The effect of neighborhood socioeconomic disadvantage on smoking status, quit attempts, and receipt of cessation support among adults with cancer: Results from nine ECOG-ACRIN Cancer Research Group trials

Abstract

Background

Tobacco use is associated with adverse outcomes among patients diagnosed with cancer. Socioeconomic determinants influence access and utilization of tobacco treatment; little is known about the relationship between neighborhood socioeconomic disadvantage (NSD) and tobacco assessment, assistance, and cessation among patients diagnosed with cancer.

Methods

A modified Cancer Patient Tobacco Use Questionnaire (C-TUQ) was administered to patients enrolled in nine ECOG-ACRIN clinical trials. We examined associations of NSD with (1) smoking status, (2) receiving tobacco cessation assessment and support, and (3) cessation behaviors. NSD was classified by tertiles of the Area Deprivation Index. Associations between NSD and tobacco variables were evaluated using logistic regression.

Results

A total of 740 patients completing the C-TUQ were 70% male, 94% White, 3% Hispanic, mean age 58.8 years. Cancer diagnoses included leukemia 263 (36%), lymphoma 141 (19%), prostate 131 (18%), breast 79 (11%), melanoma 69 (9%), myeloma 53 (7%), and head and neck 4 (0.5%). A total of 402 (54%) never smoked, 257 (35%) had formerly smoked, and 81 (11%) were currently smoking. Patients in high disadvantaged neighborhoods were approximately four times more likely to report current smoking (odds ratio [OR], 3.57; 95% CI, 1.69–7.54; p = .0009), and more likely to report being asked about smoking (OR, 4.24; 95% CI, 1.64–10.98; p = .0029), but less likely to report receiving counseling (OR, 0.11; 95% CI, 0.02–0.58; p = .0086) versus those in the least disadvantaged neighborhoods.

Conclusions

Greater neighborhood socioeconomic disadvantage was associated with smoking but less cessation support. Increased cessation support in cancer care is needed, particularly for patients from disadvantaged neighborhoods.

BACKGROUND

Tobacco use is the leading cause of preventable death and is responsible for increased cancer incidence and cancer-related mortality in the United States.^{1, 2} Socioeconomic factors have been associated with disparities in the prevalence of smoking in population subgroups.^{1, 3-5} Socioeconomic deprivation has been associated with behavioral risk factors for cancer incidence such as high rates of smoking and low rates of successful smoking cessation. Greater neighborhood deprivation has been associated with increased odds of smoking,^{6, 7} lower likelihood of quitting smoking,⁸ and reduced completion of treatment to support quit attempts.⁹ Furthermore, community environments with a high density of tobacco outlets are associated with higher smoking prevalence, increased tobacco use and smoking initiation, and less optimal cessation outcomes.¹⁰ Comprehensive tobacco use assessment and cessation support integrating the 5 As model (i.e., Asking about tobacco use, Advising users to quit, Assessing willingness to quit, Assisting in quit attempts, and Arranging follow-up contact)^{11, 12} in cancer care delivery, is essential for addressing the widening disparities in tobacco-related cancer incidence and mortality^{1, 2} and tobacco treatment counseling¹³ especially for individuals living in neighborhoods with greater socioeconomic deprivation.

National cross-sectional survey data show that between 2000 and 2017, approximately 24% of individuals diagnosed with cancer smoked at the time of first cancer diagnosis¹⁴ and 12% to 56% of these individuals continued to smoke.^{15, 16} This is problematic given that increases in all-cause mortality, cancer-specific mortality, and the development of secondary primary cancers have been reported in patients with a cancer diagnosis and cancer survivors who smoke.³ Moreover, continuing to smoke after a cancer diagnosis can result in treatment complications, diminished quality of life,^{17, 18} and reduced treatment efficacy.^19-21 Behavioral and pharmacologic cessation treatments improve prognosis and survival for individuals with cancer diagnoses,^{1, 3} and clinicians play a vital role in promoting smoking cessation treatment.²²

Despite recommendations for routine assessment and treatment of tobacco use by national oncology professional organizations such as the National Comprehensive Cancer Network,²³ the American Society of Clinical Oncology (ASCO),²⁴ and the National Cancer Institute (NCI),²⁵ and recommendations for assessment of tobacco use in clinical cancer research,²⁶ the rates of tobacco assessment and treatment in oncology care are inconsistent and integration of smoking cessation services into cancer care is suboptimal.^27-29 Patients diagnosed with cancer who are smoking often do not get cessation support during treatment.^30-35 Population-based data show that only half of individuals diagnosed with cancer who smoke report receiving counseling to quit.³⁶ Further, data on smoking history and status particularly for nontobacco associated cancers are often not systematically collected in cancer clinical trials.³⁷ This lack of attention to smoking cessation in clinical oncology can have negative implications for cancer treatment and clinical care outcomes for patients participating in clinical trials.

