Smoking Cessation Prevalence and Inequalities in the United States: 2014-2019

Abstract

Background

Smoking is a leading cause of premature death and health inequities in the United States.

Methods

We estimated cross-sectional prevalence of smoking cessation indicators among US adult recent smokers (n = 43 602) overall and by sociodemographic subgroups in the Current Population Survey Tobacco Use Supplement 2014-2015 and 2018-2019 timepoints. Respondents reported past-year quit smoking interest, attempts, sustained (successful) cessation for 6 or more months and use of e-cigarettes or behavioral or pharmacological cessation treatments to quit smoking.

Results

Past-year quit smoking attempts declined slightly from 2014-2015 (52.9%) to 2018-2019 (51.3%) overall. Quit interest (pooled = 77.1%) and sustained cessation (pooled = 7.5%) did not change across timepoints. Among smokers making past-year quit attempts, 34.4% reported using cessation treatments in 2018-2019, and using e-cigarettes to quit smoking declined from 2014-2015 (33.3%) to 2018-2019 (25.0%). Several non-White (vs White) racial and ethnic groups had higher prevalence of quit interest and attempts but lower prevalence of sustained cessation or using e-cigarettes or treatments to quit. Income, education, employment, and metropolitan residence were positively associated with sustained cessation. Sociodemographic inequalities in sustained cessation and most other outcomes did not change across timepoints.

Conclusions

Although about half of US adult smokers made past-year quit attempts from 2014 to 2019, only 7.5% reported sustained cessation, and most who made quit attempts did not report using cessation treatments. Sociodemographic inequalities in cessation were pervasive and not entirely correspondent with sociodemographic variation in motivation to quit. Smoking cessation prevalence and inequalities did not improve from 2014 to 2019. Encouraging quit attempts and equitable access to smoking cessation aids are public health priorities.

Smoking is a leading cause of premature death and health inequities in the United States (1). After decades of substantial reductions in US smoking prevalence through 2015, declines have since slowed (2), especially among certain racial and ethnic minority, socioeconomically disadvantaged, disabled, and rural subpopulations (1-3). Smoking prevalence trends and inequalities are driven, in part, by changes in smoking cessation.

Institutional and interpersonal racism and other social determinants can lead to inequalities in cessation outcomes via several mechanisms. Developing an interest in quitting smoking can be thwarted by targeted advertising and marketing of cigarette products that are more addictive (eg, menthol flavored) to communities of color (4). Frequent experience of interpersonal discrimination and other forms of stress further make quitting difficult (5,6). Limited access to high-quality health care also reduces the chances that smokers will be offered smoking cessation treatments or even be advised to quit by a physician (7). For these reasons, detailed characterization of the prevalence and sociodemographic inequalities in various indicators along the smoking cessation continuum (ie, quit smoking interest, attempts, success, and treatment utilization) can guide precision clinical and public health strategies for promoting equitable smoking cessation.

A study using National Health Interview Survey (NHIS) data found increases in quit attempts, success, and treatment utilization from 2000 to 2015 (8), a period coinciding with declining smoking prevalence (9). The NHIS analysis also identified substantial racial and ethnic and socioeconomic inequalities across various smoking cessation indicators in 2015 (8). It is unknown whether these trends have continued and if inequalities extend to rural subgroups, reported use of different cessation treatment types, and American Indians and other subgroups with prevalence that are too low to examine in NHIS. It is also unknown whether there have been changes in the prevalence and demographic makeup of US smokers who try to quit by switching to e-cigarettes, a product with substantially fewer toxins than cigarettes (10).

This study examined the prevalence and demographic inequalities in smoking cessation indicators during 2014-2019 among US adult smokers in the Tobacco Use Supplement of the Current Population Survey (TUS-CPS). As the largest nationally representative sample with detailed smoking cessation information, TUS-CPS permits the investigation of subgroups historically underrepresented in smoking cessation epidemiology.

Methods

Participants and Procedures

The TUS-CPS collects tobacco use information among noninstitutionalized US adults via in-person or telephone interviews every 3-4 years. This study used publicly available self-report data from past-year cigarette smokers pooled across 2014-2015 and 2018-2019 TUS-CPS survey timepoints and followed the American Association for Public Opinion Research reporting guideline. Because several cessation outcomes (eg, attempt to quit by switching to e-cigarettes) were not introduced until 2014-2015 and previous studies have already reported prior TUS-CPS results (11,12), we focused on 2014-2019 data here. At every timepoint, the CPS draws a new independent cross-sectional probability sample based on a stratified 2-stage sampling scheme; individual respondents participating in CPS more than once is highly unlikely (13). Person-level self-response rates among 2014-2015 and 2018-2019 TUS-eligible CPS respondents ranged from 53.1% to 58.9% (14,15). Participants provided informed consent. Given data were de-identified, the University of Nebraska Medical Center institutional review board determined this was not human subjects research.

