A SMART Approach to Treating Tobacco Use Disorder in Persons with HIV (SMARTTT): Rationale and design for a hybrid type 1 effectiveness-implementation study

Abstract

Background:

Tobacco use disorder is a leading threat to the health of persons with HIV (PWH) on antiretroviral treatment and identifying optimal treatment approaches to promote abstinence is critical. We describe the rationale, aims, and design for a new study, “A SMART Approach to Treating Tobacco Use Disorder in Persons with HIV (SMARTTT),” a sequential multiple assignment randomized trial.

Methods:

In HIV clinics within three health systems in the northeastern United States, PWH with tobacco use disorder are randomized to nicotine replacement therapy (NRT) with or without contingency management (NRT vs. NRT+CM). Participants with response (defined as exhaled carbon monoxide (eCO)-confirmed smoking abstinence at week 12), continue the same treatment for another 12 weeks. Participants with non-response, are re-randomized to either switch medications from NRT to varenicline or intensify treatment to a higher CM reward schedule. Interventions are delivered by clinical pharmacists embedded in HIV clinics. The primary outcome is eCO-confirmed smoking abstinence; secondary outcomes include CD4 cell count, HIV viral load suppression, and the Veterans Aging Cohort Study (VACS) Index 2.0 score (a validated measure of morbidity and mortality based on laboratory data). Consistent with a hybrid type 1 effectiveness-implementation design and grounded in implementation science frameworks, we will conduct an implementation-focused process evaluation in parallel. Study protocol adaptations related to the COVID-19 pandemic have been made.

Conclusions:

SMARTTT is expected to generate novel findings regarding the impact, cost, and implementation of an adaptive clinical pharmacist-delivered intervention involving medications and CM to promote smoking abstinence among PWH.

ClinicalTrials.gov identifier:

NCT04490057

1. INTRODUCTION

Antiretroviral treatment has transformed HIV into a chronic medical condition, with the life expectancy of many persons with HIV (PWH) approaching that of the general population.¹ Globally, however, the benefits of antiretroviral treatment are significantly threatened by cigarette smoking, which is highly prevalent across diverse samples of PWH.² This is specifically true in the United States, where one-third of a national sample of PWH report current smoking.³ Importantly, potentially by increasing immune activation and T cell dysfunction,⁴ cigarette smoking contributes to the leading causes of death among PWH: cardiovascular disease, malignancy, and pulmonary disease.⁵ As a result, for PWH receiving antiretroviral treatment, smoking is responsible for more years of life lost than treated HIV.^6,7

Accordingly, national organizations recommend routine use of medication and behavioral-based treatments to promote smoking cessation.^8–11 Yet research to promote smoking abstinence among PWH is needed as such widely available treatments have demonstrated only modest benefits in promoting short-term smoking abstinence with limited data to suggest long-term benefits.¹² Further, prior studies have been limited for several reasons. First, these studies had several design challenges, including small sample sizes; non-randomized treatment assignment; and focuses on behavioral interventions with and without nicotine replacement therapy (NRT),¹³ generally neglecting other important treatment options, including varenicline (a partial nicotine receptor agonist)^14–16 and contingency management (CM, the use of tangible rewards to promote verifiable changes in behavior).¹⁷ Second, despite their growing role in providing treatment to address tobacco use^18–20 and care of PWH in HIV clinics (e.g., antiretroviral treatment adherence, polypharmacy),^21–23 no prior studies have focused on clinical pharmacists for delivering tobacco treatment in HIV clinics. Third, few studies have evaluated the impact of an adaptive treatment approach for smoking,^24,25 particularly among PWH, although this approach has been successfully applied to other chronic conditions (e.g., pain, depression, alcohol)^26–32 given variability in treatment responses.^17,33 Fourth, there are mixed data regarding the impact of smoking on HIV-specific outcomes, including CD4 cell count and HIV viral load,^34–36 yet no studies have evaluated the impact of tobacco treatments on these key HIV-specific outcomes¹² or the Veterans Aging Cohort Study (VACS) Index 2.0, a more comprehensive, validated measure of morbidity and mortality.^37–43 Lastly, there is a paucity of data on factors impacting implementation of tobacco treatment delivery in HIV clinics.^12,13

To address these gaps, we are conducting a new study, entitled A SMART Approach to Treating Tobacco Use Disorder in Persons with HIV (SMARTTT), to identify the optimal adaptive approach involving first-line tobacco medications and CM to promote exhaled carbon monoxide (eCO)-confirmed smoking abstinence and its impact on HIV-specific outcomes. This Sequential Multiple Assignment Randomized Trial (SMART)⁴⁴ will be conducted within three health systems in the northeastern United States and include a complementary implementation-focused process evaluation to assess feasibility, cost, and future implementation of the intervention. The goals of this paper are to describe the rationale, aims, and study design of the SMARTTT trial.