Neighborhood is an important determinant of health that intersects several social factors to influence one’s exposure to health risks and illness, access to care, and differential health outcomes. Neighborhood characteristics including physical and social attributes affect health care utilization and health care delivery, but the impact of “place” and its associated effect on smoking status and cessation indicators in cancer care delivery has not been widely evaluated. Specifically, the relationship between social determinants of health using neighborhood level measures such as the Area Deprivation Index (ADI)³⁸ and tobacco use and cessation indicators in patients with a cancer diagnosis enrolled in clinical trials has not been explored. Further, whereas tobacco use is a modifiable risk factor for adverse outcomes among patients diagnosed with cancer, knowledge of the scope and patterns of tobacco use among patients diagnosed with cancer is limited, including among patients enrolled in cancer clinical trials. Tobacco assessment during clinical trials can help increase patients’ knowledge about the effects of tobacco use and exposure on treatment toxicity, physical and psychological symptoms. Evidence-based behavioral and pharmacological treatment of tobacco dependence can improve cancer care outcomes. Clinical trial features, such as the repeated engagement of the patient with the clinical trial team and other providers, have the potential to promote dialogue about smoking status and linkages to cessation supports. To this end, we sought to describe patterns of tobacco use and receipt of tobacco assistance, and cessation outcomes in relation to neighborhood socioeconomic status (as measured by ADI) among adult cancer patients participating in selected phase 2 and phase 3 clinical trials led by the ECOG-ACRIN Cancer Research Group (EA). Understanding tobacco use and cancer patients’ reports of their oncology providers’ assistance among patients with cancer can inform where and how best to integrate tobacco use assessment and treatment in clinical trials. This work is aligned with the priorities outlined by NCI’s Cancer Moonshot-funded Cancer Center Cessation Initiative (C3I)^{25, 39} to ensure that tobacco cessation is integrated into cancer care delivery and that disparities are addressed in the tobacco use assessment and receipt of evidence-based treatment.

METHODS

This study is a secondary analysis of baseline data from the tobacco use assessment ancillary study that evaluated tobacco use and smoking behaviors among patients with a cancer diagnosis. The ancillary study was embedded into 10 parent protocols (nine therapeutic trials and one imaging trial) actively enrolling patients at the time of the study and patients were consented as part of the parent trials (ECOG-ACRIN trials IDs E1A11, EA1131, EA3163, EA4151, EA6134, EA6141, EA8153, EA8171, EA9161, and EA5152). Enrollment for the ancillary study began in June 2017 and patient accrual closed in October 2021. One trial (EA5152) terminated accrual early, and no patients participated in the ancillary study. The trials included multiple cancer types, treatment regimens, disease stage, and participant gender. Patients with breast, head and neck, leukemia, lymphoma, melanoma, myeloma, and prostate cancers completed the modified Cancer Patient Tobacco Use Questionnaire (C-TUQ developed by NCI and the American Association for Cancer Research [NCI-AACR] Cancer Patient Tobacco Use Assessment Task Force for cancer patients and cancer survivors)³⁷ at enrollment (in the parent trials) and 3 and 6 months’ follow-up (after trial enrollment).

The research was approved by the NCI Central Institutional Review Board and local institutional review boards at the participating sites if applicable. The objectives of this cross-sectional analysis were to: (1) describe sociodemographic factors, patterns of tobacco use assessment, tobacco assistance, and cessation indicators; (2) examine the association between tobacco use assessment, tobacco assistance, cessation indicators, and neighborhood socioeconomic disadvantage (ADI) among patients with varied cancers enrolled in clinical trials of the ECOG-ACRIN Cancer Research Group.

MEASURES

Sociodemographic

Patients’ sociodemographic characteristics, including sex, age at study entry, race, ethnicity, and cancer type were extracted from the ECOG-ACRIN study record for each trial. Patients completed a self-reported questionnaire using the web-based EA System for Easy Entry of Patient Reported Outcomes (EASEE-PRO) to indicate marital status and educational attainment.

Smoking and smoking cessation

Using EASEE-PRO, patients completed a selected set of items from a modified C-TUQ, a 22-item self-report survey designed to capture information about tobacco use. The C-TUQ tool is divided into five domains and includes a core (short form of four items) and an extension (set of optional items). It is publicly available from the NCI,¹¹ and in the Supplemental Materials S1, for use.

The baseline assessment for the present study included C-TUQ items age at first cigarette use; total years smoked; average number of cigarettes smoked per day; and number of days, weeks, months, or years since last cigarette use; smoking cigarettes in the past 30 days (yes/no); number of days smoked cigarettes in past 30 days; and quit attempts in the past 30 days (yes/no). Current, former, and never smoking status variables were derived from multiple questions, as shown in Figure 1.

FIGURE 1

Smoking status of patients enrolled in nine ECOG-ACRIN clinical trials. Smoking status ascertained from tobacco use assessment baseline questionnaire. “Have you smoked at least 100 cigarettes (5 packs = 100 cigarettes) in your entire life?” distinguished patients who had never smoked "no" from those who had smoked at least 100 cigarettes in their lifetime “yes.” Former smoking status comprised of three categories “<1 year since last cigarette,” “>1 year since last cigarette,” and “don’t know/don’t remember.” Current smoking status comprised of “at least one cigarette puff today,” “1-7 days since last cigarette,” and “<1 month since last cigarette.”

The questionnaire also assessed patients’ reported receipt of 5 As brief model of cessation support. Specifically, seven items asked “Since you were first told you had cancer, has your cancer physician or nurse done any of the following?”: “Asked you about smoking cigarettes,” “Advised you to quit smoking cigarettes,” “Asked you about your interest in quitting smoking cigarettes,” “Talked with you about how to quit smoking cigarettes,” “Recommended counseling (classes, quit line) to help you quit smoking cigarettes,” “Recommended using nicotine replacement therapy (patch, gum, inhaler, lozenge, or spray), other cessation methods such as bupropion (Wellbutrin, Zyban), and/or varenicline (Chantix) to help you quit smoking cigarettes,” and “Suggested a follow-up visit or phone call to check in on your quitting smoking cigarettes.”

