Integrating tobacco treatment into lung cancer screening practices: Study protocol for the Screen ASSIST randomized clinical trial

Abstract

Background:

Integrating tobacco treatment services into lung cancer screening (LCS) has the potential to leverage a ‘teachable moment’ to promote cessation among long-term smokers and reduce disparities in tobacco treatment access. This protocol paper describes the Screen ASSIST (Aiding Screening Support In Stopping Tobacco) trial, which will identify how to best deliver evidence-driven tobacco treatment in the context of LCS.

Methods:

Screen ASSIST is a randomized clinical trial with a 3-factor, fully crossed factorial design that enrolls current smokers (any cigarette use in the past 30 days) scheduled to attend LCS at multiple sites in the Mass General Brigham healthcare system. To maximize reach, recruitment is conducted at 3 time points: 1) at the time of LCS scheduling, 2) at the LCS visit, and 3) after the participant has received their LCS results. Participants are stratified by LCS study site and recruitment point and randomly assigned into 8 groups that test intervention components varying on telehealth counseling duration (4 weeks vs. 8 weeks), nicotine replacement therapy duration (2 weeks vs. 8 weeks), and systematic screening and referral for social determinants of health via a service named ‘AuntBertha’ (referral vs. no referral). The primary study outcome is self-reported past 7-day tobacco abstinence at 6-month follow-up. This trial will also assess systems integration and evaluate implementation of the intervention.

Discussion:

Screen ASSIST will identify the most effective combination of tobacco cessation treatments within the LCS context, in order to improve the cost-effectiveness of LCS and quality of life among long-term heavy smokers.

1. Introduction

Lung cancer remains the leading cause of cancer death in the United States [1]. While there has been a gradual decline in incidence and mortality over the past three decades, an estimated 228,820 new cases and 135,720 deaths were attributed to lung cancer in 2020 [2]. The prognosis for individuals with lung cancer is generally poor, largely because only 15% are diagnosed at an early stage [3]. Cigarette smoking is responsible for 87% of lung cancer deaths and is the leading cause of preventable death in the U.S. [4–7] Adherence to annual lung cancer screening (LCS) through low-dose computed tomography (LDCT) has demonstrated a 20% reduction in lung cancer mortality among current and former smokers [8]. Since 2013, the U.S. Preventive Services Task Force (USPSTF) has recommended annual LDCT screening for individuals aged 55–80 and with 30+ pack-years smoking history. New recommendations have expanded the age range (now 50 to 80) and reduced the pack-year history to 20 pack-years of smoking [9].

LCS has the potential to detect malignancies earlier but also provides a critical opportunity to promote smoking cessation [10]. Integrating smoking cessation into the LCS process can increase access to tobacco treatment and help decrease long-standing disparities, whereas not promoting cessation may send a message to individuals that screening obviates the need to quit [11]. Most smokers require personalized tobacco treatment programs tailored to their quit readiness and risk perception. For smokers undergoing LCS, the teachable moment literature suggests a particular need to have tobacco treatment appropriately tailored to a smoker’s worry about developing lung cancer. Many smokers undergoing LCS are not ready to make a quit attempt, with previous studies suggesting smokers undergoing LCS perceive a benefit from quitting but do not possess the confidence to quit, avoid thinking about risks, and use unhelpful avoidance strategies [12–15].

Over 40% of smokers undergoing LCS are heavy smokers (>20 cigarettes per day) and need more intensive support, but it is not clear what types of interventions would be most appealing and motivating for them [12]. Clinical guidelines state that combining counseling and pharmacotherapy (e.g., nicotine replacement therapy; NRT) is more effective than either alone [16,17]. Long-term heavy smokers also have social barriers to quitting (e.g., stressful living environment) that may interfere with cessation efforts and provision of community resources may aid cessation [18]. Additionally, integrating cessation services into LCS sites (typically high-volume practices that lack cessation resources) is a challenge. Most LCS sites report screening individuals’ smoking status (99%) and advising smokers to quit (91%), but fewer provide cessation counseling or treatment referral (60%), and only one-third recommend cessation medications [19]. Thus, Screen ASSIST aims to test the effectiveness of which combination of 3 evidence-driven tobacco cessation treatments are most beneficial for smokers undergoing LCS through a factorial design.

2. Design

2.1. Conceptual framework

The components of the LCS-tailored intervention are informed by our preliminary studies and conceptual frameworks. Screen ASSIST incorporates 4 well-established models (Fig. 1). The LCS-tailored counseling content incorporates concepts of the Health Belief Model (HBM) and Self-Regulation Model (SRM) [20–22]. The HBM details attitudes that underlie behavior change and is widely used to study smoking cessation. It posits that when faced with a health threat, individuals are more likely to change a behavior if they feel the threat is serious, they are at risk, they are confident in their ability to make the change, and that there are benefits to making change (i.e., decreasing disease risk). The SRM focuses on the dynamic interplay between beliefs, emotions, and coping in the context of a health threat. The SRM postulates that individuals form illness representations (e.g., what is it?) that guide their behavioral responses to an illness and then engage in strategies (e.g., what can I do that will make me feel better?) to reduce distress. Illness representations can be influenced by environmental (e.g., others are smoking in the home) and physical (e.g., shortness of breath) factors. Parallel processing between illness beliefs and emotions leads to a coping response (e.g., quitting) and subsequent monitoring of the success of coping efforts. Thus, it posits that changes in these factors may lead smokers to quit as a coping strategy for the threat of lung cancer.