2. METHODS

2.1. Overall design

Funded by the National Cancer Institute (NCI) as part of a dedicated initiative to improve smoking cessation interventions among PWH,⁴⁵ SMARTTT is a two-stage multi-site SMART enrolling PWH who smoke tobacco and are receiving care at one of the participating HIV clinic sites (Figure 1). Potential participants are screened to determine whether they meet entry criteria for study participation and, if eligible, provide written informed consent. Enrolled participants are randomized 1:1 to receive 12 weeks of nicotine replacement therapy (NRT) with or without prize CM (i.e., NRT vs. NRT+CM) to promote smoking abstinence. Consistent with a SMART approach, responders to first stage assignment (defined as exhaled carbon monoxide (eCO)-confirmed 7-day point-prevalence smoking abstinence at week 12), continue treatment for another 12 weeks. Non-responders, are re-randomized to either switch medications from NRT to varenicline (VAR), or intensify treatment to a higher CM reward schedule (i.e., VAR vs. CM+). Interventions are delivered by clinical pharmacists embedded in HIV clinics. The overall goal of this study is to identify the optimal adaptive treatment strategy to promote smoking abstinence in PWH who smoke cigarettes. Specifically, the goals are to identify the most effective initial treatment between NRT vs. NRT+CM (measured at week 12) and then, among individuals with initial non-response, identify the most effective treatment between switch to VAR and intensify CM (measured at week 24). The primary outcome is self-reported past 7-day smoking abstinence⁴⁶ confirmed by eCO.^47–49 Secondary outcomes include CD4 cell count, HIV viral load suppression, and VACS Index 2.0 scores. Consistent with a hybrid type 1 effectiveness-implementation design⁵⁰ and informed by existing literature,⁵¹ we will conduct an implementation-focused process evaluation in parallel.

Figure 1. SMARTTT Trial Protocol Overview.

Notes: PWH=patient with HIV; NRT=nicotine replacement therapy; CM=contingency management, CM+=enhanced contingency management

2.2. Rationale for study design

Our study approach is guided by three principles. First, given that treatment strategies that are responsive to patients’ needs and outcomes are likely to be more effective than static pathways for addressing tobacco use, we are using a SMART design.^44,52–54 This approach allows tailoring of the intervention based on treatment response using a pre-specified set of decision rules, continuation of effective treatments among those with response, and alternative options for those who do not respond, to ultimately identify the overall optimal treatment strategy. Second, we are relying on intervention components, including first-line tobacco treatment medications and a behavioral intervention, each of which has a strong evidence base to support their use in the general population^55–58 and, as relevant, safety data for use among PWH,^14–16 but they have not previously been evaluated as a package. Specifically, our goal is to generate data demonstrating the impact of switching from NRT to VAR among PWH. In addition, we aim to evaluate the impact of combining first-line tobacco treatment medications with CM in HIV clinical settings. Third, our intention is to simultaneously develop and evaluate an intervention package that will likely be implementable in HIV clinics. For this reason, and because HIV clinicians may not feel that they have adequate time or clinical skills to address smoking,^59–61 we designed our approach to involve clinical pharmacists as they are increasingly delivering smoking cessation treatments in clinical and community-based settings (e.g., pharmacies)^18–20,62 and recognized to be an important part of the HIV clinical care team.⁶³ In addition, consistent with a hybrid type 1 effectiveness-implementation design,⁵⁰ we will incorporate a process evaluation grounded in the Reach, Effectiveness, Adoption, Implementation, and Maintenance (RE-AIM)^64–66 and Promoting Action on Research Implementation Science in Health Services (PARiHS)^67–69 implementation science frameworks, to directly inform future implementation efforts. This is particularly relevant given our inclusion of CM and ongoing implementation challenges (e.g., concerns about external reinforcement and economics).⁷⁰

2.3. Study aims

The primary aim of this study is to identify the optimal adaptive approach to promote 7-day eCO-confirmed smoking abstinence. The primary hypothesis is that compared to NRT alone, NRT+CM will be associated with higher rates of abstinence at week 12. The study sample was chosen based on this hypothesis (see section 2.10.1). The secondary hypothesis is that non-responders to initial treatment who switch to VAR will have higher rates of abstinence at week 24, compared to non-responders that intensify with CM. The third hypothesis is that the optimal treatment strategy to promote abstinence at week 24 Is likely to be: begin with NRT+CM; if response, continue NRT+CM; if non-response, provide VAR+CM. Our hypotheses are informed by our experiences and those of others that PWH may prefer NRT over another medication given their high pill burden⁷¹ and evidence of CM as an effective behavioral treatment for smoking cessation.⁵⁸ To our knowledge, however, dual NRT has not been compared to VAR and CM has not been compared to VAR previously to promote smoking abstinence among PWH.

The second aim is to identify the impact of various tobacco treatment regimens on HIV-related biomarkers over time, including promoting improvements in: 1) CD4 cell count; 2) HIV viral load suppression; and 3) VACS Index 2.0 scores. The hypothesis is that smoking abstinence will lead to improvements in HIV-related biomarkers over time.

The third aim is to conduct an implementation-focused evaluation of the pharmacist-delivered intervention among key stakeholders, including patient participants, clinicians, staff, and leadership at each site.

2.4. Study context, coordinating center and institutional review

The study is being conducted in the context of Yale’s Center for Interdisciplinary Research on AIDS (CIRA)-supported New England HIV Implementation Science Network, whose mission includes stimulating and supporting research and evaluating collaborations across New England and neighboring regions, and promoting implementation science. The coordinating center is located at Yale School of Medicine, in New Haven, CT, and the Yale Center for Analytical Sciences (YCAS) coordinates data management and statistical support. The three participating health systems include: Yale New Haven Hospital (YNHH)’s Nathan Smith Clinic and Haelen Center, New Haven, CT; SUNY Downstate’s STAR Health Center, Brooklyn, NY; and Mount Sinai’s Institute for Advanced Medicine’s Jack Martin Clinic and Morningside Clinic, New York, NY.