Neighborhood socioeconomic status

The national ADI, a factor-based index based on patient’s census block group, was used to ascertain neighborhood level socioeconomic status.³⁸ The ADI, an aggregate geographic area-based measure of socioeconomic position, is calculated at the census block group level using 17 U.S. census-based indicators in four key domains (poverty, education, housing quality, and employment) to rank the socioeconomic context of neighborhoods.⁴⁰ A patient’s “neighborhood” (census block group) was obtained using the U.S. census geocoder (by providing patient’s full address or the 9-digit ZIP code),⁴¹ and the corresponding ADI was obtained from the Neighborhood Atlas v3.^{38, 42}

STATISTICAL ANALYSIS

Descriptive analyses were used to calculate frequency (with proportions) and central tendency statistics for patient characteristics and selected C-TUQ items. The association of patient’s ADI (classified into tertiles: high, intermediate, and low neighborhood deprivation) with smoking status (never, former, and current smokers) was evaluated, among all patients, using multinomial logistic regression with smoking status as the outcome variable. The associations of ADI with quit attempt in the past 30 days and the seven items indicating receipt of the 5 As were evaluated, separately, using binomial logistic regression with the probability of quit attempt and 5 A receipt (coded as “yes”) being modeled as the outcome variables. Two of the outcome variables, quit attempt in the past 30 days and asked about smoking (one of the 5 A receipt indicators), were analyzed based on all patients who reported former and current smoking status. For the remaining 5 A items, the analysis was conducted using data from a subset of patients who reported smoking within the past year. All association tests were performed using multivariable models adjusting for age (as a continuous variable), sex (male vs. female), race (White vs. other), marriage status (married/living as married vs other), and education level (≤high school vs >high school but ≤4-year college vs. >4-year college). All tests were two-sided with significance level of 0.05 for the evaluation of ADI with smoking status and quit attempt separately. Because of multiple tests in patients’ reported receipt of 5 As (n = 7), Bonferroni correction was applied with family-wise error rate controlled at 0.20 for 5 As testing (i.e., each test with significance level of 0.0286 [=0.20/7]). Analyses were conducted using SAS 9.4.

RESULTS

Patient characteristics

Patient characteristics are shown in Table 1. A total of 756 patients completed the baseline survey. Smoking status for 16 patients could not be determined; these patients were excluded from the analysis. Among the 740 patients whose smoking status could be determined, 402 (54%) were classified as individuals who never smoked, 257 (35%) reported smoking in the past, and 81 (11%) were patients who reported currently smoking. The mean age at study entry was 58.8 (SD, 9.0) years, mean time since cancer diagnosis was 25.7 (SD, 40.0) months, and the mean national ADI was 42.4 (SD, 25.6). Patients were mostly male (69.7%), White (93.9%), non-Hispanic (97.4%), and reported being married/living as married (78.9%). Most patients (54.4%) had completed higher education, with 27.5% completing 4 years of college and 26.9% beyond 4 year of college. Few patients had less than or some high school education or no high school diploma (2.4%) or reported being a high school graduate or equivalent (12.9%), whereas almost one third (30.3%) had some (<4-year) college/technical/vocational/associates degree. The cancer disease sites were leukemia (35.5%), lymphoma (19.1%), prostate (17.7%), breast (10.7%), melanoma (9.3%), myeloma (7.2%), and head and neck (0.5%).

TABLE 1. Patient demography and cancer type at study entry.

	Smoking status at baseline
	Current (N = 81)		Former (N = 257)		Never (N = 402)		Total (N = 740)
	Mean (SD) or N (%)		Mean (SD) or N (%)		Mean (SD) or N (%)		Mean (SD) or N (%)
Age at study entry (in years) (mean [SD])	57.2 (9.2)		59.4 (9.1)		58.7 (8.8)		58.8 (9.0)
Time since diagnosis (in months, mean [SD])^a	26.3 (37.6)		27.1 (40.3)		24.7 (40.3)		25.7 (40.0)
National Area Deprivation Index (mean [SD])^b	56.0 (24.8)		44.3 (26.5)		38. 2 (24.0)		42.4 (25.6)
Sex
Male	55	67.9	188	73.2	273	67.9	516	69.7
Female	26	32.1	69	26.8	129	32.1	224	30.3
Race^c
White	71	91.0	236	95.2	358	93.7	665	93.9
African American	6	7.7	11	4.4	17	4.4	34	4.8
Asian	0	0.0	1	0.4	6	1.6	7	1.0
Multirace	1	1.3	0	0.0	1	0.3	2	0.3
Ethnicity^d
Hispanic	0	0.0	7	2.8	12	3.1	19	2.6
Non-Hispanic	76	100.0	246	97.2	378	96.9	700	97.4
Marital status^e
Single, never married	10	12.3	12	4.7	25	6.2	47	6.3
Married/living as married	46	56.8	212	82.8	325	80.9	583	78.9
Widowed	5	6.2	6	2.3	6	1.5	17	2.3
Divorced/separated	20	24.7	26	10.2	46	11.4	92	12.5
Education attainment^f
Less than high school; some high school/no diploma	5	6.2	6	2.3	7	1.7	18	2.4
High school graduate or equivalent	24	30.0	33	12.8	38	9.5	95	12.9
Some college, no degree; trade/technical/vocational training; associates degree	35	43.7	106	41.3	83	20.6	224	30.3
Completed 4-year college degree	9	11.3	63	24.5	131	32.6	203	27.5
Beyond 4-year college degree	7	8.8	49	19.1	143	35.6	199	26.9
Cancer type
Leukemia	27	33.3	89	34.6	147	36.6	263	35.5
Lymphoma	15	18.5	57	22.2	69	17.2	141	19.1
Prostate	11	13.6	40	15.6	80	19.9	131	17.7
Breast	10	12.3	26	10.1	43	10.7	79	10.7
Melanoma	13	16.0	23	8.9	33	8.2	69	9.3
Myeloma	3	3.7	22	8.6	28	7.0	53	7.2
Head and neck	2	2.5	0	0.0	2	0.5	4	0.5