Measures

Demographics Characteristics. Race and ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, non-Hispanic American Indian or Alaskan Native, non-Hispanic-Asian, other or multiple races), education (less than high school, high school graduate, some college [no degree], associate degree, bachelor degree, and graduate degree), income (<$25 000, $25 000-$49 999, $50 000-$74 999, $75 000-$99 999, and >$100 000), disability status (yes vs no), employment status (full-time, part-time, unemployed, and not in the workforce), and residence location (metropolitan vs rural) were assessed (see additional details about certain definitions in the Supplementary Methods, available online). Self-reported race and ethnicity was measured per US Census Bureau guidelines (16). Using self-reported race and ethnicity permits implications for policy and public health actions addressing disproportionate experience of racism and other social determinants of health not captured by other demographic indicators (17). Age (coded: 18-24, 25-34, 35-44, and 45 years or older) and sex (male vs female) were adjusted as covariates.

Combustible Cigarette Smoking Status. Among ever-users of 100 or more cigarettes, respondents were classified as current (now smokes “every day or some days”) or former (reportedly now smokes “not at all”) smokers based on previous work (18). Recent former smokers were defined as those indicating it had been less than 1 year since they completely quit smoking cigarettes (18).

Smoking Cessation-Related Outcomes and Corresponding Analytic Samples. Following previously published Centers for Disease Control and Prevention (CDC) methods (8,19), outcomes and analytic sample restrictions were as follows: 1) Interest in quitting: in the analytic sample of current smokers (n = 37 903), participants reported interest in quitting smoking on a 10-point scale (1 [not at all] to 10 [extremely]; >2 was coded positive for quit interest) (19). The raw continuous variable (range = 1-10) served as a secondary outcome. 2) Past-year quit attempts: the analytic sample included past-year smokers (n = 43 602). Quit attempts were coded “yes” among current or former smokers who quit smoking for 1 or more days or made a “serious” past-year quit attempt. 3) Past-year sustained (successful) smoking cessation: the analytic sample included past-year smokers of 2 or more years (n = 43 361) (8,19). Sustained cessation was coded “yes” for former smokers who quit 6-11 months ago and “no” for current smokers and recent quitters (<6 months ago). 4) Attempt to quit by switching to e-cigarettes: the analytic sample included smokers with 1 or more past-years quit attempt (n = 21 694). Respondents were asked, “The last time you tried to quit smoking in the past year” (current smokers) or “When you quit smoking completely, did you try to quit by switching to e-cigarettes?” (former smokers). Responses were binary (yes/no). 5) Reported smoking cessation treatment use: the analytic sample included 2018-2019 respondents with 1 or more past-years quit attempt (n = 9038) since limited cessation treatments were asked in the 2014-2015 survey timepoint. Respondents reported use of the following during their last cessation attempt: telephone helpline or quitline; internet or web-based program or tool, including smartphone apps and text messaging programs; nicotine patch, gum, lozenge, nasal spray, or inhaler; prescription medication (ie, varenicline or bupropion); one-on-one in-person counseling by a health professional; stop-smoking clinic, class, or support group. Responses were coded into a binary variable (>1 vs 0 treatments); supplementary analyses examined use of each treatment type. We also created a reported use of medication and counseling combination treatment as a secondary outcome (counseling or class and >1 pharmacotherapies, yes or no) (20).

Statistical Analysis

After descriptive analyses, we calculated weighted percentages (95% confidence intervals [CIs]) overall and stratified by demographic subgroup using pooled 2014-2015 and 2018-2019 data. Next, we tested multivariable logistic regression models for each outcome including all demographic variables and timepoint as simultaneous regressors in pooled 2014-2015 and 2018-2019 data. Then, we calculated changes across timepoints (2018-2019 vs 2014-2015) in weighted percentages (95% CIs) for each outcome with unadjusted Rao-Scott χ² tests in the overall sample and each demographic subgroup. Unadjusted demographic variable x timepoint omnibus interactions tested for widening or narrowing of disparities across time in logistic regressions for outcomes measured at both timepoints.

We used SAS 9.4 (Cary, NC)  with sampling weights designed to be representative of the US population and using replicate weights and the balanced repeated replication method to address the complex survey design (15). Missing outcome data were rare (see Supplementary Methods, available online), permitting listwise deletion. All statistical tests were 2-sided, and a P value less than .05 was considered statistically significant.

Results

Study Sample

Analytic samples varied by study outcome (see Methods). The past-year smoker sample for the quit attempts outcome was 46.1% female, varied by age (18-24 years = 10.9%; 25-44 years = 38.8%; 45-64 years = 39.0%; 65 years or older = 11.3%) and race and ethnicity (American Indian and Alaska Native = 1.3%; Asian = 2.7%; Black = 12.1%; Hispanic = 10.8%; White = 70.7%; Other = 2.4%), and included 15.4% high school nongraduates, 25.6% with annual incomes less than $25 000, 18.6% with disabilities, 6.0% with unemployment, and 19.6% rural residents. Supplementary Table 1 (available online) reports demographics for each analytic sample.