2.5. Randomized trial participants

To be eligible for the randomized trial, patients must meet the following inclusion criteria: 1. be living with HIV [confirmed by electronic medical record review], 2. receive care at one of the participating sites, 3. be ≥18 years old, 4. have smoked ≥100 cigarettes over their lifetime, 5. smoke cigarettes every day or some days, 6. smoke, on average, ≥5 cigarettes per day based on the item “How many cigarettes a day to you smoke?”, and 7. provide written informed consent. Patients are excluded if they meet any of the following criteria: 1. exclusively use non-cigarette tobacco or nicotine products, including e-cigarettes, 2. currently use NRT or VAR (defined as any use of medication in the past 7 days); 3. self-report or urine testing confirming pregnancy, nursing, or trying to conceive, 4. life-threatening or unstable medical, surgical, or psychiatric condition, 5. inability to provide at least one collateral contact for a friend or family member, 6. living out-of-state, or 7. inability to read or understand English or Spanish.

2.6. Recruitment and randomization

Potential participants will be identified using a multi-pronged approach, including proactive screening via electronic medical review of patients scheduled for routine clinical visits, self-referral via recruitment flyers, and clinician referral. Potential participants will be screened to determine whether they meet criteria for current tobacco use (see section 2.5).⁷² Individuals who are potentially eligible and express interest in study participation will undergo written informed consent and then be invited to complete baseline study assessments and undergo randomization.

In the first stage, participants are randomized in a 1:1 ratio to NRT vs. NRT+CM where randomization is stratified based on site and the Heaviness of Smoking Index, a 2-question instrument that assesses the number of cigarettes smoked daily and time to first cigarette.⁷³ Heavy tobacco use is defined as smoking more than 10 cigarettes per day or having the first cigarette within 30 minutes from waking up. In the second stage, participants who do not demonstrate treatment response, are re-randomized to either switch medications from NRT to VAR or intensify CM with a higher reward schedule (i.e., CM or CM+); re-randomization is stratified based on site and first stage treatment. Permuted block randomization sequences are implemented in REDCap data management, a secure, customizable clinical trials management system used for randomization as well as data storage and management. The randomization sequences are concealed in REDCap.

Randomized participants receive $50 equivalent (i.e., reloadable debit card, cash, Target gift card depending on the site) for completing baseline assessments, and an additional $50 gift card for completing the follow-up assessments at week 12 and week 24. The reloadable debit cards are mailed to the participants home within a few days of the baseline visit completion and subsequent funds are uploaded within 48 hours. Participants who earn rewards from CM are promptly reimbursed (upon visit completion with Target gift cards or funds on reloadable debit card). These assessments are collected via interview by the research coordinator, optimally occurring face-to-face in private clinic or research space (depending on the site and with added flexibility due to COVID-19, see Table 4). Per protocol, assessments should be collected within 2 weeks of their scheduled due date; assessments collected outside of these windows are considered unscheduled and outside of window (with greater flexibility allowed for labs, see Section 2.9). Upon enrollment, study enrollment is documented with a study note in the electronic medical record and serves as mechanism to inform clinicians of their patients’ participation in the study.

Table 4.

Protocol Modifications due to COVID-19

	Original protocol	Proposed COVID-19 modifications
Consent	• In-person	• In-person or telephone-based
Assessments	• In-person; • Clinical pharmacist collects eCO if relevant for a CM visit; otherwise research coordinator collects eCO if assessment visit	• In-person, using coVita OneBreath™ piece; • Participant is remote with videoconference and Covita iCO smokerylzers; • Participant in clinic using videotechnology to have visit with research coordinator located remotely; clinic staff facilitate eCO collection
Intervention visits	• In clinic with clinical pharmacists	• In-clinic with study physician or research coordinator; • Participant in clinic using videotechnology to have visit with clinical pharmacist remotely • Participant on videoconference and remote
Contingency management	• Fishbowl with participant draws	• In person-participant or research team member draws with gloves • Participant on videoconference and watches clinical pharmacist draw from fishbowl
Gift cards and CM rewards	• Gift cards given directly and immediately after rewards earned	• Mail gift cards or reloadable debit card

2.7. Intervention overview

Consistent with principles of SMART designs,⁴⁴ participants’ response to initial treatment will be reassessed at an a priori defined time point; re-randomization will occur based on treatment response. During the first stage of treatment, intervention components include NRT and NRT+CM. At week 12, all participants will be assessed for smoking abstinence (i.e., the primary tailoring variable as defined based on self-reported past 7-day abstinence and confirmed by eCO≤6ppm) as this is a standard approach to defining abstinence.⁴⁹ Participants who have evidence of response, will continue treatment; those with lack of response will be re-randomized to switch to VAR or intensify CM (see below, Section 2.7.4). All intervention components are designed to be delivered by clinical pharmacists embedded in the HIV clinics, as such the visits with the clinical pharmacist occur in a routine private exam room. All participants are invited to return for a total of 10 visits over the 24-week period. For all visits, recommended brief counseling to address tobacco use (e.g., patterns of tobacco use, stressors and triggers of tobacco use, coping strategies and benefits of cutting down) is incorporated into the visit. These visits are guided by structured visit forms that were developed in collaboration with the clinical pharmacists to ensure they are consistent with clinical practice. Medications are obtained per usual care at pharmacies with payment coverage based on the participant’s health insurance.

