Integrating Tobacco Use Assessment and Treatment in the Oncology Setting: Quality Improvement Results from the Georgetown Lombardi Smoking Treatment and Recovery Program

As part of the NCI’s Cancer Center Cessation (C3i) initiative, we initiated, expanded, and maintained an evidence-based tobacco treatment program at the Georgetown Lombardi Comprehensive Cancer Center. We present a quality improvement (QI) assessment of the implementation process and patient-level outcomes. At two hematology/oncology outpatient clinical sites, five oncology-based teams (clinical administrators, clinical staff, pharmacy, information technology, and tobacco treatment staff) developed implementation strategies for opt-out patient assessment and enrollment, centralized tobacco treatment, audit, feedback, and staff training. Among eligible patients (tobacco use in ≤30 days), we assessed demographic, clinical, and tobacco-related characteristics to examine predictors of enrollment (baseline completed), treatment engagement (≥one sessions completed), and self-reported 7-day abstinence (6 months post-enrollment). Across both sites, medical assistants screened 19,344 (82.4%) patients for tobacco use, which identified 1345 (7.0%) current tobacco users, in addition to 213 clinician referrals. Of the 687/1256 (54.7%) eligible patients reached, 301 (43.8%) enrolled, and 199 (29.0%) engaged in treatment, of whom 74.5% were African American and 68% were female. At the larger site, significant multivariate predictors of enrollment included African American race (vs. white/other) and clinician referral (vs. MA assessment). Treatment engagement was predicted by greater nicotine dependence, and abstinence (27.4%) was predicted by greater treatment engagement. In summary, the systematic utilization of multiple oncology-based teams and implementation strategies resulted in the development and maintenance of a high-quality, population-based approach to tobacco treatment. Importantly, these strategies addressed inequities in tobacco treatment, as the program reached and engaged a majority-African-American patient population. Finally, the opt-out patient assessment strategy has been implemented in multiple oncology settings at MedStar Health through the Commission on Cancer’s Just Ask program.


Introduction
Among patients diagnosed with cancer who smoke, an estimated 20-30% continue smoking following diagnosis [1,2]. Continued tobacco use during and following cancer treatment contributes to poorer treatment outcomes, including increased risk of cancerspecific and all-cause mortality, cancer recurrence, second primary cancers, and treatmentrelated toxicities [3]. In contrast, smoking cessation can significantly improve health outcomes for cancer survivors [3,4]. Guidelines [5,6] recommend treating tobacco dependence as a routine part of cancer care. However, the implementation of oncology-based tobacco cessation programs has been limited until recently [7].
Between 2017 and 2020, the National Cancer Institute began the Cancer Center Cessation Initiative (C3i) to support the implementation of tobacco treatment programs at 52 NCI-designated cancer centers. The goal of C3i is to ensure that all cancer patients who smoke have access to smoking cessation support during and following treatment [7]. With C3i funding, we established the Smoking Treatment and Recovery (STAR) Program to provide evidence-based tobacco treatment to hematology and oncology patients at the Georgetown Lombardi Comprehensive Cancer Center, in collaboration with its clinical partner, MedStar Georgetown University Hospital (MGUH) [8]. In 2020, supported by MedStar Health, we implemented the STAR program at the Washington Cancer Institute of the MedStar Washington Hospital Center (MWHC), a regional cancer center that serves a diverse patient population, including racial and ethnic minority groups and patients with fewer socioeconomic resources.
Several C3i programs have evaluated the barriers and facilitators of the successful implementation of smoking cessation [9][10][11][12]. One notable barrier has been the lack of involvement of information technology (IT) leadership, which is critical for completing the electronic health record (EHR) modifications needed to improve the documentation of tobacco use [13]. Additionally, patients' lack of awareness of the harms of continued smoking after a cancer diagnosis is a barrier to engaging in smoking cessation treatment [14]. Facilitators of implementing tobacco treatment programs include the involvement of cancer center and healthcare system leadership (e.g., oncology and pharmacy administrators) [12]. Providing free nicotine replacement therapy (NRT) can facilitate patients' engagement with treatment [15]. The objective of this paper is to describe the quality improvement assessment of the three primary strategies we utilized to facilitate the identification of eligible patients as well as treatment enrollment and engagement, and to present the patient outcomes that resulted following the use of these strategies [16].