• Note: This table was based on patients completing the baseline survey and with definite smoking status reported.
• ^a Unknown/missing for 166 patients (19, 63, and 84 for current, former, and never smokers, respectively) because data were not collected in all the parent trials.
• ^b Unknown/missing for 60 patients (four, 16, and 40 for current, former, and never smokers, respectively). National ADI scores range from 1 to 100 (higher scores = higher deprivation/disadvantage).
• ^c Unknown/missing for 32 patients (three, nine, and 20 for current, former, and never smokers, respectively).
• ^d Unknown/missing for 21 patients (five, four, and 12 for current, former, and never smokers, respectively).
• ^e Unknown/missing for one patient (one former smoker).
• ^f Unknown/missing for one patient (one current smoker).

Smoking related variables and receipt of 5 As since cancer diagnosis by smoking status

Table 2 illustrates smoking-related variables and receipt of the 5 As since cancer diagnosis by smoking status. The mean age of first smoking regularly was 16.6 (SD, 6.5) years with an average of 20.7 (SD, 15.6) years smoked, and patients reported an average of 14.7 (SD, 10.9) cigarettes smoked per day. Among the 81 patients who reported currently smoking, the mean number of smoking days in the past 30 days was 24.4 (SD, 8.8), and the majority (67.5%) had attempted to quit in the past 30 days. Among patients who reported current or former smoking status, most (86.7%) reported being asked about smoking since their cancer diagnosis. For patients reporting current smoking status, most had been advised to quit smoking (80.0%) and asked about interest in quitting smoking (76.3%). Almost half of patients reporting currently smoking had not received a recommendation of counseling to help quit smoking (46.9%) and 42.0% had not received a recommendation to use nicotine replacement therapy or other medications to help quit smoking. The majority (76.3%) had not received a suggestion of a follow-up visit or phone call to check in on quitting smoking.

TABLE 2. Smoking related variables and receipt of 5 As since cancer diagnosis by smoking status.

	Current (N = 81)	Former (N = 257)	Total (N = 338)
	Mean (SD) or N (%)	Mean (SD) or N (%)	Mean (SD) or N (%)
Age first began smoking regularly (mean [SD])^a,d, ^a,d	16.9 (6.2)	16.5 (6.6)	16.6 (6.5)
Total years smoked (mean [SD])	32.7 (15.0)	16.9 (13.9)	20.7 (15.6)
No. of cigarettes per day (mean [SD])	17.1 (9.9)	13.9 (11.2)	14.7 (10.9)
No. of days smoking in the past 30 days (mean [SD])^b	24.4 (8.8)	NA	NA
Quit attempt in the past 30 days^c
Yes	54 (67.5%)	42 (16.7%)	96 (28.9%)
No	26 (32.5%)	210 (83.3%)	236 (71.1%)
Receipt of 5 As since cancer diagnosis^d
Asked about smoking
Yes	74 (91.4%)	218 (85.2%)	292 (86.7%)
No	7 (8.6%)	38 (14.8%)	45 (13.3%)
Advised to quit smoking
Yes	64 (80.0%)	15 (5.9%)	79 (23.6%)
No	12 (15.0%)	10 (3.9%)	22 (6.6%)
NA/no cigarettes since diagnosis	4 (5.0%)	230 (90.2%)	234 (69.8%)
Asked about interest in quitting smoking
Yes	61 (76.3%)	11 (4.3%)	72 (21.5%)
No	15 (18.7%)	5 (2.0%)	20 (6.0%)
NA/no cigarettes since diagnosis	4 (5.0%)	238 (93.7%)	242 (72.5%)
Talked about how to quit smoking
Yes	43 (53.1%)	7 (2.8%)	50 (15.0%)
No	35 (43.2%)	8 (3.2%)	43 (12.9%)
NA/no cigarettes since diagnosis	3 (3.7%)	237 (94.0%)	240 (72.1%)
Recommend counseling to help quit smoking
Yes	40 (49.4%)	5 (2.0%)	45 (13.4%)
No	38 (46.9%)	13 (5.1%)	51 (15.2%)
NA/no cigarettes since diagnosis	3 (3.7%)	237 (92.9%)	240 (71.4%)
Recommend using nicotine replacement therapy or other medications to help quit smoking
Yes	43 (53.1%)	7 (2.7%)	50 (14.8%)
No	34 (42.0%)	10 (3.9%)	44 (13.1%)
NA/no cigarettes since diagnosis	4 (4.9%)	239 (93.4%)	243 (72.1%)
Suggested a follow-up visit or phone call to check in on quitting smoking
Yes	13 (16.2%)	1 (0.4%)	14 (4.2%)
No	61 (76.3%)	14 (5.5%)	75 (22.4%)
NA/no cigarettes since diagnosis	6 (7.5%)	240 (94.1%)	246 (72.4%)

• Abbreviations: 5As, Ask, Advise, Assess, Assist, and Arrange follow-up; NA, not available.
• ^a Unknown/missing for one former smoker.
• ^b Unknown/missing for five current smokers.
• ^c Unknown/missing for six patients (one current smoker and five former smokers).
• ^d Patient with unknown/missing response were not presented.