Prevalence and Demographic Differences in Smoking Cessation Indicators

Interest. Among current smokers, quit smoking interest prevalence (pooled = 77.1%) did not change from 2014-2015 (77.5%) to 2018-2019 (76.6%) (change = -1.0%, 95% CI = -2.0% to 0.1%; P = .08). Collapsed across timepoints, quit-smoking interest was higher in high school graduates or higher education levels vs high school nongraduates, annual incomes $75 000 or more vs less than $25 000 employed full-time vs those not in workforce, and metropolitan vs rural residents after adjusting for other demographics. There was no evidence of differential changes across time for education, income, or employment (non-statistically significant time x demographic interactions; Table  1). Quit smoking interest was more common in Black vs White individuals collapsed across timepoints and changed across time differentially by race and ethnicity (P_interaction = .002). Quit smoking interest prevalence from 2014-2015 to 2018-2019 increased in American Indian and Alaska Native (65.1% to 79.7 %; P = .001), decreased in White (77.0% to 75.5%; P =.01), and did not statistically significantly change in other racial and ethnic groups (Table  1). Disability status was not associated with quit-smoking interest. Supplementary Table 2 (available online) presents the sensitivity analysis based on the continuous 10-point scale of interest in quit, with findings consistent with the binary outcome.

Table 1.

Prevalence and demographic factors associated with quit smoking interest, 2014-2015 and 2018-2019^a

Demographic factors	Pooled across timepoints			At each timepointb				Demographic variable × time interaction, Pg
	% (95% CI)b	Demographic difference aOR (95% CI)c	Pd	2014-2015, %	2018-2019, %	Change, % (95% CI)e	Pf
Overall sample	77.1 (76.5 to 77.6)	N/A		77.5	76.6	−1.0 (−2.0 to 0.1)	.08	N/A
Race and ethnicity								.002
American Indian or Alaskan Native	72.2 (66.9 to 77.0)	0.87 (0.68 to 1.12)	.28	65.1	79.7	14.6 (5.5 to 23.7)	.001
Asian	80.5 (76.7 to 83.9)	1.10 (0.88 to 1.38)	.40	83.4	77.1	−6.3 (−13.1 to 0.4)	.06
Black	79.6 (78.3 to 80.9)	1.27 (1.16 to 1.38)	<.001	80.2	78.8	−1.5 (−4.5 to 1.5)	.33
Hispanic	78.3 (76.2 to 80.3)	1.12 (0.99 to 1.27)	.08	77.3	79.5	2.2 (−1.3 to 5.8)	.21
Others	79.9 (75.6 to 83.5)	1.15 (0.89 to 1.48)	.28	79.4	80.4	1.0 (−6.6 to 8.6)	.80
White	76.3 (75.7 to 76.9)	Referent		77.0	75.5	−1.6 (−2.8 to 0.3)	.01
Education								.54
< H.S. graduate	72.5 (71.1 to 73.8)	Referent		73.3	71.5	−1.8 (−4.8 to 1.3)	.25
H.S. graduate	75.2 (74.3 to 76.1)	1.13 (1.04 to 1.22)	.004	75.5	74.9	−0.7 (−2.4 to 1.1)	.47
Some college	79.8 (78.7 to 80.9)	1.42 (1.29 to 1.56)	<.001	80.3	79.3	−1.0 (−3.1 to 1.1)	.34
Associate degree	78.7 (76.9 to 80.4)	1.32 (1.16 to 1.51)	<.001	80.4	76.9	−3.5 (−6.6 to 0.4)	.02
Bachelor degree	83.1 (81.5 to 84.5)	1.71 (1.49 to 1.96)	<.001	82.9	83.3	0.4 (−2.6 to 3.5)	.77
Graduate degree	80.3 (77.8 to 82.6)	1.44 (1.23 to 1.69)	<.001	80.8	79.8	−1.0 (−6.3 to 4.4)	.72
Annual income, $								.48
<25 000	74.9 (73.8 to 76.0)	Referent		75.8	73.6	−2.2 (−4.5 to 0.1)	.06
25 000−49 999	76.4 (75.5 to 77.3)	1.05 (0.97 to 1.14)	.23	76.5	76.3	−0.2 (−2.0 to 1.7)	.88
50 000−74 999	77.5 (76.3 to 78.6)	1.05 (0.96 to 1.16)	.30	78.8	76.2	−2.6 (−5.0 to 0.1)	.04
75 000−99 999	80.1 (78.2 to 81.9)	1.19 (1.03 to 1.36)	.02	81.0	79.2	−1.8 (−5.2 to 1.6)	.29
≥100 000	81.3 (79.8 to 82.7)	1.24 (1.09 to 1.40)	.001	81.8	80.9	−0.9 (−4.1 to 2.3)	.57
Disability status								.64
No	77.7 (77.0 to 78.3)	Referent		78.1	77.1	−1.0 (−2.2 to 0.2)	.09
Yes	74.6 (73.4 to 75.8)	1.07 (0.99 to 1.16)	.07	74.8	74.4	−0.4 (−3.0 to 2.2)	.75
Employment status								.79
Full-time	79.1 (78.3 to 79.9)	Referent		79.6	78.6	−1.0 (−2.4 to 0.4)	.15
Part-time	79.1 (77.4 to 80.7)	1.04 (0.93 to 1.15)	.51	79.6	78.5	1.0 (−0.1 to 2.0)	.51
Unemployed	80.7 (78.4 to 82.9)	1.13 (0.97 to 1.32)	.11	80.1	81.8	1.6 (−3.4 to 6.6)	.52
Not in workforce	73.0 (72.1 to 73.8)	0.88 (0.82 to 0.95)	.001	73.4	72.5	−0.8 (−2.7 to 1.0)	.37
Metropolitan area								.34
Yes	78.4 (77.7 to 79.0)	Referent		79.0	77.7	−1.3 (−2.5 to 0.1)	.04
No	72.0 (70.7 to 73.2)	0.77 (0.71 to 0.82)	<.001	72.1	71.9	−0.2 (−2.5 to 2.1)	.89