Fig. 1.

Integration of proposed intervention components and theoretical models.

The systems integration and implementation evaluation of the intervention are informed by the Chronic Care Model and the RE-AIM framework [23–25]. The Chronic Care Model gives health systems a structure for organizing care of chronic diseases to improve outcomes. This model’s components inform our healthcare system-wide integration and implementation strategy. Our intervention has the support of health system leaders, integrates into the LCS care delivery system (i.e., the EHR ordering system and patient records) with the inclusion of primary care and radiology, uses information systems to provide timely information, includes decision support (videos of primary care and radiology clinicians and tobacco coach), and provides links to community services. The RE-AIM Model framework will evaluate the impact of the intervention in the LCS setting. (1) Reach (proportion of eligible patients and characteristics of enrollees); (2) Effectiveness (smoking outcomes); (3) Adoption (proportion of providers/sites who use the intervention); (4) Implementation (adherence to intervention, cost). The final phase of the project will focus on development of a Maintenance plan (extent to which the intervention becomes part of the LCS healthcare system settings).

3. Methods

3.1. Overview

The primary aim of Screen ASSIST is to test the effectiveness of which combination of evidence-driven tobacco cessation treatments are most beneficial achieving cessation among smokers undergoing LCS. To do so, this randomized controlled trial (N = 720) employs a factorial design to assess 3 intervention components that vary on: telehealth counseling duration (4 weeks vs. 8 weeks), nicotine therapy duration (2 weeks vs. 8 weeks), and systematic screening and referral for social determinants of health (referral vs. no referral). The primary study outcome is 7-day point prevalence of self-reported tobacco abstinence at 6 months. Secondary aims are to: a) develop a centralized systems-based smoking cessation intervention at study LCS screening sites; and b) evaluate the reach, adoption, implementation, and maintenance of the intervention.

3.2. Setting

Participants are currently recruited from 8 LCS screening sites in the Mass General Brigham (MGB) integrated health care delivery system in Massachusetts. These include 5 sites affiliated with Massachusetts General Hospital (MGH) (2 on the hospital campus and 3 at community sites in the surrounding cities of Waltham, Chelsea, and Danvers), as well as sites at Brigham and Women’s Faulkner Hospital, Newton-Wellesley Hospital, and Martha’s Vineyard Hospital. Further sites are being evaluated for study inclusion. Screen ASSIST began enrolling participants in April 2019.

3.3. Inclusion/Exclusion criteria

Participants who are scheduled to undergo LCS at a participating MGB-LCS site are eligible for study enrollment if they have any cigarette use in the last 30 days, speak English or Spanish, and fulfill the 2015 Medicare coverage criteria (age 55–80 years and 30+ pack/years). Age and pack-year eligibility criteria were expanded in March 2021 to reflect updated USPSTF recommendations. Participants are eligible if they are willing to discuss their smoking behavior; they are not required to be willing to quit at enrollment. Exclusion criteria include participants who are undergoing lung CT as part of a diagnostic or abnormal follow-up evaluation, unable to give informed consent due to significant psychiatric or cognitive impairment as determined via EHR chart review in consultation with study PI or treating clinician, lack of access to a telephone or cannot communicate by telephone, or do not reside within the United States of America.

3.4. Recruitment

To maximize reach of the study, individuals undergoing LCS at a participating study site are proactively offered study enrollment at 3 points in the LCS process. The study research assistants (RAs) recruit participants by telephone: (1) after the LCS has been ordered, (2) at LCS visit, and (3) after the result has been communicated to the participant. Our recruitment strategy is designed to require little additional effort at busy radiology sites, which do not typically have the resources to dedicate staff for in-person recruitment. Instead, our systematic, multipronged, and repeated recruitment efforts are designed to achieve maximal reach. All recruitment materials (consent, etc) are offered in Spanish. Materials were translated by a native Spanish speaking study staff member.

Recruitment Point 1:

LCS Scheduled – After a participant’s clinician orders a LCS through the electronic health record (EHR), the test is scheduled by Radiology. Study research assistants (RA) obtain a live feed of the newly scheduled LDCTs that displays scheduled tests up to 3 weeks in advance. The RA reviews the EHR to determine if the participant meets the study eligibility criteria and mails and/or emails eligible participants an opt-out letter with a study pamphlet and a detailed information sheet about the study. If participants have an email address recorded in their EHR, they also receive a link to a short video recruitment message, embedded in a REDCap survey and hosted on the MGB approved YouTube account. The short video recruitment message included two members of the trial team talking directly into the camera: a primary care physician, who is also a study investigator, and a tobacco treatment specialist, who provides cessation counseling in the trial. The video reinforces the importance of completing their LCS, the benefits of cessation, and informs the smoker about the goals of Screen ASSIST and their potential eligibility to participate.

