	U = 746.00	$8808 (2745)	$8790 (1152)	U = 160.00
Fagerstrom Score	5.6 (1.8)	5.3 (1.9)	t(74) = 0.63	5.0 (1.8)	5.4 (1.8)	t(40) = −0.54
Age first regular smoking	16.9 (4.4)	18.1 (5.1)	t(74) = −1.09	18.5 (4.5)	16.0 (4.1)	t(40) = 1.96
Cigarettes/day	16.7 (8.2)	15.0 (6.0)	t(74) = 1.04	14.5 (7.5)	12.8 (5.2)	t(40) = 0.87
Baseline carbon monoxide (ppm)	19.7 (11.1)	15.9 (8.6)	t(74) = 1.66	12.7 (7.9)	11.9 (6.6)	t(40) = 0.37
Phase I treatment assignment			χ2(1) = 0.21			χ2(1) = 0.35
SoC	20 (52.6)	22 (57.9)		9 (40.9)	10 (50.0)
HM-CM	18 (47.4)	16 (42.1)		13 (59.1)	10 (50.0)

All p values >.05.

^**Fisher’s exact test nonsignificant..

Phase I Outcomes

The number of sessions attended did not differ by group (M = 2.5 for both groups). At the end of the fourth treatment week, 32.6% of participants were treatment responders but rates did not differ by group (35.9% for CM and 29.2% for SoC [χ²{1, N = 129}] = 0.66, p = .42). In total, 91.5% (n = 118/129) of the original phase I sample was randomized in phase II.

Phase I Nonresponder Phase IIa Outcomes

There was no difference in the number of sessions attended (MS+CM M = 4.1 [SD = 4.5] vs. MS M = 3.7 [SD = 3.8]). The 7-day point prevalence showed no difference by MS+CM versus MS group at posttreatment (7.9% and 5.3%), 6 months (2.6% and 2.6%), and 12 months (5.3% and 0%; ps > .05).

Phase I Responder Phase IIb Outcomes

Participants in the LI+CM condition attended 3.5 (SD = 2.3) out of six possible sessions, whereas participants in the NAT condition were not required to attend any sessions. Posttreatment 7-day point prevalence data showed that treatment responders who received CM during phase II were more likely to be abstinent (50%) than those who did not receive CM (13.6%; χ²[1, N = 42] = 6.72, p = .01; OR = 6.33; 95% CI = 1.41 to 28.39). Those who received CM also were more likely to be abstinent at 6 months (45.0% and 13.6%; χ²[1, N = 42] = 5.20, p < .05; OR = 5.18; 95% CI = 1.15 to 23.29) and 12 months (40.0% and 9.1%; χ²[1, N = 42) = 5.78; OR = 6.67; 95% CI = 1.21 to 36.74) compared to NAT participants.

Treatment Satisfaction

At the end of phase I, participants in both treatment conditions endorsed a high degree of treatment satisfaction (SoC M = 6.3, SD = 1.1; HM-CM M = 6.2, SD = 1.2; p = .74). Similarly, there was no difference in mean satisfaction for phase I treatment nonresponders (MS M = 6.4, SD = 1.0; MS+CM M = 6.0, SD = 1.4; p = .25) or phase I treatment responders (NAT M = 6.7, SD = 0.57; LI+CM M = 6.8, SD = 0.54; p = .35).

Incentive Earnings

During phase I, participants treated with CM earned an average of $52 (SD = $101.88) in incentives across 4 weeks. Participants assigned to the CM condition among phase I nonresponders received an average of $3 (SD = $15.33) in incentives across 8 weeks of phase IIa. Participants assigned to the CM condition among the phase I responders received an average of $73 (SD = $102.49) in incentives across 8 weeks of phase IIb.

Discussion

This is the first study to examine the use of a prize-based CM for smoking cessation among PWH. The study hypotheses received mixed support. Once-weekly CM did not significantly improve early smoking cessation nor did continuing twice-weekly CM increase cessation among individuals who did not initially respond to treatment. However, among those initially responsive regardless of phase I intervention (35.9%), a small, infrequent CM payment during the later 8 weeks of treatment facilitated continued smoking abstinence among more participants, and these gains were maintained at 6- and 12-month post-baseline. If replicated, these findings could encourage the use of incentives for smoking cessation.