a

Interest in quitting defined: score of >2 on 10-point scale (1 [not at all interested] to 10 [extremely interested]). Analytical sample included all current smokers who responded to the question regarding interest in quitting (n = 37 903 pooled across both survey timepoints; n = 22 163 for 2014-2015, n = 15 740 for 2018-2019). aOR = adjusted odds ratios; CI = confidence interval; H.S. = high school; N/A = not applicable.

b

Weighted estimates within row.

c

Multivariable logistic regression model included pooled 2014-2015 and 2018-2019 data using interest in quitting as the outcome variable with age, sex, year (2018-2019 vs 2014-2015), and all variables in this table as simultaneous regressors. Age and sex-specific results are reported in Supplementary Table 4 (available online).

d

P value from multivariable logistic regression model. All statistical tests were 2-sided.

e

Unadjusted difference 2018-2019 minus 2014-2015 estimate.

f

P value from Rao-Scott χ² test for changes in weighted estimates. All statistical tests were 2-sided.

g

P value from 2-sided unadjusted logistic regression model omnibus interaction test.

Attempts. Among past-year smokers, self-reported past-year quit-attempt prevalence declined from 2014-2015 (52.9%) to 2018-2019 (51.3%) (change = -1.6%, 95% CI = -2.7% to -0.4%; P = .01). Collapsed across timepoints and adjusted for other demographics, the past-year quit-attempt prevalence was higher among Black, Hispanic, American Indian and Alaska Native, Other (non-Asian) racial and ethnic vs White respondents; college attendees or higher education levels vs high school nongraduates; those with disabilities vs those without; part-time employed and unemployed vs full-time employed; and metropolitan vs rural residence (Table  2). Income was not associated with quit-attempt prevalence. Changes in quit attempt prevalence across timepoints did not differ by any demographic variable (non-statistically significant interactions; Table  2).

Table 2.

Prevalence and demographic factors associated with past-year quit attempts, 2014-2015 and 2018-2019^a

Demographic factors	Pooled across timepoints			At each timepointb				Demographic variable × time interaction, Pg
	% (95% CI)b	Demographic difference aOR (95% CI)c	Pd	2014-2015, %	2018-2019, %	Change, % (95% CI)e	Pf
Overall sample	52.2 (51.6 to 52.7)	N/A		52.9	51.3	−1.6 (−2.7 to 0.4)	.01	N/A
Race and ethnicity								.34
American Indian or Alaskan Native	56.7 (51.9 to 61.5)	1.29 (1.06 to 1.58)	.01	55.3	58.2	2.9 (−6.3 to 12.1)	.53
Asian	53.5 (49.2 to 57.8)	1.08 (0.91 to 1.30)	.38	51.7	55.5	3.8 (−3.7 to 11.4)	.32
Black	55.5 (53.7 to 57.3)	1.22 (1.12 to 1.33)	<.001	55.3	55.8	0.6 (−2.9 to 4.1)	.75
Hispanic	53.7 (51.8 to 55.5)	1.13 (1.04 to 1.22)	.01	55.1	52.1	−3.0 (−7.1 to 1.1)	.15
Others	59.6 (55.0 to 64.0)	1.29 (1.07 to 1.56)	.01	61.7	57.4	−4.4 (−12.2 to 3.4)	.27
White	51.0 (50.3 to 51.6)	Referent		51.9	49.9	−2.0 (−3.4 to 0.6)	.004
Education								.99
< H.S. graduate	49.2 (47.6 to 50.8)	Referent		49.9	48.3	−1.6 (−4.9 to 1.8)	.35
H.S. graduate	49.9 (49.0 to 50.7)	1.07 (1.00 to 1.15)	.05	50.7	48.8	−1.9 (−4.0 to 0.2)	.07
Some college	55.3 (54.1 to 56.5)	1.29 (1.19 to 1.41)	<.001	55.9	54.6	−1.4 (−3.9 to 1.2)	.29
Associate degree	55.3 (53.7 to 57.0)	1.34 (1.22 to 1.48)	<.001	55.8	54.9	−0.9 (−4.3 to 2.5)	.59
Bachelor degree	54.6 (52.7 to 56.4)	1.32 (1.17 to 1.49)	<.001	55.6	53.5	−2.2 (−5.6 to 1.2)	.20
Graduate degree	53.5 (50.3 to 56.6)	1.30 (1.12 to 1.51)	.001	53.9	53.0	−0.9 (−6.9 to 5.1)	.76
Annual income, $								.05
<25 000	53.7 (52.6 to 54.7)	Referent		53.4	54.1	0.8 (−1.5 to 3.1)	.51
25 000−49 999	51.0 (50.0 to 51.9)	0.95 (0.89 to 1.01)	.08	52.0	49.7	−2.4 (−4.4 to 0.4)	.02
50 000−74 999	51.3 (49.9 to 52.8)	0.98 (0.90 to 1.06)	.54	52.2	50.5	−1.8 (−4.7 to 1.1)	.23
75 000−99 999	52.8 (50.7 to 54.9)	1.03 (0.92 to 1.14)	.64	52.4	53.2	0.8 (−3.1 to 4.7)	.69
≥100 000	53.1 (51.1 to 55.1)	1.04 (0.94 to 1.16)	.44	56.0	50.7	−5.3 (−9.1 to 1.4)	.01
Disability status								.31
No	51.4 (50.8 to 52.0)	Referent		52.3	50.4	−1.9 (−3.2 to 0.6)	.004
Yes	55.4 (53.9 to 56.8)	1.34 (1.24 to 1.44)	<.001	55.5	55.2	−0.4 (−3.0 to 2.3)	.78
Employment status								.51
Full-time	51.4 (50.6 to 52.2)	Referent		52.3	50.4	−1.9 (−3.5 to 0.2)	.03
Part-time	56.3 (54.3 to 58.2)	1.14 (1.04 to 1.25)	.01	55.9	56.9	1.0 (−2.8 to 4.8)	.60
Unemployed	56.9 (54.6 to 59.2)	1.14 (1.03 to 1.27)	.01	58.1	54.6	−3.6 (−9.7 to 2.6)	.25
Not in workforce	51.1 (50.2 to 52.1)	1.01 (0.94 to 1.07)	.83	51.5	50.6	−0.9 (−2.8 to 1.1)	.37
Metropolitan area								.81
Yes	53.0 (52.3 to 53.6)	Referent		53.7	52.2	−1.5 (−3.0 to 0.1)	.03
No	48.8 (47.6 to 49.9)	0.87 (0.82 to 0.92)	<.001	49.6	47.7	−1.9 (−4.4 to 0.6)	.13