Approximately 3 business days after the study recruitment materials have been sent, an RA calls potential participants to reinforce attendance at the upcoming LCS appointment, describes the study, screens interested participants for eligibility and obtains verbal informed consent using a standard script. Enrolled participants complete a baseline survey verbally or through a Research Electronic Data Capture (REDCap) link sent via email or SMS text message. A mailed survey can be sent to participants if they do not have an email address and prefer a hard copy of the survey. RAs ask participants who are uncertain about participating in the study for permission to re-contact them at subsequent recruitment timepoints. If they decline, the study team does not approach them at later recruitment points.

Recruitment Point 2:

LCS Exam – Participants who have been screened as chart eligible by study staff are given iPads by front desk staff upon check-in. iPads are offered to the participant as an opportunity to self-report their eligibility status through a short REDCap survey (e.g., current smoking status) and indicate interest in the study or opt out of the study. At the end of the survey, a short video of a study radiologist, tailored to the participant’s responses, is displayed. Participants who have already enrolled in the study are identified by the iPad and receive a video reinforcing the importance of continuing with the study. Participants who indicate they are willing to join the study are contacted by the study RAs for recruitment within approximately 3 days of their LCS exam. Consent and enrollment procedures are then conducted. If participants are uncertain, they can ask to be contacted at the next recruitment point. If they decline, the study team does not approach them at the next recruitment point.

Due to institutional safety measures to prevent the spread of COVID-19, this recruitment point was adapted on March 13, 2020 to remove participant contact with the study iPad. At specific study sites, study pamphlets are currently being given to all LCS patients to increase awareness of the study and maintain clinic presence. Current consent and enrollment procedures detailed are conducted within approximately 3 days after a participant completes the LCS exam.

Recruitment Point 3:

LCS Results – After the participant has received their LCS result, participants who have not previously enrolled or not already declined enrollment are sent a short video recruitment message in a secure REDCap survey after approximately 7 days from LCS completion date. The video, tailored to the participant’s LCS result, has a study radiologist invite them to join the study and offer brief cessation advice. Participants are then contacted by the study RAs to initiate the recruitment process. Participants who enroll at an earlier recruitment point receive a short video embedded in a secure REDCap survey tailored to their test result to address why quitting and study participation remain important. Consent and enrollment procedures are then conducted.

3.5. Assignment to treatment group

Participants are randomly assigned at the time of enrollment (i.e., completed consent and baseline survey), stratified into 9 randomization strata (3 groups of LCS study sites x 3 recruitment points), to a LCS-tailored intervention treatment arm. The intervention arms consist of 8 groups of equal proportions, based on a 3-factor, fully crossed factorial design to efficiently test components that vary on counseling duration, NRT duration, and provision of community referral (Table 1).

Table 1.

Intervention groups (3-factor fully crossed factorial design).

Condition	Counseling duration (4 vs. 8 sessions)	NRT duration (2 vs. 8 weeks)	Systematic screening for social determinants of health and community referral (Yes vs. No)
1	Shorter	Shorter	Yes
2	Shorter	Longer	Yes
3	Longer	Shorter	Yes
4	Longer	Longer	Yes
5	Shorter	Shorter	No
6	Shorter	Longer	No
7	Longer	Shorter	No
8	Longer	Longer	No

Note. Shorter durations of counseling and NRT are modelled after Quitline durations.

3.6. Tobacco treatment interventions

3.6.1. Counseling

The team developed an evidence-driven tobacco cessation counseling intervention for LCS in English and Spanish that built on their prior research, was grounded in theoretical models, and adapted prior tobacco cessation counseling and medication treatment protocols used for low SES smokers, cancer patients, and LCS patients. We ensured all patient-facing materials were developed at a 6th grade reading level (Flesch-Kincaid) for all protocols. Once the treatment protocols were finalized, the MGB translation services back translated all materials, to assure equivalency and cultural appropriateness.

Participants randomized to the short duration counseling conditions are offered 4 weekly sessions over 1 month. Those assigned to the long duration counseling are offered 4 additional biweekly sessions over months 2 and 3. The tobacco treatment specialist (TTS) promotes medication use, described below.

3.6.2. Counseling style and content

The TTS uses Motivational Interviewing (MI) as an empathic and supportive treatment style, which can deliver personalized risk counseling, enhance motivation for behavior change, and is effective for participants not ready to quit and medically and socio-economically vulnerable smokers. MI is well suited for LCS participants for several reasons: 1) MI focuses on building self-confidence and resolving ambivalence; 2) MI tools (i.e., open-ended questions, affirmations, reflections, and summarizing statements) are effective when addressing sensitive topics; and 3) MI’s strategy is effective when communicating about risk. Undergoing LCS might increase or decrease smokers’ risk perceptions, motivation to quit, and cessation efforts.