Smoking status, receipt of 5 As since the diagnosis of cancer, quit attempts in past 30 days by ADI and the association of ADI with smoking status, being asked about smoking (assessment), and being recommended counseling to help quit (assistance)

Table 3 shows the distribution of smoking status, 5 As and quit attempts by the category of ADI, and the association of ADI with smoking status, being asked about smoking (assessment), and being recommended counseling to help quit (assistance). A significantly higher proportion of people who currently smoke was observed in the high disadvantaged geographic areas relative to low (18.6% vs 5.3%; p < .001), and in the intermediate areas relative to low (10.1% vs 5.3%; p = .039). We evaluated quit attempts in the past 30 days and being asked about smoking among patients reporting former and current smoking status by ADI. The majority of patients in geographic areas with high ADI tertile (93.7%) were asked about smoking compared with those reflecting intermediate (85.7%) and low (79.1%) ADIs. We then evaluated the remaining 5 As among patients who reported smoking within the past year (i.e., ≤1 year) by ADI. Patients in geographic areas with low ADI (i.e., least disadvantaged were more likely to be advised to quit [100%]) compared with those with intermediate (86.4%) and high (77.8%) ADIs. Similarly, more patients in geographic areas with low ADI were asked about interest in quitting smoking (92.9%) compared with those in intermediate (81.8%) and high (79.1%) ADI geographic areas. Only half (50.0%) of patients in high ADI geographic areas received advice about quitting smoking compared with 68.2% and 66.7% in intermediate and low ADI geographic areas, respectively. In addition, almost half of all patients who identified as smoking within the past year were not recommended counseling to help quit (48.1%), and more were in geographic areas with a high ADI (59.1%) followed by intermediate (45.5%) and low (20.0%) ADIs. Nicotine replacement therapy or other medications to help quit smoking was recommended to the majority of currently smoking patients in high (61.9%), intermediate (59.1%), and low (53.3%) ADI geographic areas. Providers did not suggest a follow-up visit or phone call to check in on quitting smoking for many of the patients who reported smoking within the past year (82.9%), with a high proportion of these patients being in geographic areas with high a ADI (85.0%) followed by intermediate (81.0%) and low (80.0%) ADI geographic areas. Patients in high ADI geographic areas made more attempts to quit in the past 30 days (33.1%) followed by those in low (27.7%) and intermediate (22.9%) ADI ranked geographic areas.

TABLE 3. Smoking status, quit attempts in past 30 days, receipt of 5 As since the diagnosis of cancer, by area deprivation index (ADI) and association of ADI with smoking status, being asked about smoking, and being recommended counseling to help quit.

	ADI national rank					ADI: high/low		ADI: intermediate/low
	High (N = 226)	Intermediate (N = 228)	Low (N = 226)	Total (N = 680)
Outcome	N (%)	N (%)	N (%)	N (%)	p (overall) and N	OR (95% CI)	p	OR (95% CI)	p
Among all patients with ADI available
Smoking status					.0145*
Current	42 (18.6, 54.5)	23 (10.1, 29.9)	12 (5.3, 15.6)	77 (11.3)	678	3.57 (1.69–7.54)	.0009	2.02 (0.92–4.45)	.0800
Former	85 (37.6, 35.3)	82 (36.0, 34.0)	74 (32.7, 30.7)	241 (35.5)		1.42 (0.93–2.16)	.1033	1.20 (0.80–1.80)	.3827
Never	99 (43.8, 27.3)	123 (53.9, 34.0)	140 (62.0, 38.7)	362 (53.2)

	High (N = 127)	Intermediate (N = 105)	Low (N = 86)	Total (N = 318)
	N (%)	N (%)	N (%)	N (%)
Among patients who reported former and current smoking status and with ADI available
Quit attempt in the past 30 days^a					.3855
Yes	41 (33.1, 46.6)	24 (22.9, 27.3)	23 (27.7, 26.1)	88 (28.2)	311
No	83 (66.9, 30.0)	81 (77.1, 36.2)	60 (72.3, 26.8)	224 (71.8)
Asked about smoking^b					.0117*	4.24 (1.64–10.98)	.0029	1.65 (0.73–3.71)	.2300
Yes	118 (93.7, 42.8)	90 (85.7, 32.6)	68 (79.1, 24.6)	276 (87.1)	315
No	8 (6.3, 19.5)	15 (14.3, 36.6)	18 (20.9, 43.9)	41 (12.9)