2.7.1. NRT

In the initial stage of the study, all participants are counseled on and offered a prescription for 12 weeks of long-acting NRT with patches (84 count) and short-acting NRT along with brief advice. Following clinical practice guidelines, NRT dosing is tailored to participants’ cigarette consumption such that participants are prescribed a 14mg patch if they smoke 5–9 cigarettes per day or 21mg patch if they smoke 10 or more cigarettes per day. If they smoke the first cigarette after 30 minutes of waking, they are prescribed 2mg gum/lozenge, or if within 30 minutes of waking, they are prescribed 4mg gum or lozenge. Instead of gum or lozenges, participants may receive alternative short-acting NRT based on preferences and insurance coverage (nasal spray or inhaler). Combination long and short-acting NRT (“dual” NRT) is used because of its generally greater efficacy in promoting smoking abstinence.⁷⁴ In addition, the intervention starts with NRT instead of VAR due to the relative accessibility of NRT⁷⁵ and concerns regarding polypharmacy and pill burden, which are of particular relevance among PWH.^76,77 Also, prior studies have reported greater acceptability of NRT compared to oral medications (e.g., VAR) among PWH.⁷¹ During this first stage of treatment, participants will be invited to return for five follow-up visits with the clinical pharmacist, occurring approximately every week for the first two weeks of treatment and then every three weeks to address medication adherence, side effects, and progress towards smoking abstinence. Participants who respond to NRT at the end of week 12 will receive an additional 12 weeks of NRT with dose adjustment as indicated.⁷⁸

2.7.2. VAR

VAR, a partial agonist at the nicotine receptor, is a Food and Drug Administration (FDA)-approved medication for tobacco use disorder. It is the most effective form of monotherapy for tobacco use disorder treatment and has been demonstrated to be safe and effective in PWH, but has not yet been evaluated for use after patients do not respond to NRT.^14–16 Participants who do not respond to NRT or NRT+CM in the initial randomization may be randomized in stage 2 to switch to VAR, during which participants are counseled on and prescribed VAR. VAR dosing will be titrated upward, as is customary, from 0.5mg daily to 1mg twice daily by the end of week 1. Participants will be invited to return for visits with the clinical pharmacist, occurring approximately every week for the first two weeks of treatment and then every three weeks to address medication adherence, side effects, and progress towards smoking abstinence.

2.7.3. CM and CM+

Participants may be randomized in stage 1 to receive NRT+CM; in stage 2, those who do not respond to initial treatment, may be re-randomized to intensify CM (i.e., NRT+CM or NRT+CM+, respectively). Rooted in behavioral economics and with data to support its use to improve outcomes among PWH and address tobacco use in the general population,^58,79–82 CM involves the use of tangible, immediate rewards to promote verifiable behavior change.^70,83 In contrast to many other prior CM-based interventions to address tobacco use where monitoring occurs on a weekly or even daily basis in settings such as opioid treatment programs or the workplace,⁵⁸ CM in this study is delivered in HIV clinics. As such, the visit schedule is designed to minimize patient demands and to be consistent with other clinic-based practices, where weekly monitoring and assessment reflects high-intensity monitoring and visits are coupled with the medication-focused visits with the clinical pharmacist. Rewards will be obtained through “draws” of paper slips from a fishbowl, with slips replaced into the fishbowl after each individual draw. The slips in the fishbowl are set up in accordance with preset probabilities of rewards such that the average price per prize is $22.30 (Table 1); the rewards are provided as gift cards to local stores or using a reloadable debit card, based on site-specific preferences.⁸³ The prize-per-award magnitude was specifically designed to be higher than most other studies to account for less frequent monitoring.^84,85 Participants will earn draws if they report past 7-day smoking abstinence, confirmed with an eCO≤6ppm.^48,49 During CM, participants earn one draw at the first visit they achieve this goal, and draws earned escalate by one draw for evidence of continued abstinence to a maximum of 5 draws per session. During CM+, potential draws start at five with a maximum of 8 draws per session upon continued abstinence. If participants do not achieve abstinence, the reward schedule resets.

Table 1.

Fishbowl Rewards for Contingency Management.

Fishbowl Rewards
Reward	Total Slips	Value	Probability of Draw	Average expected price/prize
Small	20	$5	0.20	$1
Medium	64	$20	0.64	$12.8
Large	15	$50	0.15	$7.5
Jumbo	1	$100	0.01	$1
Total	100		1	$22.30

2.7.4. Determining treatment response: tailoring variable

At 12 weeks, response to stage 1 of treatment is determined using a combination of self-reported past 7-day tobacco use and eCO monitoring to verify abstinence. Participants are considered to have responded to treatment if they report no past 7-day tobacco use and have an eCO ≤6ppm and then continue in their initial treatment assignment. Participants are considered to have not responded to treatment if they: 1) report past 7-day tobacco use; 2) report no past 7-day tobacco use, but do not have an eCO ≤6ppm; or 3) do not present for week 12 visit within two week window; these participants are re-randomized to switch medications or intensify CM.