Table 2 gives an overview of the counseling content. Protocols for the counseling sessions follow the 5As format for smoking cessation. In addition, the counseling protocol incorporates participants’ risk perceptions before and after the LCS results. The risk counseling content emphasizes understanding the personal implications of risk beliefs and aims to reduce cognitive biases (e.g., protective benefits of a negative screen). The counseling protocol elicits 1) contrasts between personal and comparative risk, 2) reasons underlying perceived risk, and 3) beliefs about the connections between risk, smoking, and lung cancer. The TTS: 1) assesses the participant’s preferences for health information/numeracy; 2) explores the personal meaning of risk; 3) personalizes risk messaging based on patient preference for negative (e.g. costs of not quitting) and/or positive framing (e.g., benefits of quitting); 4) emphasizes personal magnitude of risk reduction; and 5) acknowledges ambivalence.

Table 2.

Overview of counseling content by session.

Sessiona	Factor	Counseling Content	Smoking Cessation Medication Content
1	Short	• Beliefs about lung screening • Review smoking history • Assess emotional distress	• NRT introduction and discuss 1st NRT dose
2	Short	• Lung screen visit experience • Reinforce perceived benefits • Nicotine cravings and withdrawal symptoms	• Reinforce medication benefits/asses side-effects • Discuss 2nd NRT dose for participants in longer NRT duration group
3	Short	• Discuss lung screen results • Social support • Coping with smoking triggers	• Medication side effects and use
4	Short	• Adherence to follow-up screen recommendations • Internalized shame • Environmental smoke	• Medication adherence • Discuss 3rd NRT dose for participants in longer NRT duration
5	Long	• Adherence to follow-up screen recommendations • Addiction & nicotine • Coping with slips and relapses	• Medication adherence
6	Long	• Adherence to follow-up screen recommendations • Risks of other forms of tobacco/nicotine • Distress management	• Medication adherence
7	Long	• Adherence to follow-up screen recommendations • Preparing to stay quit	• Review need for additional or extended medications
8	Long	• Expectations for annual lung screen follow-up • Relapse prevention	• Discuss plans for future medication needs and use

^aSessions 1–4 are conducted weekly and Sessions 5–8 are conducted every 2 weeks.

3.6.3. Counseling delivery

Participants are given a choice of phone-based or HIPAA-compliant videoconferencing-based virtual visits, with participant choices tracked. Participants are sent a secure hyperlink to take them to their videoconferencing session.

3.6.4. Treatment fidelity

Treatment fidelity is monitored using recommendations by the Treatment Fidelity Workgroup of the NIH Behavior Change Consortium (BCC) and a peer supervision model that mirrors the Smokefree Support Clinical Service at the MGH Cancer Center, which is also directed by one of the study MPIs. The BCC cites monitoring 5 areas to promote reliability of behavioral interventions: Design. Each week a list of counseling sessions completed are reviewed by the TTS and senior investigator. Training. The TTS underwent national training to become a certified tobacco treatment specialist and was trained internally on the counseling protocol and MI delivery. Treatment Delivery. All sessions are recorded, and each week one of the MPIs and the TTS review selected session recordings, MI counseling delivery, and complete an MI checklist. Receipt of Treatment. The session topics covered are reviewed during weekly supervision and documented in the study database. Enactment of Treatment Skills/Knowledge. The protocol is tailored to participants’ stage of change, and participants’ treatment goals and progress are monitored.

3.7. Smoking cessation medication

All participants receive the same information about the risks and benefits of cessation medication use. Participants in the short duration arm are offered one 2-week kit. Participants in the long duration arm are offered one 4-week kit and are eligible for one additional 4-week refill, using a tapering dose schedule for a total NRT duration of 8 weeks. Participants in both short and long arms receive NRT by mail and at no cost.

3.7.1. Medication dispensing procedures

At the first counseling session, the TTS promotes medication use, describes all FDA-approved options, assesses appropriateness of using NRT, and offers free NRT patches to smokers without contraindication. Study medication is mailed to the participant with an instruction sheet for usage. NRT dose is determined by daily cigarette use: 21 mg/d patch for ≥10 cig/day, 14 mg/d or 7 mg/d patch <10 cig/day. The TTS also assists smokers who choose other medications to obtain prescriptions from their clinicians. Although use of medication is promoted, it is not required for study participation. Participants are not required to use NRT or may request NRT in the quantity assigned at subsequent sessions.

3.8. Systematic screening for social determinants of health and community referral (Aunt Bertha)

For participants who are randomly assigned to receive this resource and agree to the systematic screening and referral for social determinants of health, the TTS completes a 16-question assessment within a web-based platform (Aunt Bertha; https://www.auntbertha.com/). This assessment is completed during the first counseling session and includes the following domains: transportation, food, housing, paying for healthcare, employment, education, child/family care, legal services, phone service, health education/fitness programs, and social support. A branching logic suggests local, free or low cost services that meet the participant’s needs and the TTS selects the most appropriate referrals. An RA then mails out the information about the referrals to the participant. Some participants who elect to search for information themselves, are emailed instructions on how to do so by the TTS. The TTS discusses whether the participant utilized the referrals or was able to navigate the website on their own in the second counseling session.