Opt-Out Identification of Eligible Patients
Identifying patients who may benefit from tobacco treatment using a proactive, "optout" approach requires the assessment of all patients in a given population [17,18]. In traditional clinical care models that employ an "opt-in" approach, patients either accept or decline a clinician's referral. This method typically limits tobacco treatment referrals to those who indicate that they are ready to quit, while excluding those who are not ready to quit but who may be willing to receive smoking cessation support [19]. Opt-out approaches can simultaneously reduce clinician burden (i.e., clinicians do not need to identify and refer patients) and reduce patient burden (i.e., patients do not need to request a referral).
The methods used to identify eligible patients for tobacco treatment have important implications, with preliminary research indicating that opt-out approaches result in the identification of a larger number of patients, higher reach and engagement [19], and higher quit rates [20], compared to opt-in approaches. Furthermore, an opt-out approach does not preclude clinician referrals.

Centralized Delivery of Tobacco Treatment
The centralized delivery of tobacco treatment can involve the provision of services across multiple clinic and hospital locations and via multiple treatment modalities, in-cluding counseling delivered in-person or via phone or video telehealth, and the use of electronic resources (e.g., texting programs) [17]. Centralized treatment delivery, particularly by phone and video telehealth, may require fewer staff members while also providing treatment across multiple sites. Furthermore, offering multiple treatment modalities may appeal to patients' personal preferences while also having similar abstinence rates [21,22]. Although telehealth visits have been shown to have lower no-show rates [23] and similar patient satisfaction levels [24] compared to in-person visits, disparities in digital access continue to exist [25,26], highlighting the need to offer both (non-video) phone and in-person counseling [27].

Staff Training and Audit/Feedback
Ongoing quality improvement (QI) procedures are integral to maintaining evidencebased tobacco assessment and treatment [28,29]. While individual training is important, team training interventions can improve clinical care teamwork (e.g., communication and coordination), as well as patient mortality and morbidity [28]. Within settings of team collaboration, particularly when providing feedback, it is important to ensure a setting in which interpersonal risk-taking is supported, in order to promote communication and positive teamwork outcomes [30]. Another example of a QI method is 'audit and feedback', a practice used to assess performance on an individual or clinical level and to then relay this information back to the individual or practice [31][32][33]. Providing regular feedback and training improves patient care as well as collaboration among team members [28].
We utilized these three implementation strategies to achieve patient identification, outreach and enrollment, and centralized tobacco treatment and follow-up. To better understand the program's strengths and areas in need of improvement, we conducted a QI assessment of the strategies used for program development and maintenance. Furthermore, we present the demographic, clinical, and tobacco-related characteristics of eligible patients, and the predictors of patient enrollment, treatment engagement, and abstinence in the STAR program.

Overview of Program Development
As recommended by the C3i leadership [7], we developed a population-based registry within specified clinics to offer tobacco treatment to all patients who currently smoke. Prior to the development of the STAR program, there had not been an oncology-based tobacco treatment program at MedStar Health. The Cancer Center leadership, oncology clinical and administrative staff, the STAR program lead, and an expert consultant (Dr. Graham Warren) identified the relevant team members and clinical champions.
The teams made four central decisions necessary for program implementation: (1) Patient identification: To maximize the identification of patients who currently smoke, we selected an opt-out method rather than a referral-based method. One approach is for medical assistants (MAs) to document tobacco use at every outpatient visit while collecting other vital signs. This method provides more complete data collection, including showing the change in tobacco status over time, particularly important when patients were unreachable for 6-month outcome assessments. (2) Clinic location: We considered providing point-of-care tobacco treatment in conjunction with an oncology visit vs. providing separate appointments in the STAR clinic. In consultation with the clinical champions (thoracic oncologists) and clinical administrators, we determined that the busy oncology clinics and lack of available consultation rooms would not allow for point-of-care tobacco treatment. (3) Staffing of TTP: As the oncology clinicians and nursing staff were already overcommitted with oncology-related care, we determined that adding tobacco treatment to their duties would be unlikely to result in consistent, high-quality treatment. As an alternative, we elected to provide centralized treatment using multiple modalities, delivered by a full-time tobacco treatment specialist (TTS) and a part-time nurse practitioner (NP) with tobacco treatment training [34,35]. (4) EHR documentation: We determined that locating smoking status in the EHR vital signs section [36] and using a standardized single item [37] would assist with MA workflow and consistent documentation. In March 2021, MedStar's EHR changed from Aria to Cerner, which initially reduced tobacco assessments due to the new MA workflow, but ultimately resulted in improved assessments and EHR-based communication with clinicians. Table 1 details the relevant teams, facilitators and barriers of implementation, and the outcomes of the three implementation strategies. The TTSs were the only team members who were hired to implement the STAR program. The hematology/oncology clinical staff (e.g., MAs, NPs) and clinical administrators were all part of the existing cancer center staff with other duties. The data manager received a small percent for their effort in obtaining the numbers needed for audit and feedback. The procedures were identical across locations (MGUH and MWHC) and are described below (Table 1). Of note, although we have described the program procedures and outcomes in the past tense, the STAR program is ongoing.  Providers are reminded of the rationale for tobacco treatment, institutional program support, and to refer their patients.