a

Past-year quit attempts were defined as 1) current smokers who reported that they had made past 12-months quit attempts and 2) former smokers who quit completely within the past year. Analytical sample included current smokers and former smokers who quit within the past year and reported past-year quit attempts (n = 43 602 pooled across both survey timepoints; n = 25 529 for 2014-2015, n = 18 073 for 2018-2019). aOR = adjusted odds ratios; CI = confidence interval; H.S. = high school; N/A = not applicable.

b

Weighted estimates within row.

c

Multivariable logistic regression model included pooled 2014-2015 and 2018-2019 Tobacco Use Supplement data using past-year quit attempts as the outcome variable with age, sex, year (2018-2019 vs 2014-2015), and all variables in this table as simultaneous regressors. Age and sex-specific results are reported in Supplementary Table 4 (available online).

d

P value from multivariable logistic regression model. All statistical tests were 2-sided.

e

Unadjusted difference 2018-2019 minus 2014-2015 estimate.

f

P value from Rao-Scott χ² test for changes in weighted estimates. All statistical tests were 2-sided.

g

P value from 2-sided unadjusted logistic regression model omnibus interaction test.

Sustained Cessation. Among past-year smokers, prevalence of self-reported past-year sustained cessation for 6 or more months (pooled = 7.5%) did not change from 2014-2015 (7.6%) to 2018-2019 (7.5%) (change = -0.2%, 95% CI = -0.8% to 0.5%; P = .62). Collapsed across timepoints and adjusted for other demographics, sustained cessation was higher in White vs Black, Hispanic vs White, high school graduates or higher education levels vs high school nongraduates, $25 000 higher vs less than $25 000 income brackets, full-time employed vs unemployed, and metropolitan vs rural residents. Disability status was not associated with sustained cessation. Changes in sustained cessation across timepoints did not differ by demographics (Table  3).

Table 3.

Prevalence and demographic factors associated with sustained cessation, 2014-2015 and 2018-2019^a