3.9. Assessments

Participants complete assessments 3 times during the study period, at study entry and at 3 and 6 months after randomization. Assessments are done either electronically (through the secure web application, REDCap), administered over the phone, or via hard copy mailed survey in accordance with participant preferences. The baseline survey is completed following informed consent and prior to initiating counseling; follow-up surveys are completed 3 and 6 months following baseline survey completion. For participants who prefer Spanish-language intervention and materials, existing and validated Spanish language versions of study measures are used whenever possible. At 3-and 6-month follow-ups, research staff initiate survey outreach attempts (e.g., phone calls, and emailed electronic surveys) approximately two weeks prior to the target assessment time point. A minimum of 5 call attempts are made for each survey follow-up, with calls stopped 60 days from target survey date. For participants who do not complete their follow-up surveys in a timely manner (approximately 3 weeks after target date), a mailed paper survey that assesses primary and secondary smoking outcomes is mailed. All assessments take approximately 10–15 min to complete and participants are provided $20 in remuneration for each completed survey.

3.10. Participant surveys

The Smoking Cessation at Lung Examination (SCALE) Collaboration is an initiative sponsored by the National Cancer Institute to conduct research on long-term smokers who are screened for lung cancer using LCS. Screen ASSIST is a study site for this cross-project research collaboration and is participating in a collection of common measures (the “SCALE Measures Special Collection”). These measures include: sociodemographic, family medical history, psychological variables, historical and current smoking behavior, smoking cessation attitudes/experiences, smoking cessation outcomes, implementation, and medical outcomes. All study measures are summarized in Table 3.

Table 3.

Summary of study measures across study timepoints.

Construct	Measures	Source	Baseline	3 months	6 months
Sociodemographic factors	Sex, age, race/ethnicity, education, and preferred language (English or Spanish).	Patient, EHR	X
Medical history	Comorbid tobacco-related disease, ambulatory visits.	Patient, EHR	X
Smoking characteristics	Current cigarettes/day, number of years smoked, e-cigarette use, past and current use of cessation medication and other tobacco products, 24-h intentional quit attempt, nicotine dependence.	Patient	X	X	X
Nicotine dependence	2-item Heaviness of Smoking Index from the Fagerstrom Test for Nicotine Dependence [26]	Patient	X	X	X
Screening results	Lung RADS Score results of any diagnostic tests (CT or biopsy)	EHR
Smoking cessation beliefs	Readiness to quit (1 item 10-point ‘contemplation ladder) [27], importance and confidence to quit (1 item 10-point scales), 4 item perceived lung cancer risk, and 2 item perceived benefits of quitting.	Patient	X	X	X
Emotional symptoms	Anxiety and depression symptoms (PROMIS Anxiety and Depression 4a short-forms) [28], worry about lung cancer (1 item about level).	Patient	X	X	X
Environmental factors	Living with a smoker in the household, 1 item 4-point scale about rules about smoking in the household, environmental workplace/hobby exposure and social support (PROMIS emotional and informational 4a short forms) [28].	Patient	X	X	X
Health and screening beliefs	Self-reported overall health status (1 item from the SF-36 survey) [29], 2 item perceived benefits of screening	Patient	X	X	X

3.11. Exit interviews

A sample of 72 participants (approximately 10% of anticipated 6-month survey completers) are invited to participate in an in-depth individual interview after completing their 6-month survey assessment. Interviews will help further explore key constructs within the RE-AIM framework. Participants are stratified by LCS site, intervention group, and quit status. Interviews are conducted on the telephone and use a semi-structured interview guide. All participants are asked about domains related to: 1) smoking and quitting behaviors, 2) the enrollment process (e.g., timing, video recruitment messages, iPad use), 3) coordination with the LCS process, 4) understanding of their test results, and 5) adherence to future LCS recommendations. Participants are asked for feedback on the intervention components (counseling content and duration, medication dose and access, and the community resources referral process) and about their overall satisfaction with the tobacco cessation counseling received. Interviews last approximately 30 min, and patients are provided $20 remuneration. Interviews are audio-recorded and transcribed verbatim by a professional transcription service. A thematic analysis is being conducted in Dedoose by two study RAs, who double code each transcript. Emergent codes are operationally defined and entered into a formal codebook. An MPI and the project director guide the coding and development of a coding framework. To establish trustworthiness, an audit trail is kept and reviewed in biweekly meetings to examine the analysis process, provide feedback on emergent codes, and reach consensus on principal themes that are identified. To ensure coding reliability, discrepancies are resolved through discussion and comparison to raw data. Finally, a qualitative framework method analysis is being conducted, in wich codes are summarized and recatgorized into broader concepts to reflect the most salient themes. These themes are then charted by strata (LCS site, intervention group, and quit status), and comparisons by strata are made to aid interpretation.