Implementation Strategies and Teams
The three implementation strategies were as follows: (1) The opt-out strategy required the involvement of IT, oncology, clinical, and STAR teams. Each week, the STAR team downloaded the patients newly identified by the MAs as having smoked in <30 days [36,37]. STAR outreach staff called patients (≤3 attempts) to describe the program, enroll them and conduct a brief baseline assessment, and schedule an initial counseling session. Patients uncertain about enrolling were offered another outreach call in two months. (2) The centralized treatment strategy required actions by the STAR, IT, clinical, and pharmacy teams. The STAR staff provided treatment to both MGUH and MWHC patients (see patientlevel procedures below). (3) The audit/feedback/staff training strategy required actions by the STAR, oncology administration, and clinical teams. Each month, the STAR data manager assessed the number of unique patients seen in the hematology/oncology clinics, the percentage of the patients assessed for tobacco use, and the number of patients identified as currently smoking. Using audit and feedback, we contacted the nurse manager when the percentage of patients assessed declined, who then (re)trained the MAs. Additionally, each year the STAR staff held 1-2 in-person or Zoom training sessions. The TTSs received monthly supervision from an experienced clinical psychologist. The clinicians received (1) monthly EHR updates about the status of their patients identified as currently smoking, and (2) an overview of all program metrics at the annual faculty presentation.

The STAR Clinical Program
Tobacco treatment specialists (TTS; 1.2 FTE) and nurse practitioners (NP; 0.2 FTE) provided up to four 30 min sessions (plus the offer of two booster sessions at the six-month assessment) of evidence-based behavioral counseling and medication, as needed. Readiness to quit was not an eligibility criterion. All patients were encouraged to engage in treatment, including those not ready to quit, and the treatment was personalized to patients' goals. We offered first-line treatments of behavioral counseling, NRT, varenicline, and bupropion [5].
The TTS began by conducting a complete smoking history assessment, patients' experiences with prior cessation methods, their current smoking patterns, smoking-related goals (i.e., timeframe for quitting or cutting down), and interest in medication use (NRT and/or prescription medications). From this assessment, an individualized treatment plan was developed in collaboration with the patient. Standard behavioral strategies were utilized (e.g., use of cigarette tracking, methods for reducing cigarettes per day, practice quit dates). Motivational interviewing was used to guide conversations on behavioral strategies and medication management. Quit dates were set by patients, when (and if) they became ready to make a quit attempt. For patients who were willing to consider medication use, initial and follow-up appointments were scheduled with the STAR nurse practitioner, who was also trained as a TTS. Following telehealth visits, hospital pharmacies delivered medications to patients' homes. Due to an in-kind donation from CVS Health, since 2020 we have provided free nicotine patches and lozenges, as appropriate. Usual co-pays were required for prescription medications.

Patient-Level Procedures
Using a population-based approach, the denominator includes all hematology/oncology outpatients who had an appointment within the given time period (Figures 1 and 2). Using the implementation strategies described above, MAs screened patients at each outpatient visit for tobacco use. The STAR team contacted patients who had smoked in <30 days by phone, mail, and/or EHR portal for program enrollment. All assessments were completed by phone and tracked using the secure REDCap database. We contacted patients six months post-enrollment to assess current smoking status, to offer two booster sessions to patients who continued to smoke, and to solicit program feedback.