2.8. Intervention training and monitoring

Clinical pharmacists embedded in the participating HIV clinics conduct the intervention. Their training includes an orientation to: the overall study and procedures; review of the adverse impact of tobacco use on health among PWH; appropriate clinical use of NRT and VAR; eCO monitoring; and principles of CM. Prior to enrollment, all members of the study team from the participating sites participated in a full-day, in-person training that included didactics and role play. This training was videotaped to facilitate training of new individuals who join the study team at a future date. The clinical pharmacists were assessed for competence in study procedures using a quiz, which focused on CM procedures. Then to ensure ongoing adherence to the study protocol, structured visit forms are used guide intervention visits and the clinical pharmacists participate in monthly videoconferences and receiving ongoing supervision and feedback. This feedback is based on review of digital recordings of sessions and visit tracking forms, both of which are rated based on an adapted version of the Contingency Management Competence Scale⁸⁶ (Appendix 1).

2.9. Data collection protocol

Assessments are conducted by trained research coordinators at baseline, week 12, and week 24 with participant interview and objective markers; at month 12, laboratory data is additionally collected. The goal of these assessments is to evaluate baseline characteristics to ensure patients meet eligibility criteria, and important factors that may impact the effect of the intervention; and to measure process outcomes and study endpoints (Table 2). The primary outcome is smoking abstinence based on a negative response to “Have you smoked even a puff in the last 7 days?”⁴⁶ confirmed with eCO ≤6ppm.^47,48 Secondary smoking-related outcomes include: quit attempts, daily cigarette consumption, and nicotine dependence using the Heaviness of Smoking Index (HSI).^73,87 Secondary HIV-related outcomes include: change in CD4 cell count, proportion with an undetectable HIV viral load (defined as <50 copies/mL), and change in VACS Index 2.0 scores. The VACS Index 2.0 score is a validated composite biomarker that integrates routinely-available, laboratory-based test results to predict morbidity and mortality risk based on measures of organ system injury.^43,88 Previously validated in individuals with and without HIV infection, the VACS Index score includes age, HIV biomarkers (CD4 cell count, HIV viral load), hemoglobin, FIB-4 (a non-invasive measure of liver fibrosis calculated based on aspartate aminotransferase, alanine aminotransferase, and platelets), estimated glomerular filtration rate (determined with serum creatinine), white blood cell count, albumin, and presence of hepatitis C virus (HCV) infection. The VACS Index score is responsive to changes in antiretroviral medication³⁹ and varying levels of alcohol and opioid use,^41,88,89 with higher scores reflecting a greater risk of morbidity and mortality. Baseline labs are collected 60 days prior to assessment due date, or up to 30 days after to be considered within the preferred window period; follow-up labs are collected 30 days before or after their scheduled due date to be considered within the preferred window period. Labs collected outside this window are noted as such and the date noted.

Table 2.

Assessments, Outcome, and Process Measures

Assessments, Outcome, and Process Measures
Instrument	Baseline	Week 12	Week 24	Month 12
Baseline measures and key covariates
Intake History/Patient Demographics	X
Self-reported comorbidities107	X
AUDIT-C108	X	X	X
ASSIST Lite109	X	X	X
Marijuana Assessment	X	X	X
Patient Health Questionnaire (PHQ)-8110	X	X
PROPr91	X	X	X
Medication Adherence Questionnaire111	X	X	X
Short UPPS-P112	X
Medication nonadherence due to medication costs113	X
Tobacco-related measures
Exhaled carbon monoxide level48,49	X	Xa	Xa
Wisconsin Predicting Patients’ Relapse Questionnaire114	X	X	X
Heaviness of Smoking Index73,87	X	X	X
Self-report of past 7 day tobacco use95	X	X	X
Tobacco and e-cigarette use	X	X	X
Readiness to change smoking and quit status	X
Quit attempts		X	X
HIV-related measures
CD4 count	X		X	X
HIV viral load	X		X	X
VACS Index 2.043	X		X	X
Symptoms
HIV Symptom Index90,115,116	X	X	X
Process Measures
adapted Treatment Service Review (TSR),117 including self-reported days of NRT, VAR use	X	X	X
CM session attendance, rewards earned.51,84	during intervention visits b
Implementation related Measures
Cost related data	X	X	X
Patient Satisfaction Survey.51,93		X	X

Notes:

^a.among those who report past 7 day smoking abstinence only

^b.among those receiving CM only

Additional assessments are being collected to evaluate baseline characteristics of study participants that may impact the effectiveness of the intervention on outcomes (e.g., depressive symptoms, alcohol use) and allow for exploratory subgroup analyses. We will assess bothersome symptoms using the HIV Symptom Index⁹⁰ and catalogue any study related and non-study related adverse events. We will assess use of tobacco use disorder treatments, including medications (i.e., NRT, VAR and bupropion); participation in smoking cessation support groups; use of texting programs and other Quitline services; and use of electronic cigarettes. To inform subsequent cost-effectiveness analyses, we will assess the costs of implementing and sustaining each intervention strategy, health related quality of life using the PROMIS-Preference score (PROPr),⁹¹ treatment services with an adapted version of the Treatment Services Review,⁹² and rewards earned from CM. A participant satisfaction survey will be used to assess participant perceptions of all aspects of the intervention.^51,93