3.12. Cost

At the end of the study, we will calculate the incremental cost per quit (ICQ) of the interventions over the 6-month follow-up period. In order to guide stakeholder decision-making and planning, we will focus on costs associated with implementing and providing the intervention in the short or medium term but will exclude longer term costs relevant to a full societal-perspecitve analysis such as future health care costs. ICQ calculations will proceed as follows: (total per-person costs of LCS-tailored intervention – total per-person costs of the lowest intensity comparator/(cessation rate with the LCS-tailored intervention – cessation rate with the lowest intensity comparator). From these comparisons, we will also be able to construct a “league table” comparing the relative cost-effectiveness of different intervention components. While the study is specifically powered to detect differences across factorial dimensions, we will develop incremental cost per quit estimates for all intervention comparisons that are proven effective, whether it be for individual factorial dimensions’ main effects (counseling duration, NRT duration, or systematic screening), for specific permutations of the intervention components versus usual care or for intervention permutations versus each other.

Costs of implementation included in our analyses will be: 1) personnel time related to intervention delivery (training, minutes of counseling, contact attempts, maintaining data and information systems), 2) NRT costs (medication, delivery), and 3) systematic referral to community resources. We will also include participant time as an indirect cost. Cost data collection methods will adapt procedures used in our prior studies [30–32] as well as those methods outlined by members of the SCALE Collaboration [33]. The TTS database will document all implementation costs.

All intervention costs incurred will be included in the analysis, even for participants who do not complete their intervention course. Research costs will be excluded. Effectiveness (smoking cessation) for the cost analyses will be based on the primary outcome – self-reported 7-day point prevalence smoking cessation at 6 months.

Uncertainty in cost and effectiveness inputs will be incorporated into the incremental cost per quit comparisons using Monte Carlo methods allowing us to determine whether these ratios are significantly different from zero. The robustness of the cost-effectiveness ratio estimates will be further examined in sensitivity analyses in which each parameter is varied, singly and in combination, through plausible ranges. Sensitivity analyses will also consider how cost-effectiveness changes based on payment arrangements and stakeholder (provider, payer, and participant) perspectives to identify where incentives to adopt the interventions differ. For example, overhead costs such as counseling training and data/information systems are borne by the provider, and indirect costs are borne by participants, but treatment costs (counseling, NRT) and the cost of systematic referral-style resources may be borne by providers, payers, and/or participants (e.g., copayments). Assessing how assignment of costs to different stakeholders changes cost-effectiveness from different perspectives will be critical to understanding how to scale up the intervention to other LDCT screening settings and what payment policies will produce the best public health outcomes.

3.13. Outcomes measures

3.13.1. Primary smoking outcome

Participants who enrolled in Screen ASSIST from April 15, 2019 to June 30, 2020 were required to provide biochemically-verified abstinence if they had self-reported as abstinent for at least 7 days during the 3-and 6-month follow-up assessments. However, the COVID-19 pandemic forced a change due to the infeasibility of obtaining in-person samples of carbon monoxide for biochemical verification after the Commonwealth of Massachusetts’ state-wide stay-at-home advisory, imposed on March 13, 2020, began. Biochemical verification was removed as a criterion of assessment for participants enrolled after July 1, 2020. The primary outcome was changed to self-reported 7-day abstinence at 6 months. This change was approved on May 18, 2020, in agreement with the NCI Project Officer.

Prior to this change, participants were required to provide a 7-day point-prevalence tobacco abstinence at 6-month follow-up confirmed biochemically by saliva cotinine (<15 ng/ml) or expired air CO (<10 ppm). Participants were mailed a saliva collection kit with instructions for use and a pre-addressed, pre-stamped return envelope. Saliva samples were then sent to J2 Laboratories (Tucson, AZ) for cotinine assay. Participants who reported smoking abstinence but were using nicotine-based smoking cessation medication or e-cigarettes were asked to complete an in-person, expired air carbon monoxide sample. Participants who enrolled during April 15, 2019 to June 30, 2020 were also remunerated $20 for each cotinine or expired air sample provided.

3.13.2. Secondary smoking outcomes

(1) self-reported 7-day point prevalence abstinence at 3-month follow-up; (2) significant reduction in cigarettes per day by self-report (>50% decrease from baseline in cigarettes) at 3-and 6-month follow-up; and (3) prevalence of making at least one >24-h intentional quit attempt at 3-month and 6-month follow-up.

3.14. Data analysis

3.14.1. Factorial design

In a factorial research design, 2 or more independent intervention components (factors) are concurrently examined within the same trial. For the Screen ASSIST trial, the 3 factors are crossed with one another to create a total of eight experimental conditions. An equal number of subjects are randomly assigned to each condition, using a random number generator. This is not an 8-arm randomized controlled trial, but instead allows for an efficient examination of main effects for each variable over the entire sample of 720 subjects. If there is no 3-way interaction between factors, this approach allows examination of 3 key treatment development questions in a much more time-efficient and economical manner by simultaneously performing 3 studies within the single trial. This achieves power to detect between-group differences that is equivalent to performing 3 separate randomized trials.

3.14.2. Analysis

All analyses are intent-to-treat; we will classify participants who are lost to follow-up as current smokers. We will explore whether the mechanism of missing data is missing at random by comparing participant characteristics between those who complete follow-up vs. those who do not. We will perform sensitivity analysis: 1) limited to those who have complete data, and 2) multiple imputation for missing data.