Patient-Level Measures
At the baseline assessment, we assessed demographic, clinical, and tobacco-related characteristics. Variables included race, ethnicity, age, sex, insurance status, primary diagnosis, and cancer stage (when applicable). Tobacco-related characteristics included smoking history (pack-years, prior quit attempts, and usage of medication and other cessation strategies). We evaluated nicotine dependence using the 2-item Heaviness of Smoking Index [38] (time to first cigarette after waking and cigarettes smoked per day), and assessed readiness to quit on a 10-point scale [39] (1 = not ready to quit, 10 = already quit). We assessed self-rated distress using a 10-point scale [40] (1 = low distress, 10 = high) wherein individuals who rated distress as a 7 or higher were referred to the psychosocial oncology team. We also ascertained whether the patient lived with others who smoked.
At the six-month assessment of MGUH patients, we evaluated 7-day point prevalence self-reported abstinence, treatment engagement (number of sessions completed and whether medications were prescribed), and program satisfaction (counseling sessions and medication offered). For patients who were unreachable at six months, we used the most recent smoking status from the MA assessments, and patients with no available data (N = 22) were classified as currently smoking.

Patient-Level Data Analyses
We conducted the analyses separately by site due to the differences in demographics, tobacco use, and clinical characteristics. As the MWHC program began two years after the MGUH program, we had fewer completed six-month follow-up assessments and thus predictors of abstinence were not assessed at MWHC. We used descriptive statistics and bivariate analyses (t-tests and X 2 tests) to describe the associations between baseline characteristics and program enrollment (i.e., completed baseline assessment), treatment engagement (i.e., completion of ≥1 counseling session), and abstinence at MGUH (i.e., 7-day self-reported abstinence, 6 months post-enrollment). We conducted adjusted logistic regression analyses to assess predictors of engagement and abstinence at MGUH and enrollment at both sites. All regression analyses controlled for baseline demographic and clinical characteristics associated with the outcome (p < 0.10). We used SPSS v.27 to conduct the analyses.

Implementation Results
Employed together, three strategies (Table 1) provided the necessary structure for program implementation, maintenance, and quality improvement. At both sites, compared to clinician referrals, the opt-out procedures resulted in the identification of the vast majority of patients who had smoked in <30 days. Over   clude too sick to participate as assessed via clinician, has quit/not interested in rela no cancer/hematology dx, never smoker, second opinion, identified outside of 2 hospice consultation/in hospice, no longer an MGUH patient, not fluent in Engl hearing impairment, marijuana-only smoker, vaping only. Note: as of 2022, patie are not fluent in English are now eligible for STAR.  The centralized treatment strategy resulted in an economy of scale as the TTSs and NPs provided evidence-based behavioral counseling and medication to patients across the two sites, through the use of phone, telehealth, and in-person visits. Due to late cancellations and no shows, phone sessions required less administrative time for TTSs than telehealth or in-person visits.
Using the audit/feedback/staff training strategy, we alerted the oncology clinical team to provide feedback and training as needed to the MAs, which improved the percentage of patients assessed and identified as currently using tobacco (Figures 3 and 4). For example, due to the decline in the percentage of patients assessed at MGUH in fall 2021 (Figure 3), we met with the lead MA and learned that staff shortages had led to a decline in tobacco assessments. We held a training session with the MAs and the lead MA provided individual feedback to those missing a high percentage of assessments, which increased the number of patients assessed and eligible patients identified (Figure 3). team to provide feedback and training as needed to the MAs, which improved the percentage of patients assessed and identified as currently using tobacco (Figures 3 and 4). For example, due to the decline in the percentage of patients assessed at MGUH in fall 2021 (Figure 3), we met with the lead MA and learned that staff shortages had led to a decline in tobacco assessments. We held a training session with the MAs and the lead MA provided individual feedback to those missing a high percentage of assessments, which increased the number of patients assessed and eligible patients identified (Figure 3). In Year 1 (2/2020 to 1/2021) at MWHC (Figure 4), due to the onset of the pandemic, we piloted the MA assessments only among thoracic oncology patients, which explains the disparity between the total number of patients seen and the percentage assessed for tobacco use. In Year 1 (2/2020 to 1/2021) at MWHC (Figure 4), due to the onset of the pandemic, we piloted the MA assessments only among thoracic oncology patients, which explains the disparity between the total number of patients seen and the percentage assessed for tobacco use.