Demographic factors	Pooled across timepoints			At each timepointb				Demographic variable × time interaction, Pg
	% (95% CI)b	Demographic difference aOR (95% CI)c	Pd	2014-2015,%	2018-2019, %	Change, % (95% CI)e	Pf
Overall sample	7.5 (7.3 to 7.8)	N/A		7.6	7.5	−0.2 (−0.8 to 0.5)	.62	N/A
Race and ethnicity								.32
American Indian or Alaskan Native	6.6 (4.3 to 10.0)	0.96 (0.59 to 1.54)	.85	6.0	7.2	1.2 (−4.3 to 6.8)	.66
Asian	8.8 (6.5 to 11.7)	0.91 (0.64 to 1.30)	.61	6.9	10.7	3.7 (−1.4 to 8.8)	.13
Black	5.0 (4.3 to 5.9)	0.74 (0.62 to 0.90)	.002	4.9	5.2	0.4 (−1.3 to 2.0)	.67
Hispanic	8.9 (7.9 to 10.2)	1.22 (1.04 to 1.42)	.02	8.6	9.3	0.6 (−1.7 to 3.0)	.59
Others	8.4 (6.3 to 11.1)	1.02 (0.73 to 1.44)	.89	7.4	9.4	2.0 (−2.0 to 6.0)	.32
White	7.7 (7.3 to 8.1)	Referent		8.0	7.4	−0.6 (−1.3 to 0.1)	.10
Education								.47
< H.S. graduate	4.8 (4.1 to 5.5)	Referent		4.5	5.1	0.6 (−0.8 to 1.9)	.42
H.S. graduate	6.1 (5.7 to 6.6)	1.23 (1.03 to 1.47)	.03	6.3	5.9	−0.4 (−1.3 to 0.6)	.48
Some college	8.5 (7.7 to 9.3)	1.64 (1.35 to 1.98)	<.001	8.6	8.3	−0.3 (−1.8 to 1.2)	.68
Associate degree	9.3 (8.4 to 10.3)	1.92 (1.57 to 2.35)	<.001	8.9	9.8	0.9 (−1.3 to 3.1)	.43
Bachelor degree	11.9 (10.8 to 13.1)	2.37 (1.92 to 2.93)	<.001	12.8	11.0	−1.8 (−4.1 to 0.5)	.12
Graduate degree	10.8 (9.0 to 13.0)	2.22 (1.70 to 2.90)	<.001	11.7	10.0	−1.7 (−5.6 to 2.3)	.41
Annual income, $								.34
<25 000	5.8 (5.2 to 6.3)	Referent		5.8	5.8	0.0 (−1.2 to 1.2)	1.00
25 000−49 999	7.2 (6.7 to 7.7)	1.15 (1.00 to 1.33)	.046	7.3	7.0	−0.4 (−1.4 to 0.7)	.51
50 000−74 999	8.2 (7.5 to 9.0)	1.25 (1.08 to 1.43)	.003	8.5	7.9	−0.6 (−2.1 to 1.0)	.48
75 000−99 999	9.2 (8.2 to 10.4)	1.35 (1.13 to 1.62)	.001	8.9	9.6	0.7 (−1.5 to 2.9)	.52
≥100 000	10.2 (9.2 to 11.3)	1.43 (1.20 to 1.70)	<.001	11.4	9.2	−2.2 (−4.3 to 0.2)	.04
Disability status								.36
No	7.9 (7.5 to 8.2)	Referent		8.0	7.7	−0.3 (−1.0 to 0.4)	.45
Yes	6.2 (5.5 to 6.9)	1.05 (0.92 to 1.21)	.47	6.0	6.4	0.4 (−0.9 to 1.8)	.51
Employment status								.67
Full-time	8.4 (8.0 to 8.8)	Referent		8.5	8.3	−0.1 (−1.1 to 0.8)	.77
Part-time	8.4 (7.4 to 9.6)	1.06 (0.91 to 1.24)	.46	8.2	8.7	0.6 (−1.8 to 2.9)	.64
Unemployed	5.3 (4.2 to 6.5)	0.68 (0.53 to 0.87)	.003	5.7	4.4	−1.3 (−3.6 to 1.1)	.31
Not in workforce	6.4 (6.0 to 6.9)	1.05 (0.94 to 1.19)	.38	6.6	6.2	−0.4 (−1.4 to 0.6)	.43
Metropolitan area								.23
Yes	8.0 (7.6 to 8.3)	Referent		8.0	7.9	−0.1 (−0.8 to 0.7)	.89
No	5.8 (5.3 to 6.3)	0.77 (0.70 to 0.86)	<.001	6.1	5.3	−0.8 (−1.8 to 0.3)	.16

a

Recently sustained smoking cessation was defined as quitting smoking 6-11 months ago. Analytical sample included current smokers who smoked for ≥2 years and former smokers who quit during the past year (n = 43 361; pooled across both survey timepoints; n = 25 349 for 2014-2015, n = 18 012 for 2018-2019). aOR = adjusted odds ratios; CI = confidence interval; H.S. = high school; N/A = not applicable.

b

Weighted estimates within row.

c

Multivariable logistic regression model included pooled 2014-2015 and 2018-2019 Tobacco Use Supplement data using sustained cessation as the outcome variable with age, sex, year (2018-2019 vs 2014-2015), and all variables in this table as simultaneous regressors. Age and sex-specific results are reported in Supplementary Table 4 (available online).

d

P value from multivariable logistic regression model. All statistical tests were 2-sided.

e

Unadjusted difference 2018-2019 minus 2014-2015 estimate.

f

P value from Rao-Scott χ² test for changes in weighted estimates. All statistical tests were 2-sided.

g

P value from 2-sided unadjusted logistic regression model omnibus interaction test.

Attempt to Quit by Switching to E-cigarettes. Among smokers who made past-year quit attempts, the prevalence of trying to quit smoking by switching to e-cigarettes in the past year declined from 2014-2015 (33.3%) to 2018-2019 (25.0%) (change = -8.3%, 95% CI = -9.9% to -6.7%; P < .001) (Table  4). Collapsed across timepoints and adjusted for other demographics, trying to quit by switching to e-cigarettes was higher in groups with high school degrees, some college, or associate degrees vs high school nongraduates, $100 000 or higher vs less than $25 000 incomes, those with disabilities vs without, and metropolitan vs rural residents, with no interactions with time for these variables (Table  4). Black, Hispanic, American Indian and Alaska Native, and Asian (vs White) subgroups had a lower prevalence of quit attempts involving switching to e-cigarettes collapsed across time. Race and ethnicity differentiated time changes in this outcome (P_interaction < .001), whereby prevalence from 2014-2015 to 2018-2019 did not change in Hispanic, American Indian and Alaska Native, or Asian participants but decreased in White (change = -9.6%, 95% CI = -11.4% to -7.9%; P < .001), Black (change = -7.3%, 95% CI = -11.4% to -3.3%; P < .001), and other/multiracial (change = -24.2%, 95% CI = -34.5% to -13.8%; P < .001) racial and ethnic groups. Employment was not associated with using e-cigarettes to quit smoking.