Outcome measures will be assessed at baseline, 3-, and 6-months follow-up. At the outset, we will examine the frequency distributions of all variables. Data from all MGB sites will be pooled for analysis, after confirming that there is no significant heterogeneity among sites or adjusting as needed. We will compare the baseline characteristics to assess whether randomization distributed covariates evenly. We will determine whether there is differential dropout in the groups and consider developing probability-of-completion weights to obtain unbiased estimates of treatment effect. The primary analysis will assess the effects of each intervention component in terms of its association with the primary outcome (self-report of cessation at 6-months follow-up), secondary smoking outcomes and proximal outcomes/mediators: (a) short vs long counseling (conditions 1/2/5/6 vs. 3/4/7/8), (b) short vs long NRT (conditions 1/3/5/7 vs conditions 2/4/6/8), and (c) provision of systematic referral or not (conditions 1–4 vs. 5–8) (see Table 1 for condition descriptions). Given the balanced complete factorial design, each of the 3 main effect estimates will be based on the full sample size of 720. Cross-sectional analyses will be conducted for outcomes assessed at 3 and 6 months separately. For binary outcomes, chi-square tests will be used to compare the outcomes between groups for each follow-up time. For continuous outcomes, two-sample t-tests or Wilcoxon rank sum tests, whichever is more appropriate, will be used to compare between groups. A longitudinal analysis using Generalized Estimating Equations (GEE) techniques will also assess the overall impact of each component by including data from all follow-up time points.

The secondary analysis will examine the interaction between components, which will provide insight into whether certain combinations of components are most useful. We will test the 3-way and 2-way interaction terms in the regression models. If there are component interactions, we will determine whether the components are synergistic or antagonistic. Exploratory analyses will examine whether treatment effectiveness is moderated by participant characteristics and other covariates. Of primary interest are sociodemographic characteristics (sex, race/SES, age), smoking characteristics (quit motivation, nicotine dependence), medical history, LCS-related factors (test results, and point of study entry), delivery modality selected (phone vs. video), language of participation (English vs. Spanish) and timing of enrollment (enrolled before or after March 13, 2020). We will test the effects of these factors in multivariable logistic regression models to determine their association with the primary and secondary smoking outcomes. We will test for interactions between intervention and these factors to determine whether intervention effects vary among subgroups. Those with significant interactions (p < .15) will be considered as candidates for identifying subpopulations.

3.14.3. Sample size and power calculations

Power calculations are based on our primary analysis, which is to assess the effect of each intervention component (e.g., shorter counseling vs. longer counseling) on self-reported 7-day point-prevalence tobacco abstinence at 6-months follow-up. This is a two-sample comparison between participants that receive one intervention component (e.g., longer counseling) and those that receive the other (e.g., shorter counseling). We aim to recruit a total sample of 720 participants. This sample size allows us to allocate N = 360 to each treatment component comparison (main effect). Our outcome of interest is also assessed at 3-month follow-up, which allows us to use a longitudinal generalized estimating equations approach wherein power is proportional to the within-person correlation. We estimate the correlation between 3 month and 6 month smoking cessation to be 0.6. Assuming N = 360 participants in each condition, a Type 1 error rate of 0.05, a 25% attrition rate by 6 month follow-up, and a 15% cessation rate for the reference group, we have 92.3% power to detect a risk ratio of 1.6 (15% vs. 21%). Under those same conditions, but with a 10% cessation rate for the reference group, we have 79.0% power to detect a risk ratio of 1.6 (10% vs. 16%). Given that this is full factorial design, we have the same power to detect two-way interactions between components.

We will evaluate the reach, adoption, effectiveness implementation, and maintenance of the treatment program and will use constructs within RE-AIM framework to project implementation capacity in the MGB Healthcare System. Our objective is to use these data for MGB as well as have these data available for health care systems that are considering adopting aspects of Screen ASSIST to determine its feasibility within their systems (e.g., cost, personnel burden). To assess the representativeness of the sample, we will compare participant characteristics between smokers who enroll in the study and those who are deemed eligible but decline to enroll. We will document reasons for ineligibility and refusal and document EHR-recorded sociodemographic and biological characteristics, medical history, and cancer history of refusers. While the results of the effectiveness of the intervention are critical, the eventual impact of any program will depend upon its ability to be implemented in practice. We will assess Proctor’s recommended outcomes for implementation exploration. Specifically, we will measure acceptability (satisfaction with content/delivery), adoption (program uptake), cost, fidelity (implementation), penetration (reach), and sustainability (maintenance).