Baseline Predictors of Enrollment
Compared to eligible patients who did not enroll (including never reached and eligible but declined; N = 359; 66.9%), those who did enroll (N = 178; 33.1%) were significantly more likely to be female, African American (vs. white/other racial group), to have smoked a cigarette today (vs. 1-30 days ago), to have early (vs. late) stage cancer, to have been referred by a clinician (vs. identified by MAs), and to have government-supported (vs. private) insurance ( Table 2).   Notes. * p = 0.10, ** p < 0.05, *** p < 0.001. 1 Enrolled/not engaged are those who completed the baseline assessment but did not engage in counseling; 2 NHPI: Native Hawaiian and Pacific Islander; MA: medical assistant; 3 tobaccorelated cancers included lung, head, and neck, stomach, kidney, pancreas, liver, bladder, cervix, colorectal, and acute myeloid leukemia; 4 patients reporting a score of 7 to 10 were offered a referral to the Psychosocial Oncology Program for evaluation and treatment. Bolded text indicates significant findings.

Baseline Predictors of Treatment Engagement
Among the 178 enrolled at MGUH, there was significantly greater engagement among individuals with greater nicotine dependence, greater distress, and not living with someone who was smoking (Table 2). Approaching significance, patients with early stage disease (p < 0.10) or who were referred by their clinicians (p < 0.10) were more likely to engage. A median of two counseling sessions were completed and 47.2% received NRT or prescription medications. Table 3 presents the adjusted logistic regression model predicting engagement. Individuals with greater baseline nicotine dependence were significantly more likely to engage in treatment (p = 0.009). The other bivariate predictors approached significance (p < 0.10).

Baseline Predictors of Abstinence at Six Months Post-Enrollment
Among the 178 enrolled, 146 (82%) reached the 6-month assessment (the remainder were not yet due) ( Table 2). Of those, 40 (27.4%) individuals self-reported 7-day point prevalence abstinence. Furthermore, among the 146 individuals who reached the six-month assessment, 41 (28.1%) had not engaged in treatment and 105 (71.9%) had engaged in treatment. The quit rates were 8/41 (19.5%) and 32/105 (30.5%), respectively. Due to the small Ns, we have not conducted significance testing, but these differences suggest that treatment engagement may be having the expected impact on cessation.
Compared to individuals who continued to smoke, the baseline assessment indicated that those who quit were significantly less likely to have used tobacco/nicotine today (vs. 1-30 days ago), further along on the readiness to quit continuum, less likely to live with someone who smoked, and smoked fewer CPD. Furthermore, those who quit completed significantly more counseling sessions. Table 4 presents the adjusted logistic regression model predicting 7-day abstinence at 6 months at MGUH. After adjusting for age, nicotine dependence, and living with someone who smokes, we found that those who completed two+ counseling sessions were almost three times as likely to quit smoking, compared to those who completed zero or one session (OR = 2.80; 95% CI 1.19, 6.58). At the 6-month assessment, we also assessed satisfaction with and engagement in counseling: 69.8% (37/53) who completed counseling were very satisfied with the treatment and 24.5% (13/53) were somewhat satisfied. Of those engaged in treatment, 40.9% (52/127) received NRT or prescription medication. The offer of two booster sessions was accepted by 69.6% of those who had not quit, and by 40% of those who had quit (for relapse prevention).

MedStar Washington Hospital Center (MWHC)
Of the 510 patients who we attempted to reach, 108 were ineligible or undecided, and we reached 285 (70.9%) eligible patients (Figure 2). Of those, 123 (43.2%) enrolled in the program, and 72 (25.3%) engaged in ≥one counseling session. The treatment modalities included 33 (45.8%) phone only, 2 (2.8%) in-person only, 10 (13.9%) video telehealth only, and 27 (37.5%) combination (of which 25 (92.6%) included ≥one video telehealth visit). High treatment engagement (defined as three+ sessions) was greater among those who used a combination of modalities (44%) compared to those who used phone only (30.3%). The chi square test was not valid as there were no individuals who completed three+ sessions in the other two groups. (Table 5) Of the 402 eligible MWHC patients, compared to the 279 (69.4%) who did not enroll, the 123 (30.6%) who enrolled were significantly more likely to be female, to have smoked today (vs. 1-30 days ago), and to have been referred (vs. identified by MAs).  Notes. * p = 0.10, ** p < 0.05, *** p < 0.001. 1 Enrolled/not engaged are those who completed the baseline assessment but did not engage in counseling. 2 NHPI: Native Hawaiian and Pacific Islander; MA: medical assistant; 3 tobacco-related cancers included lung, head, and neck, stomach, kidney, pancreas, liver, bladder, cervix, colorectal, and acute myeloid leukemia; 4 patients reporting a score of 7 to 10 were offered a referral to the Psychosocial Oncology Program for evaluation and treatment. Bolded text indicates significant findings.