Table 4.

Prevalence and demographic factors associated with switching to e-cigarettes, 2014-2015 and 2018-2019^a

Demographic factors	Pooled across time points			At each time pointb				Demographic variable × time interaction, Pg
	% (95% CI)b	Demographic difference aOR (95% CI)c	Pd	2014-2015,%	2018-2019, %	Change, % (95% CI)e	Pf
Overall sample	29.6 (28.8 to 30.3)	N/A		33.3	25.0	−8.3 (−9.9 to 6.7)	<.001	N/A
Race and ethnicity								<.001
American Indian or Alaskan Native	17.6 (12.9 to 23.5)	0.45 (0.30 to 0.66)	<.001	20.3	15.0	−5.3 (−15.3 to 4.8)	.32
Asian	21.2 (17.3 to 25.8)	0.54 (0.42 to 0.70)	<.001	21.9	20.6	−1.3 (−10.5 to 8.0)	.79
Black	19.2 (17.2 to 21.3)	0.49 (0.42 to 0.57)	<.001	22.6	15.3	−7.3 (−11.4 to 3.3)	<.001
Hispanic	22.8 (20.5 to 25.4)	0.57 (0.49 to 0.67)	<.001	21.8	24.1	2.3 (−2.6 to 7.3)	.35
Others	36.4 (31.1 to 42.1)	1.06 (0.83 to 1.36)	.64	47.6	23.5	−24.2 (−34.5 to 13.8)	<.001
White	32.8 (31.9 to 33.7)	Referent		37.1	27.5	−9.6 (−11.4 to 7.9)	<.001
Education								.68
< H.S. graduate	25.2 (23.3 to 27.3)	Referent		27.8	21.7	−6.1 (−10.3 to 2.0)	.005
H.S. graduate	29.5 (28.3 to 30.7)	1.14 (1.01 to 1.28)	.04	33.5	24.6	−8.9 (−11.4 to 6.4)	<.001
Some college	31.7 (30.0 to 33.4)	1.21 (1.06 to 1.39)	.01	35.8	26.3	−9.5 (−12.7 to 6.3)	<.001
Associate degree	32.0 (29.8 to 34.2)	1.26 (1.07 to 1.47)	.005	36.9	26.7	−10.3 (−14.6 to 6.0)	<.001
Bachelor degree	29.9 (27.7 to 32.2)	1.12 (0.96 to 1.30)	.16	33.0	26.5	−6.6 (−10.9 to 2.2)	.003
Graduate degree	25.4 (21.9 to 29.2)	0.94 (0.75 to 1.19)	.61	27.7	23.2	−4.5 (−12.3 to 3.4)	.26
Annual income, $								.62
<25 000	27.9 (26.5 to 29.3)	Referent		31.1	23.2	−7.9 (−10.8 to 5.0)	<.001
25 000−49 999	29.9 (28.7 to 31.2)	1.06 (0.97 to 1.17)	.18	33.2	25.6	−7.6 (−10.5 to 4.8)	<.001
50 000−74 999	29.9 (28.1 to 31.7)	1.04 (0.92 to 1.17)	.56	33.8	25.9	−7.9 (−11.5 to 4.4)	<.001
75 000−99 999	28.9 (26.4 to 31.6)	0.99 (0.84 to 1.18)	.95	34.7	23.2	−11.6 (−17.0 to 6.1)	<.001
≥100 000	32.1 (29.9 to 34.5)	1.17 (1.02 to 1.34)	.03	37.9	26.8	−11.1 (−15.7 to 6.6)	<.001
Disability status								.96
No	29.6 (28.7 to 30.4)	Referent		33.3	24.9	−8.3 (−10.1 to 6.6)	<.001
Yes	29.6 (28.0 to 31.2)	1.27 (1.16 to 1.40)	<.001	33.5	25.2	−8.3 (−11.5 to 5.1)	<.001
Employment status								.13
Full-time	30.6 (29.5 to 31.7)	Referent		34.3	26.3	−8.0 (−10.2 to 5.7)	<.001
Part-time	31.3 (29.1 to 33.5)	1.00 (0.89 to 1.14)	.96	34.4	27.0	−7.4 (−12.1 to 2.7)	.003
Unemployed	34.2 (30.6 to 37.9)	1.16 (0.98 to 1.37)	.08	34.9	32.5	−2.4 (−9.8 to 4.9)	.52
Not in workforce	26.5 (25.4 to 27.6)	0.96 (0.87 to 1.07)	.46	31.0	21.3	−9.7 (−12.4 to 7.0)	<.001
Metropolitan area								.72
Yes	29.8 (29.0 to 30.7)	Referent		33.6	25.3	−8.3 (−10.1 to 6.5)	<.001
No	28.3 (26.6 to 30.1)	0.84 (0.77 to 0.93)	.001	32.1	23.4	−8.7 (−12.0 to 5.3)	<.001

a

Analytic sample included current smokers with ≥1 past-year quit attempt or former smokers who quit during the past year and responded to the question regarding switching to e-cigarettes (n = 21 694; pooled across both survey time points; n = 12 937 for 2014-2015, n = 8757 for 2018-2019). aOR = adjusted odds ratios; CI = confidence interval; H.S. = high school; N/A = not applicable.