	High (N = 51)	Intermediate (N = 26)	Low (N = 19)	Total (N = 96)
	N (%)	N (%)	N (%)	N (%)
Among patients who reported smoking within the past year and with ADI available
Advised to quit smoking^c					.4907
Yes	35 (77.8, 50.7)	19 (86.4, 27.6)	15 (100.0, 21.7)	69 (84.1%)	82
No	10 (22.2, 76.9)	3 (13.6, 23.1)	0 (0.0, 0.0)	13 (15.9%)
Asked about interest in quitting smoking^d					.4576
Yes	34 (79.1, 52.3)	18 (81.8, 27.7)	13 (92.9, 20.0)	65 (82.3)	78
No	9 (20.9, 64.3)	4 (18.2, 28.6)	1 (7.1, 7.1)	14 (17.7)
Talked about how to quit smoking^e					.2166
Yes	22 (50.0, 46.8)	15 (68.2, 31.9)	10 (66.7, 21.3)	47 (58.0)	80
No	22 (50.0, 64.7)	7 (31.8, 20.6)	5 (33.3, 14.7)	34 (42.0)
Recommend counseling to help quit smoking^f					.0225*	0.11 (0.02–0.58)	.0086	0.28 (0.05–1.54)	.1441
Yes	18 (40.9, 42.8)	12 (54.5, 28.6)	12 (80.0, 28.6)	42 (51.9)	80
No	26 (59.1, 66.7)	10 (45.5, 25.6)	3 (20.0, 7.7)	39 (48.1)
Recommend using nicotine replacement therapy or other medications to help quit smoking^g					.8958
Yes	26 (61.9, 55.3)	13 (59.1, 27.7)	8 (53.3, 17.0)	47 (59.5)	78
No	16 (38.1, 50.0)	9 (40.9, 28.1)	7 (46.7, 21.9)	32 (40.5)
Suggested a follow-up visit or phone call to check in on quitting smoking^h					.8230
Yes	6 (15.0, 46.1)	4 (19.0, 30.8)	3 (20.0, 23.1)	13 (17.1)	75
No	34 (85.0, 54.0)	17 (81.0, 27.0)	12 (80.0, 19.0)	63 (82.9)

• Note: For percent figures in parentheses of each cell, the first one refers to column percent and the second one refers to the row percent. Multivariable models were fitted for each outcome adjusting for age, sex (male vs. female), race (White vs. other), marital status (married/living as married vs other), and education level (≤high school vs >high school but ≤4-year college vs. >4-year college). Patients with the not applicable and missing responses in the outcome variable, and with missing values in ADI and any of the covariates were excluded in the analysis.
• ^a n = 6; did not complete this item (unknown/missing); data are not presented.
• ^b n = 1; did not complete this item (unknown/missing); data are not presented.
• ^c n = 1; did not complete this item (unknown/missing) and n = 13; not applicable/no cigarettes since diagnosis; data are not presented.
• ^d n = 1; did not complete this item (unknown/missing) and n = 16; not applicable/no cigarettes since diagnosis; data are not presented.
• ^e n = 15; not applicable/no cigarettes since diagnosis; data are not presented.
• ^f n = 15; not applicable/no cigarettes since diagnosis; data are not presented.
• ^g n = 17; not applicable/no cigarettes since diagnosis; data are not presented.
• ^h n = 1; did not complete this item (unknown/missing) and n = 19; not applicable/no cigarettes since diagnosis; data are not presented.
• *Statistically significant with Bonferroni correction (α = 0.20 for seven tests on the 5 As).

Multivariable analysis adjusting for patient sociodemographic characteristics indicate that ADI was significantly associated with patient’s smoking status (p = .0145). Patients in a high ADI geographical area were three to four times more likely to report currently smoking than patients in a low ADI geographic area (OR, 3.57; 95% CI, 1.69–7.54; p = .0009). For patients who reported former and current smoking status, being asked about smoking was associated with ADI (p = .0117). The odds of a patient in a geographic area with high ADI being asked about smoking was four times that of a patient in a low ADI geographic area (least disadvantaged) (OR, 4.24; 95% CI, 1.64–10.98; p = .0029). Among patients who reported smoking within the past year, being recommended counseling to help quit smoking was associated with ADI (p = .0225). The odds of a patient with high ADI being recommended counseling to help quit smoking was one tenth of a patient with low ADI (OR, 0.11; 95% CI, 0.02–0.58; p = .0086). Patient’s report of being advised to quit smoking, asked about interest in quitting smoking, talking about how to quit smoking, being recommended use of nicotine replacement therapy or other medications to help quit smoking, being suggested a follow-up visit or phone call to check in on quitting smoking, and quit attempts in the past 30 days were not significantly associated with ADI geographic rank.

DISCUSSION

Despite guidelines for standardized assessment and routine treatment of tobacco use in cancer care,^{23, 26} our findings demonstrate that gaps in implementation and disparities exist by ADI in assessing smoking and providing cessation support among adults diagnosed with cancer participating in phase 2 and phase 3 therapeutic (n = 8) and imaging (n = 1) trials led by the EA. In adjusted multivariable analysis, being in a geographic area with high ADI (most disadvantaged) was associated with higher current smoking, higher assessment of smoking (i.e., being asked about smoking), but lower odds of being recommended for counseling to help quit smoking. The unequal distribution of social determinants inherent in the ADI measure (i.e., education, income/employment, housing quality, and household characteristics) has the greatest impact on cancer risk factors such as smoking and access to high quality tobacco treatment.⁴³ Our findings indicate disparities in smoking and cessation support among patients diagnosed with cancer from high disadvantaged geographic areas. Specifically, despite patients in geographic areas with greater disadvantage being more likely to report current smoking, and provider assessment of their tobacco use, they are less likely to be recommended counseling to help quit when compared to their counterparts in less disadvantaged areas. The accumulation of life experiences with structural disadvantages in the four domains of the ADI (poverty, education, housing quality, and employment), means that patients in neighborhoods with greater disadvantage face unique challenges obtaining optimal cessation support even when participating in clinical trials.