2.10. Statistical considerations

2.10.1. Justification of sample size

The primary aims are to identify the most efficacious initial treatment (Aim 1, Hypothesis 1), confirmed with biochemical measurement of eCO at 12 weeks, and the most efficacious treatment (switch vs intensify) for non-responders (i.e. participants who continue to smoke) to the initial randomization (Aim 1, Hypothesis 2), confirmed with measurement of eCO at week 24. For the effect sizes of the treatments, VAR, combination NRT, and CM, we used estimates taken from a Cochrane meta-analysis of tobacco dependence treatment in PLWH, focusing on data pooled from 11 clinical trials involving a total of 1785 participants. In these studies, tobacco dependence treatment interventions (behavioral, pharmacologic, or both) had a Mantel-Haenszel risk ratio of abstinence at up to 6 months, relative to controls, of 1.51 (95%CI 1.15, 2.00).⁹⁴ An additional Cochrane meta-analysis of incentive payments for smoking cessation in general populations of adult smokers, pooling data from 17 trials involving 7,715 participants, found an odds ratio for abstinence at 6 months or more of 1.42 (95%CI 1.19, 1.69), relative to controls.⁵⁸ Using estimates from these studies, we assume a 12 week biochemically confirmed abstinence rate of 16% for the first-stage treatment of combination NRT+CM, and 7% for NRT alone. For the second stage, in which non-responders from both NRT and NRT+CM arms are randomized to 12 weeks of either switching medication or intensifying CM, we assume an end-of-treatment biochemically confirmed abstinence rate of 16% for switching, and 7% for intensifying. Given these parameters and a 2-sided type I error of 0.025 (corrected for 2 primary hypotheses), a total sample size of 538 participants will provide 85% power to detect superiority of NRT+CM over NRT alone for initial treatment. To account for a 15% loss to attrition, we will enroll a total of 632 participants. Given the expected proportions of non-response to initial treatment (i.e. 88.5% total; 84% and 93% in NRT+CM and NRT alone, respectively), the sample size will provide 80% power to detect a difference of 9% (i.e. 16% in switch, 7% in intensify). The sample size calculation was performed with PASS v12 (Kaysville, UT).

2.10.2. Statistical analyses

2.10.2.1. Primary outcomes.

The primary efficacy endpoint for this study will be biochemically verified 7-day abstinence at 12 weeks (Hypothesis 1) and 24 weeks (Hypothesis 2).⁹⁵ Tobacco use will be assessed by self-report and confirmatory biochemical testing with exhaled carbon monoxide.

Analyses will be conducted using intention to treat principles,⁴⁷ such that all randomized participants will be included in the denominator for calculating abstinence rates with the exception of unavoidable loss to follow-up (i.e., participants who died).⁴⁷ We will report the number of participants in these categories who are excluded separately for each condition.⁴⁷ Multiple imputation will be used to account for missing data in participants lost to follow-up.

For our first primary hypothesis (Aim 1, Hypothesis 1), we will evaluate the impact of the initial treatment strategy on 12-week abstinence. A logistic regression will be used including initial treatment condition (NRT vs. NRT+CM) along with covariates for site, age, gender and HSI. Odds ratios, predicted probabilities of 12 week abstinence, and 97.5% confidence intervals will be estimated. Baseline characteristics and abstinence outcomes assessed at other timepoints will be used in the missing data model. The difference between probabilities of 12-week abstinence between NRT and NRT+CM will be evaluated at the 0.025 two-sided significance level (i.e. Bonferroni-corrected for the 2 primary hypotheses).

A similar analysis will used for our other primary hypothesis (Aim 1, Hypothesis 2), to evaluate the impact of the second randomization (i.e., switch vs. intensify) among participants with non-response. This analysis will only include participants for whom the initial treatment failed to produce abstinence. Logistic regression will include second stage treatment as well as covariates for site, age, gender, HSI, reason for 12-week non-response (i.e., non-abstainer vs. loss to follow-up) and initial treatment group. Odds ratios, predicted probabilities of 24-week abstinence and 97.5% confidence intervals will be estimated. Baseline characteristics will be used in the missing data model. The difference between probabilities of 24 week abstinence between switching and intensifying will be evaluated at the 0.025 two-sided significance level (i.e. Bonferroni corrected for the 2 primary hypotheses).

2.10.2.2. Secondary outcomes.

For secondary outcomes of CD4 cell count, HIV viral load and VACS Index 2.0, evaluation of initial treatment strategies at 12 and 24 weeks will be performed using linear mixed models. These analyses will include fixed effects for initial treatment, time, and their interaction; covariates for baseline outcome, site, age, and gender; and a random effect for subject repeated measures. Similar linear mixed models of only those for whom the treatment fails to produce a response will be used to evaluate second stage treatment. Least squares means and 95% confidence intervals will be estimated.