3.14.4. Reach

To assess participant engagement in smoking cessation, we will measure the proportion of eligible participants willing to enroll. We will assess reach in two ways: (1) the number of eligible participants undergoing LCS who were approached about the study at each timepoint and (2) the number of eligible participants enrolled, and characteristics of refusers and enrollees. Adoption. The proportion of participants who participate at each LCS site. Effectiveness Implementation. To assess participants’ participation with the intervention, we will track delivery modality selected (phone or video) and language of participation. Within each intervention group, we will assess a) number of counseling sessions completed, b) weeks on NRT, and c) use and type of referrals from AuntBertha. We will also assess the fidelity of the delivery of the intervention components for each participant (see Treatment Fidelity above). At the 3-and 6-month surveys, participant satisfaction with the program will be assessed with four multiple choice questions with text for comments: (1) “To what extent has the Screen ASSIST Study Program met your needs?”; (2) Did you get the kind of smoking cessation assistance that you wanted?; (3) “How would you rate the quality of the smoking cessation assistance that you received?”; (4) “How helpful has Screen ASSIST been for you?” Maintenance. To guide our assessment of the extent to which the program becomes part of organizational practices/policies, we will conduct meetings with stakeholders who include the 1) MGB Population Health leadership, 2) radiology leadership, and 3) primary care practice leadership. The leadership teams will review cost outcomes and effectiveness results and evaluate program sustainability. They will advise on how to advocate for the LCS specific integrated model of smoking cessation coverage among payers.

A convergent parallel mixed methods design will enhance the evaluation of the program (methodological triangulation); specifically, quantitative and qualitative data will be combined to determine the convergence, divergence, and relationships between the survey and interview results. Once data analyses and interpretations are completed, we will leverage expert clinical review from our study team to discuss the implementation of smoking cessation in the LCS setting at MGB and wider dissemination (investigator triangulation). We will triangulate data from the different sources (participant survey, exit interview and investigator data) to strengthen effectiveness, adoption, and implementation findings. For Effectiveness, we will assess if the intervention components that are associated with the highest quit rates are also described as useful in the interviews, and if proximal outcomes/mediators (i.e., benefits of screening) compare with descriptions of understanding of test results and adherence to screening recommendations. For Adoption, we will understand perspectives of the enrollment process and coordination with the LCS process. For Implementation, we will assess how the intervention component utilization and intervention satisfaction compares with participants’ descriptions.

4. Discussion

Screen ASSIST offers an innovative model to address the need to integrate smoking cessation support into LCS. Integration is crucial to leverage a “teachable moment” to increase quit rates, ensure that smokers don’t view LCS as a protective alternative, attempt to reduce disparities in access to tobacco treatment, and to enhance the cost-effectiveness of LCS. As many smokers undergoing LCS are not ready to make a quit attempt, providing person-centered cessation care to increase willingness and confidence to quit among heavy smokers may have a profound effect on quit success.

Screen ASSIST has several strengths. First, the 3-point recruitment design permits study staff to attempt multiple points of recruitment but also offers a systems-based approach for sustainable digital outreach. For example, RP1 has the complementary goals of recruiting a smoker prior to their LCS test, but through our digital outreach videos, also underlines the importance of quitting and attending the upcoming LCS test. In doing so, we are also trying to marginally improve overall LCS adherence rates. Second, we purposefully partnered with lung screening community sites that serve a large Hispanic/Latinx population, a population that reports greater barriers to accessing tobacco treatment [34]. Due to limited resources, patients at these sites are often not offered access to free tobacco treatment at the point of care. We, therefore, believe that this is another way in which to reduce disparities among this population. Third, we offer all study materials in Spanish, so as to increase participation among marginalized patient groups and our TTS is a native Spanish speaker. Four, we have leveraged the above rationale and received a successful NCI Diversity Supplement (PI Flores) to further improve patient-centered and culturally tailored outreach videos for Hispanic/Latinx smokers, so as to increase participation rates. A final strength is use of the multiphase optimization strategy factorial design, permitting economical testing of 3 different treatment factors.

Despite these strengths, there are some limitations of Screen ASSIST. Due to COVID-19, we had to change a number of study processes after trial launch. For example, at RP2, we were no longer able to have a study iPad at each LCS site, reducing awareness of the trial at the point of care. At specific study sites, we have trialled study pamphlets to try and maintain awareness and clinic presence. Further, we are no longer able to biochemically validate abstinence. We do expect this may increase quit rates because not all participants who self-report abstinence will provide a biological sample and not all samples will verify abstinence. As a result, there was deliberated decision in whether to remove biochemical validation with the NCI Project Officer in Spring 2020. The decision was influenced by the forced and indefinite removal of expired air CO as a form of validation (i.e., study staff were not able to collect in-person samples from participants due to fear of COVID-19 transmission) and the reduced completion rate of mailed salivary cotinine (i.e., we believe participants were fearful of providing a saliva sample for risk of transmitting COVID-19). While biochemical validation is desirable, it proved to be impossible to safely and reliably collect samples. This resulted in other trials within the NCI SCALE Collaboration making the same decision to remove biochemical validation as a primary outcome.

With expanded USPSTF guidelines for LCS coming into effect in March 2021 and covered by Medicaid from 2023, Screen ASSIST offers a timely understanding of how to integrate and implement this care into the LCS process. There have been numerous challenges presented as a result of COVID-19, and in particular, the deferment of all elective procedures and preventive screening at MGB. Despite these challenges, Screen ASSIST has demonstrated a promising model of remote tobacco treatment delivery.