Baseline Predictors of Enrollment and Engagement
Among those who enrolled (N = 123), there were no significant demographic, clinical, or tobacco-related differences between those who engaged (N = 72; 58.5%) vs. those who did not engage (N = 51; 41.5%) in treatment. As there were only two nonsignificant trends, we did not conduct a logistic regression analysis. Regarding engagement, patients completed a median of two counseling sessions and 59% received NRT or prescription medication.
Finally, of the 123 enrolled, only 81 (65.9%) had reached the six-month assessment point, and of those, 12 (14.8%) self-reported 7-day point-prevalence abstinence. Due to the limited numbers available at present, we did not assess predictors of abstinence at MWHC. Furthermore, among the 81 individuals who had reached the six-month assessment, 29 (35.8%) had not engaged in treatment and 52 (64.2%) had engaged in treatment. The quit rates were 2/29 (6.9%) and 10/52 (19.2%), respectively. As above, these Ns are too small to conduct significance testing, but are in the expected direction. Table 6 presents the adjusted logistic regression models predicting enrollment. At MGUH, compared to those who declined enrollment, those who enrolled were significantly more likely to be African American (vs. white/other) and referred to STAR (vs. identified by MAs). At MWHC, there were similar results, with those referred more likely to enroll, and trends for African Americans and older adults in terms of being more likely to enroll (vs. white/other races and younger adults, respectively).