Previous studies have found positive results for the efficacy of CM for smoking cessation in various populations. A meta-analysis examining CM for smoking among individuals with substance use disorders found CM to be superior to comparison groups with medium short-term effect sizes.¹⁹ Our overall abstinence rate at the end of treatment was roughly 22.4% for people who received CM during phase IIA or IIB, which is somewhat lower than the average 36% found across CM studies.¹⁹ In contrast, our end-of-treatment cessation rate was 8.3% for those who received no CM in phase II, which is consistent with the average 7.8% found across studies.¹⁹ Our study findings are also consistent with another meta-analysis that focused on a broader range of incentive programs in mixed populations,⁴⁰ which also revealed that CM can have longer-term effects on smoking cessation. Importantly, given the inclusion requirement of ≥10 cigarettes per day, ours may be considered to be a relatively heavier smoking sample.

CM delivered in our study may have differed appreciably from previous investigations. Our initial treatment phase included once-weekly CM delivery. Earlier CM smoking studies used more frequent visits up to 2–3 visits daily.^17,18,23 Our study findings may be more consistent with others that used a less frequent reinforcement schedule for CM, although some studies using less frequent schedules had null findings.⁴¹ Given myriad transportation and other barriers experienced by PWH,⁴² a frequent visit schedule is not feasible with our sample. We attempted to offset the less frequent visit schedule with higher-magnitude reinforcement for each visit, consistent with research showing that higher-magnitude reinforcement augments behavior change,⁴³ although magnitude effects are not universal.⁴⁴

Our procedures worked well with those who experienced early cessation, as those who went on to receive infrequent booster CM sessions were less likely to return to smoking during and after treatment. However, it is possible that a more frequent visit schedule might have led to greater cessation. Given the challenges of administering CM on an outpatient basis, future research should explore the use of remote mobile health interventions, which have demonstrated efficacy in CM for smoking cessation with other populations.^45,46

It is also important to consider our findings in light of cessation rates using other forms of treatment among PWH. Biologically verified 7-day point-prevalence cessation estimates vary substantially across studies and treatments, with combined psychosocial+medication interventions showing the highest efficacy (see Ashare et al.⁷ for review). A systematic review identified 28 intervention studies in PWH consisting of counseling, pharmacotherapy, and information technology approaches.⁴⁷ The included studies differed in several ways: only 15 were randomized clinical trials, smoking outcomes varied and not all studies used biological abstinence, and counseling and pharmacotherapy approaches differed. Thirteen studies were included that used both face-to-face counseling and pharmacotherapy with only 7 showing improvements in cessation. Thus, behavioral smoking cessation for PWH remains in a nascent stage, but we have noted a recent concerted effort led by the United States National Cancer Institute has increased the number and breadth of rigorous controlled trials to develop and identify efficacious interventions.⁷

CM study findings are mixed with regard to longer-term outcomes after removal of incentives, with some studies demonstrating no long-term differences from usual care,¹⁹ and others finding longer-term cessation differences.^40,48 In our study, those who initially responded to CM were less likely to return to smoking and experienced better long-term outcomes if they received low-magnitude incentives during the final 8 weeks of treatment. This finding appears consistent with work demonstrating graded removal of reinforcers to lead to longer-term behavior change in token economies (Review⁴⁹). To our knowledge, no previous study of CM for smoking cessation has investigated a tailored CM for those who were initial treatment responders. Thus, our results have important implications for preventing return to use among individuals once the cessation intervention ends. Specifically, it is important to explore the use of booster CM sessions as one method to maintain abstinence among individuals who are trying to stop smoking.

Limitations

Our findings are limited by the small sample size and recruitment from a single urban clinic. Thus, the present findings may not generalize to other populations or other clinics in other areas. Furthermore, we were unable to conduct outcome analyses by subgroups (eg, sex or other substance use). Our CM intervention, which was weekly throughout phase I, used a less frequent monitoring schedule than is typical of past studies that demonstrated greater smoking cessation success.^17,18,23 We also did not assess the co-use of nicotine replacement or electronic nicotine delivery systems. Our finding showing that treatment responders maintain their abstinence with booster CM sessions is important but needs replication.

Conclusions

This study is the first to examine the use of a prize CM intervention for smoking cessation among PWH. It is also the first to explore incentives for smoking cessation within the context of a SMART design. Participants who received the weekly reinforcement schedule in phase I were not more likely than those without weekly reinforcement to be early treatment responders. Furthermore, high-magnitude incentives later in treatment did not lead to greater smoking cessation among participants who were nonresponsive in the early phase of treatment. Lack of response to treatment may be due to the less frequent reinforcement schedule than typically used in CM studies, which has broader implications for CM implementation. Our most important finding is that those who initially responded to treatment were significantly less likely to return to smoking if they continued to receive low-magnitude incentives throughout the treatment period. These individuals also experienced the benefits of CM at the 6- and 12-month follow-up. Taken together, these findings help inform decisions on how CM may be used to promote smoking cessation among PWH.

Supplementary Material

Supplementary material is available at Nicotine and Tobacco Research online.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.