2.10.2.3. Exploratory outcomes.

We will determine the overall best strategy with regard to 24-week abstinence. There are four strategies to compare: 1) NRT followed by VAR in non-responders; 2) NRT followed by NRT+CM in non-responders; 3) NRT+CM followed by VAR+CM in non-responders; 4) NRT+CM followed by NRT+CM+ in non-responders. Choosing the optimal strategy is an estimation rather than hypothesis testing problem.⁹⁶ Weighted logistic regression will be used to estimate 24-week abstinence rates for each of the strategies. Inverse probability weighting will account for non-responders being re-randomized and thus split in two groups, as they will be underrepresented relative to responders. A random effect will be included for repeated measures and the analysis will include indicators for first and second randomization, time, the interactions of first/second randomization indicators with time, as well as covariates for baseline outcome, site, age, gender, and HSI. Predicted probabilities and 95% confidence intervals will be estimated for each of the strategies. The strategy with the highest predicted probability will be identified as the optimal strategy. We will additionally examine “dose response effects” and examine the impact of number of completed intervention sessions as measures of the intensity of intervention on outcomes, including number of completed visits.¹² Similar analyses will be conducted for secondary outcomes.

To examine whether treatment differences vary among subgroups based on demographic (e.g., age, gender, race/ethnicity) and clinical characteristics (e.g., depressive symptoms, alcohol use), we will conduct exploratory and hypothesis-generating analyses.

2.11. Implementation-focused process evaluation

In parallel with testing the impact of our designed intervention on promoting smoking abstinence and improving HIV-related outcomes, we will gather information on its delivery to inform implementation in real-world settings.⁵⁰ We will model our approach based on prior work⁵¹ and grounded it in RE-AIM^64–66 and PARiHS.^67,68,97 Data sources for this evaluation will be primarily based upon experiences with potential and recruited trial participants (Table 3). In addition, we will conduct an online REDCap-based survey upon trial initiation (Appendix 2) and post-trial completion to examine perspectives on the intervention package and its components among clinicians, staff and clinical leadership at each of the participating sites. Items included in this survey were drawn from the existing literature^98–100 and based upon validated tools as possible^101,102 and then piloted and refined as needed by the investigative team. Number (percent) will be used to present data from categorical variables. Continuous data will be presented as means (+/− standard deviations); nonnormally distributed data will be characterized using medians and IQRs. Differences pre-post trial responses will be assessed using appropriate parametric or non-parametric methods. To assess costs associated with conducting the intervention, we will track clinical pharmacist time for training and then delivering the intervention, costs of eCO monitoring, and rewards earned per CM session. We will assess cost variation based on source of the intervention (e.g., physician, advanced practice practitioner, social worker).^51,84 In addition, we will track costs from the perspective of the participants as well as changes in health service utilization and quality of life to allow future cost effectiveness analyses.

Table 3.

Overview of Implementation-Focused Process Evaluation^a

Implementation Framework Elements to Guide Process Evaluation Questions and Tools
Element	Questions	Data Sources	Tools
RE-AIM
Reach	What % of patients approached agree to participate in the intervention?	Recruitment rates	Patient screening database
	Do those that agree to participate differ systematically from those that do not?	Demographics of those agreeing vs. declining participation
	Patients’ likes/dislikes about intervention?	Perceptions of participants	Patient satisfaction survey
Effectiveness	Effect of intervention on patient outcomes?	Main study outcomes	Abstinence rates
			Study retention Fidelity monitoring (e.g., CM protocol adherence; medication receipt)
Adoption	Barriers to adoption?	Perceptions of clinical pharmacists, clinicians, staff and clinical leadership	Survey with clinicians, staff and clinical leadershipb
	Supports needed for clinic to adopt intervention?
	How do perceptions change pre-/post-trial?
Implementation	Is the intervention delivered as intended?	Process outcomes	Study retention Fidelity monitoring (e.g., CM protocol adherence; medication receipt)
	What supports need to be in place to ensure consistent delivery of intervention?	Perceptions of clinicians, staff, clinical leadership	Survey with clinicians, staff, clinical leadershipb; notes from standing weekly and monthly calls (e.g., medication access)
	Tools needed for consistent intervention delivery?	Perceptions of clinical pharmacists and research assistants
	What does the intervention cost?	Cost data	Records of incentives awarded, costs of supplies, staff time
Maintenance	Resources needed to maintain intervention?	Perceptions of clinical pharmacists, clinical leadership	Post-trial survey
	What adaptations are needed to integrate intervention into regular practice?
PARiHS
Evidence	What are clinician and staff perceptions of the evidence supporting the intervention?	Perceptions of clinical pharmacists, clinicians, staff and clinical leadership	Survey with clinicians, staff, clinical leadershipb
	What are attitudes toward intervention?
	Does intervention fit in current practice?
	Does it meet perceived need of patients?
Context	What are characteristics of culture in clinic?	Perceptions of clinical pharmacists, clinicians, staff and clinical leadership	Survey with clinicians, staff, clinical leadershipc
	Characteristics of leadership of clinic?
	What resources are available to clinic?

Notes.

^a.Table adapted from Hagedorn et. al. Addiction Science and Clinical Practice⁵¹ and published with permission from the publisher.