Discussion
Tobacco use among patients with cancer results in poorer treatment outcomes, increased mortality, and treatment-related toxicities [3]. However, as the necessary resources and expertise are often unavailable in the oncology setting, patients frequently do not receive tobacco use treatment to stop smoking. As part of the NCI's C3i initiative and in collaboration with MedStar Health, we implemented, expanded, and have sustained the STAR program in two hematology/oncology settings. This has required the involvement of multiple oncology-based teams, including cancer center leadership, IT programmers and informaticists, clinical administrators, clinical staff, pharmacists, and the tobacco treatment team. The teams' utilization of implementation strategies to identify, reach, and enroll patients in an evidence-based tobacco treatment program has resulted in program expansion and maintenance.
To assess the program's strengths and limitations, we conducted a QI assessment of the implementation strategies and an analysis of the patient-level predictors of enrollment, treatment engagement, and abstinence. A total of 19,344 unique patients were assessed for tobacco use over two clinical sites, which averaged to a 7% rate of current tobacco use, comparable to other C3i sites [41] and to national rates of smoking among older patients with cancer [42]. The difference in smoking rates by site is unclear, but may be explained in part by the small sample size at MWHC and the higher percentage of individuals with private insurance at MGUH [43].
By leveraging key implementation strategies, including the opt-out approach and audit/feedback with the MAs, we sought to address inequities in cessation treatment, which resulted in reaching and engaging a majority-African-American patient population [19,44]. This is important, as African Americans and other minoritized groups are less likely to receive assistance to quit smoking from their clinicians [44,45]. The QI results revealed the importance of both the opt-out approach for tobacco assessment and the inclusion of clinician referrals. The opt-out approach provided a population-based registry of tobacco use across the specified clinics, the identification of more patients than is feasible in a referral-based program, and an assessment of the program's reach, treatment engagement, and abstinence outcomes. Involving clinicians in the referral process and regularly informing them of their patients' outcomes gave clinicians the information needed to engage with their patients about tobacco use. Importantly, patients referred by their clinicians were more likely to enroll in the program at both sites, and marginally more likely to engage in treatment at the MGUH site. Based on these results showing significantly higher enrollment following a clinician referral, we plan to modify the EHR so that clinicians are prompted to refer their currently smoking patients to STAR.
The percentage of patients who engaged in treatment was 31.6% at MGUH (Figure 1) and 25.3% at MWHC (Figure 2), both higher than the 18.4% median across 28 C3i programs [41]. Similarly, the 6-month abstinence rate at MGUH was 27.4% higher than the 18.4% median of the C3i programs, and with data collected on 85% of those enrolled. The abstinence rate at MWHC was lower at 14.8%, which we attribute in part to the smaller sample that had reached the 6-month assessment and thus requires continued assessment as the MWHC matures. Although there is certainly room for improvement, we attribute these successes to the teams-based approach and the implementation strategies utilized to implement, expand, and maintain the program.
The multivariate analyses predicting engagement at MGUH indicated that patients with greater nicotine addiction at baseline were significantly more likely to engage in treatment. This finding suggests the importance of offering pharmacotherapy to all patients engaged in treatment, particularly those with greater addiction. Our data indicated that we can improve in this area, as an average of one half of the engaged patients received medication from the program, possibly suggesting that not all patients understood the benefits of medication and/or its availability (i.e., free NRT or prescription medication that is covered by insurance). Furthermore, improved communication regarding the importance of engaging in behavioral counseling may be needed for patients smoking a few cigarettes a day, as their lower engagement may suggest less motivation to stop smoking and/or have a plan to quit on their own [41]. Efforts to increase patient engagement may benefit from communicating our findings to patients who are uncertain about engaging in treatment, such that patients who completed two+ counseling sessions were almost three times as likely to report abstinence compared to those who completed ≤1 session.
Based on the implementation and success of the STAR program, there are two new programs at MedStar Health designed to assess and treat tobacco use. First, with funding from the DC Department of Health, we are expanding tobacco assessment (conducted by MAs) to primary care departments, which will prompt clinicians to consider tobacco treatment, including medication prescriptions and/or e-referral to the tobacco quitline. Clinicians are also prompted to consider referral for lung cancer screening among potentially eligible patients. Second, as part of the Commission on Cancer's 'Just Ask' Program [46], MedStar and Lombardi administrators have employed the same single item opt-out tobacco assessment [37] in all oncology clinics across MedStar Health. As above, clinicians are prompted to consider cessation medications and/or e-referral to the quitline. Additional funding will be required in order for the STAR program to expand and provide tobacco use treatment for oncology patients across MedStar Health.
The program's limitations include the following. First, the data associated with the clinical program are not fully integrated into the EHR. Due to the time and expense required for making additional EHR modifications at the program's outset, we elected to record patient responses in a parallel secure database (REDCap). Although the MAs' tobacco assessment is accessible in the EHR, it is not widely utilized by clinicians, and thus requires an EHR modification to increase STAR referrals and clinician-initiated discussions about cessation. Second, despite regular audit, feedback, and training of staff, we have only recently approached 80% completion of tobacco assessment for all patients. This finding indicates that program maintenance requires continuous audit, feedback, and training, regardless of the amount of time since program outset. Finally, we have not yet conducted an economic analysis of the program and will undertake this effort once both sites have reached a steady state. In a recent paper, Salloum and colleagues [47] described an economic analysis of 15 of the C3i programs and concluded that they were "within the range of historical cost-effectiveness estimates of tobacco treatment . . . ".

Conclusions
In summary, the systematic utilization of multiple oncology-based teams resulted in the implementation, expansion, and maintenance of the STAR program. From this QI assessment and presentation of patient-level results, we have demonstrated the program's successes in engagement and cessation rates, as well as areas in need of improvement. For example, although the opt-out method was central for identifying patients who were using tobacco, bolstering referrals from all oncology clinicians is needed to expand treatment engagement among patients. Oncology clinics and departments planning to develop a tobacco use treatment program may benefit from our lessons learned and from employing the teams and implementation strategies described here. Funding: The program described was originally supported by the National Cancer Institute of the National Institutes of Health under Award Number P30CA051008-26S1 (PI: Weiner). Presently, the program is supported by MedStar Health, CVS Health (with an in-kind donation of nicotine replacement therapy), and by the Lombardi Comprehensive Cancer Center's Survey, Recruitment, and Biospecimen Shared Resource, which is partially supported by P30CA051008.
Institutional Review Board Statement: Ethical review and approval were waived for this study by the MedStar Georgetown Oncology IRB, as the study does not meet the definition of human subject research as outlined in 45 CFR 46.102. This a quality improvement project.
Informed Consent Statement: Patient consent was waived as the work is a quality improvement project necessary for the development and maintenance of evidence-based tobacco treatment in the cancer care setting. Data Availability Statement: Deidentified data presented in this quality improvement project are available upon request from the corresponding author. The data are not publicly available as this was a quality improvement analysis and not a research study.