Patients diagnosed with cancer might have better treatment outcomes (e.g., treatment toxicity, symptom burden, overall quality of life) with improved cessation support, especially for patients in highly disadvantaged geographic areas. Cancer care providers, particularly those working in high ADI settings, need to increase routine delivery of all of the components of 5 As, consistent with an evidence-based approach to treating tobacco use. Research shows that for patients receiving cancer treatment, the final 2 As (Assist and Arrange) are pivotal for promoting smoking cessation.²² Patients enrolled in clinical trials have repeated interactions with providers and thus there are several windows of opportunity for health professionals to deliver the 5 As or briefer models of intervention such as the Ask-Advise-Refer (AAR)⁴⁴ and Ask-Advice-Connect (AAC)⁴⁵ to facilitate linkages to tobacco treatment resources. Implementing strategies to improve adoption of clinical practice guidelines for treating tobacco use is needed, particularly for patients in high ADI geographic areas. Our study findings are consistent with other cancer-related research indicating that high ADI is associated with poor health outcomes such as worse symptoms of anxiety,⁴⁶ lung cancer prevalence and mortality,⁴⁷ and suboptimal quality of care such as lower likelihood of receiving postoperative (adjuvant) therapy for pancreatic cancer,⁴⁸ and lower rates of physician referral to screening for breast, cervical, and colorectal cancers.⁴⁹

LIMITATIONS

Although this study contributes to the knowledge base of social determinants of health, smoking status and tobacco treatment delivery among adults participating in cancer-related clinical trials, there are important methodological limitations. First, the sample size is modest and not racially and ethnically diverse. Second, the ADI metric is a composite indicator using patient ZIP code and does not allow for the specificity of analysis offered at individual level. It should be noted that physical attributes of a neighborhood such as transportation, pharmacy, and hospital density, etc., which are robustly measured at the census level and important contributors to health care utilization are currently uncaptured. Third, our survey did not assess whether patients were connected to a counselor or provided a prescription for a smoking cessation medication. Fourth, our ability to make inferences about causation and to generalize study findings beyond these practice settings are limited. Future research should examine barriers and facilitators of implementation of evidence-based tobacco treatment guidelines in oncology practices at the patient, provider, system, and policy levels.

CONCLUSIONS

This work has implications for oncology clinical practice and clinical trial design. People with a cancer diagnosis are more likely to report being asked about their tobacco use, yet less likely to have counseling to support quitting recommended when they reside in a more disadvantaged geographic area. Patients in high ADI geographic areas have compounding social determinant of health needs that can impact their smoking cessation success, engagement in a clinical trial, and cancer care–related outcomes. The observed disparities in access to tobacco treatment among patients in high ADI geographic areas participating in clinical trials demonstrate the heightened need to provide comprehensive tobacco use assessment and treatment (5 As model) with an emphasis on assisting and arranging treatment to improve health outcomes. Briefer models consisting of the 3As (Ask, Advise, Refer/Connect) should be considered in resource limited health care settings where other health care professionals are able to refer and connect patients to tobacco treatment. Future research should examine uptake of briefer models of intervention such as Ask-Advise-Refer or Ask-Advice-Connect in community oncology and whether these models yield improved linkages to tobacco treatment, reduce disparities in tobacco treatment access, and improve cancer care outcomes for patients in high ADI geographic areas participating in clinical trials.

With the provision of cessation support in clinical trials, we may be able to mitigate these effects by adopting the American Society of Clinical Oncology/American Association for Cancer Research recommendations that tobacco assessment occur at study entry and, at a minimum, at the end of treatment protocol,²⁶ and that the 5 As comprehensive model of tobacco treatment be offered as part of all clinical trials. Broad adoption of a standardized tobacco use assessment, such as the C-TUQ, will help advance knowledge regarding the clinical significance of persistent tobacco use and cessation for cancer patients. Providers in oncology practice should integrate evidence-based guidelines for addressing tobacco dependence into their clinical practice, including recommending counseling to help quit, assisting, and arranging follow-up to eliminate tobacco related disparities and achieve health equity in treatment outcomes and oncology clinical trials.

AUTHOR CONTRIBUTIONS

Angela Wangari Walter: Conceptualization, Methodology, Visualization, Writing - original draft, and Writing - review & editing. Ju-Whei Lee: Conceptualization, Formal analysis, Methodology, Validation, Visualization, Writing - original draft, and Writing - review & editing. Joanna M. Streck: Conceptualization, Methodology, Visualization, Writing - original draft, and Writing - review & editing. Ilana F. Gareen: Conceptualization, funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, and Writing - review & editing. Benjamin A. Herman: Conceptualization, Data curation, Methodology, Project administration, Software, Validation, Visualization, Writing - original draft, and Writing - review & editing. Sheetal M. Kircher: Investigation and Writing - review & editing. Ruth C. Carlos: Writing - review & editing. Shaji K. Kumar: Investigation and Writing - review & editing. Ingrid A. Mayer: Investigation and Writing - review & editing. Nabil F. Saba: Investigation and Writing - review & editing. Timothy S. Fenske: Investigation and Writing - review & editing. Joel W. Neal: Investigation and Writing - review & editing. Michael B. Atkins: Investigation and Writing - review & editing. Frank S. Hodi: Investigation and Writing - review & editing. Christos E. Kyriakopoulos: Investigation and Writing - review & editing. Clare M. Tempany-Afdhal: Investigation and Writing - review & editing. Tait D. Shanafelt: Investigation and Writing - review & editing. Lynne I. Wagner: Funding acquisition, Investigation, Resources, and Writing - review & editing. Stephanie R. Land: Writing - review & editing. Jamie S. Ostroff: Conceptualization, funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Writing - original draft, and Writing - review & editing. Elyse R. Park: Conceptualization, funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Writing - original draft, and Writing - review & editing.