^b.survey assesses experiences and perspectives on addressing tobacco use and treatment options^{98,99,118,119} and includes an adapted version of the Contingency Management Beliefs Questionnaire [CMBQ]¹⁰¹)

^c.survey includes Organizational readiness for implementing change[ORIC]¹⁰²

2.12. Protection of participants

The Institutional Review Board (IRB) at Icahn School of Medicine at Mount Sinai has served as the single IRB for this protocol and approved this study for all sites and per reliance agreements with Yale School of Medicine and SUNY Downstate. The study is HIPAA compliant. A Data Safety and Monitoring Board (DSMB) will review study progress every 6 months starting 6 months after enrollment initiation to review enrollment, baseline characteristics of enrolled participants, delivery of the intervention, outcome ascertainment, and adverse events.

2.13. Current status of SMARTTT

In preparation for study launch, all research team members participated in a full day, in-person training on March 3, 2020 in New Haven, CT. Due to the COVID-19 pandemic, study start was delayed until July 27, 2020, when the first participant was enrolled at Mount Sinai; the first enrollment occurred on October 2, 2020 at Yale-New Haven Hospital and on November 3, 2020 at STAR. As of January 7, 2020, 27 participants total have now been enrolled in the study. To enhance safety of study participants and all individuals involved (i.e., research team and clinical team members), we have incorporated modifications into study procedures with intention to decrease risk of COVID-19 disease transmission, optimize social distancing, and maintain flexibility given uncertainty associated with the circumstances and variable access to technology among potential study participants (Table 4). In addition, the web-based survey of clinicians, staff and clinical leadership was completed from November, 4, 2020 through December 15, 2020; analyses are underway.

3. DISCUSSION

SMARTTT will generate timely and needed data on a new approach for promoting smoking abstinence among PWH and its potential for implementation in HIV clinical settings. The methods are innovative for several reasons. First, this is the first SMART that includes first-line tobacco treatment medications for PWH, and examines these medications in combination with CM, and thus will allow for the generation of needed data on the optimal tobacco treatment algorithm incorporating these evidence-based interventions.^{17,33,103,104} In addition, it will extend the literature on whether CM can effectively be delivered in clinical settings with less intense monitoring to promote smoking abstinence. Second, it will be the first evaluation of a pharmacist-led intervention to address tobacco use in HIV clinical settings. This is relevant giving the growing evidence demonstrating the importance of integrating HIV and substance use disorder-related care.^31,32,105 Third, by including the VACS Index 2.0 as an outcome, this will be among the first studies to prospectively examine the impact of a tobacco treatment intervention on a comprehensive validated laboratory-based measure of overall disease severity. Fourth, with the hybrid type 1 effectiveness-implementation design, we allow study findings to directly inform future implementation efforts. Lastly, given the unexpected challenges associated with COVID-19, we are iteratively refining protocols to optimize safety along with study rigor and integrity. These experiences may generate important insights regarding implementation of remote CM to address tobacco use among PWH, a population for whom technology is generally less accessible.¹⁰⁶ The findings from SMARTTT will have important implications for informing whether CM should be added to NRT to promote smoking abstinence; for determining whether it is more effective to switch to VAR or intensify CM for those who are unable to achieve timely smoking abstinence; and whether and how to enhance implementation.

3.1. Limitations

There are several expected limitations to our study. First, the COVID-19 pandemic has resulted in dramatic transformation in healthcare delivery with ongoing uncertainty in how the pandemic and its impact will evolve to threaten recruitment efforts and adherence to the protocol. Second, if CM does not show the expected effect, it will not be clear if CM is not effective in this context or this is a function of the lower frequency of CM visits that was chosen to be more consistent with other clinic-based practices. Third, in addition to smoking tobacco, cannabis use as well as environmental exposures may cause elevated eCO levels. Participants are advised to avoid combustible cannabis for the 24 hour period prior to any visits during which eCO may be measured and we assess cannabis use by self-report during assessment visits. Although eCO levels may be elevated for these other reasons, we have no reason to believe that the risk for an elevated eCO for other reasons should differ by treatment group.. Fourth, this is a non-blinded study and, due to resource constraints, research coordinators who are performing data collection will be aware of participant study assignment. In addition, as this is an effectiveness study, participants and pharmacists will be aware of the medications being prescribed. However, our randomization procedures include allocation concealment and our outcomes are based on self-report in conjunction with objective, biomarker-based testing of tobacco use. Third, our intervention may not be scalable to all settings, such as those that lack clinical pharmacists. Our intervention, however, will be manualized and could be performed either by another prescriber type (e.g., physician assistant) or by a prescriber with a non-prescriber (e.g., social worker); our planned cost analyses will serve to inform alternative models for intervention delivery.

3.2. Conclusion

SMARTTT will generate data on the impact of a novel adaptive intervention, including medications and CM, to promote smoking abstinence among PWH. Findings generated from this study will be directly relevant for informing delivery of tobacco treatment interventions in HIV clinical settings, as well as relevance for primary care and community-based organizations serving PWH.

HIGHLIGHTS.

1. Existing strategies to address tobacco use among individuals with HIV are inadequate.

2. Medications and contingency management may help promote smoking abstinence.

3. Clinical pharmacists play a growing role in tobacco cessation interventions.

4. SMART designs are useful for identifying the optimal sequential adaptive treatment strategy.

5. Hybrid type 1 designs generate timely information to inform implementation efforts.