ACKNOWLEDGMENTS

This study was conducted by the ECOG-ACRIN Cancer Research Group (Peter J. O'Dwyer, MD, and Mitchell D. Schnall, MD, PhD, Group Co-Chairs) and supported by the National Cancer Institute of the National Institutes of Health under the following award numbers: U10CA180794, UG1CA189828, U10CA180820, UG1CA232760, UG1CA233180, UG1CA233198, UG1CA233247, UG1CA233270, UG1CA233277, UG1CA233290, UG1CA233320, UG1CA239758. Dr. Ostroff’s contribution was supported by P30 CA008748 (Thompson) and additional funding for Dr. Streck was provided by NIDA K12 DA043490 (Rigotti/Evins). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

CONFLICT OF INTEREST STATEMENT

Ruth C. Carlos reports receiving salary support from JACR as editor-in-chief. Lynne I. Wagner reports consulting on patient-reported outcomes design, analysis, and interpretation at Celgene/Bristol-Myers Squibb and Athenex. Elyse R. Park reports receiving royalties from UpToDate, “Behavioral Approaches to Smoking Cessation.” Ingrid A. Mayer reports being currently employed at AstraZeneca. Shaji K. Kumar reports research funding for clinical trials to the institution from: AbbVie, Amgen, Allogene, AstraZeneca, BMS, Carsgen, GSK, Janssen, Novartis, Roche-Genentech, Takeda, Regeneron; consulting/advisory board participation: (with no personal payments) with AbbVie, Amgen, BMS, Janssen, Roche-Genentech, Takeda, AstraZeneca, Bluebird Bio, Secura Biotherapeutics, Trillium, Loxo Oncology, K36, Sanofi, ArcellX, and (with personal payment) Oncopeptides, Beigene, Antengene. Joel W. Neal reports research funding to the institution from: Genentech/Roche, Merck, Novartis, Boehringer Ingelheim, Exelixis, Nektar Therapeutics, Takeda Pharmaceuticals, Adaptimmune, GSK, Janssen, and AbbVie; consulting/advisory board participation at: AstraZeneca, Genentech/Roche, Exelixis, Jounce Therapeutics, Takeda Pharmaceuticals, Eli Lilly and Company, Calithera Biosciences, Amgen, Iovance Biotherapeutics, Blueprint Pharmaceuticals, Regeneron Pharmaceuticals, Natera, Sanofi/Regeneron, D2G Oncology, Surface Oncology, Turning Point Therapeutics, Mirati Therapeutics, Gilead Sciences, and AbbVie; receiving royalties from: UpToDate. Frank S. Hodi reports grants and personal fees from Bristol-Myers Squibb, personal fees from Merck, grants and personal fees from Novartis, personal fees from Surface, personal fees from Compass Therapeutics, personal fees from Apricity, personal fees from 7 Hills Pharma, personal fees from Bicara, other from Pieris Pharmacutical, personal fees from Checkpoint Therapeutics, personal fees from Bioentre, personal fees from Gossamer, personal fees from Iovance, personal fees from Catalym, personal fees from Immunocore, personal fees from Kairos, personal fees from Rheos, personal fees from Zumutor, personal fees from Corner Therapeuitcs, personal fees from Curis, personal fees from AstraZeneca, outside the submitted work; in addition, he has a patent Methods for Treating MICA-Related Disorders (#20100111973) with royalties paid, a patent Tumor antigens and uses thereof (#7250291) issued, a patent Angiopoiten-2 Biomarkers Predictive of Anti-immune checkpoint response (#20170248603) pending, a patent Compositions and Methods for Identification, Assessment, Prevention, and Treatment of Melanoma using PD-L1 Isoforms (#20160340407) pending, a patent Therapeutic peptides (#20160046716) pending, a patent Therapeutic Peptides (#20140004112) pending, a patent Therapeutic Peptides (#20170022275) pending, a patent Therapeutic Peptides (#20170008962) pending, a patent Therapeutic Peptides Therapeutic Peptides Patent number: 9402905 issued, a patent Methods of Using Pembrolizumab and Trebananib pending, a patent Vaccine compositions and methods for restoring NKG2D pathway function against cancers Patent number: 10279021 issued, a patent Antibodies that bind to MHC class I polypeptide-related sequence Patent number: 10106611 issued, and a patent Anti-Galectin Antibody Biomarkers Predictive of Anti-Immune Checkpoint and Anti-Angiogenesis Responses Publication number: 20170343552 pending. The other authors declare no conflicts of interest.

Open Research

DATA AVAILABILITY STATEMENT

Individual-level data associated with this article can be requested by contacting the ECOG-ACRIN Cancer Research Group.