b

Weighted estimates within row.

c

Multivariable logistic regression model included pooled 2014-2015 and 2018-2019 Tobacco Use Supplement data using switching to e-cigarettes as the outcome variable with age, sex, year (2018-2019 vs 2014-2015), and all variables in this table as simultaneous regressors. Age and sex-specific results are reported in Supplementary Table 4 (available online).

d

P value from multivariable logistic regression model. All statistical tests were 2-sided.

e

Unadjusted difference 2018-2019 minus 2014-2015 estimate.

f

P value from Rao-Scott χ² test for changes in weighted estimates. All statistical tests were 2-sided.

g

P value from 2-sided unadjusted logistic regression model omnibus interaction test.

Cessation Treatment. Among smokers who made past-year quit attempts in 2018-2019, 34.4% reported using some type of smoking cessation treatment, including an Food and Drug Administration (FDA)–approved nicotine replacement therapy (24.7%) or prescription medication (12.1%), telephone quitline (3.6%), in-person one-on-one counseling (5.2%), class or group counseling (2.1%), digital web or smartphone application (2.6%), or medication and counseling combination treatment (6.1%) (Table  5). After adjusting for demographics, reported use of 1 or more treatments was higher in White than Black, Hispanic, or Asian respondents; people with disabilities vs those without; and those not in workforce vs full-time employed. Education, income, and metropolitan residence were not associated with reported treatment use. Prevalence of using specific cessation treatments stratified by demographic variables are reported in Supplementary Table 3 (available online). Age and sex-specific results are reported in Supplementary Table 4 (available online).

Table 5.

Prevalence and demographic factors associated with use of smoking cessation treatments, 2018-2019^a

Use of smoking cessation treatments	% (95% CI)d	aOR (95% CI)e	Pf
Use of specific treatments in overall sampleb
Nicotine replacement therapy	24.7 (23.5 to 25.8)	N/A	N/A
Varenicline or bupropion medication	12.1 (11.3 to 12.9)	N/A	N/A
Telephone counseling quitline	3.6 (3.1 to 4.1)	N/A	N/A
One-on-one in-person counseling	5.2 (4.7 to 5.8)	N/A	N/A
Group counseling or class	2.1 (1.8 to 2.5)	N/A	N/A
Internet or smartphone application	2.6 (2.2 to 3.0)	N/A	N/A
Counseling + medication combination treatmentc	6.1 (5.5 to 6.7)	N/A	N/A
Use ≥1 treatment by demographic factors
Overall sample	34.4 (33.1 to 35.6)	N/A	N/A
Race and ethnicity
American Indian or Alaskan Native	34.2 (25.8 to 43.8)	0.96 (0.63 to 1.46)	.85
Asian	24.3 (18.4 to 31.4)	0.61 (0.42 to 0.88)	.01
Black	30.7 (27.2 to 34.5)	0.71 (0.58 to 0.86)	.001
Hispanic	25.6 (21.8 to 29.8)	0.64 (0.51 to 0.81)	<.001
Others	35.6 (27.9 to 44.1)	0.91 (0.62 to 1.31)	.60
White	36.9 (35.6 to 38.3)	Referent
Education
< H.S. graduate	35.0 (31.4 to 38.8)	Referent
H.S. graduate	32.9 (30.8 to 35.0)	0.95 (0.76 to 1.17)	.61
Some college	34.7 (32.1 to 37.4)	1.04 (0.83 to 1.31)	.75
Associate degree	36.7 (33.2 to 40.3)	1.08 (0.85 to 1.37)	.53
Bachelor degree	33.9 (30.5 to 37.5)	1.01 (0.79 to 1.29)	.93
Graduate degree	38.6 (32.1 to 45.6)	1.17 (0.81 to 1.68)	.41
Annual income, $
<25 000	37.5 (35.0 to 40.1)	Referent
25 000-49 999	31.8 (29.8 to 33.8)	0.91 (0.78 to 1.07)	.24
50 000-74 999	34.4 (31.5 to 37.4)	1.06 (0.87 to 1.30)	.56
75 000-99 999	35.8 (31.7 to 40.0)	1.14 (0.90 to 1.45)	.28
≥100 000	34.1 (30.4 to 37.9)	1.00 (0.79 to 1.26)	.97
Disability status
No	31.4 (30.1 to 32.8)	Referent
Yes	45.4 (42.8 to 48.0)	1.44 (1.26 to 1.65)	<.001
Employment status
Full-time	30.6 (29.0 to 32.3)	Referent
Part-time	31.5 (27.9 to 35.3)	1.05 (0.86 to 1.29)	.60
Unemployed	30.5 (24.8 to 36.8)	1.11 (0.80 to 1.53)	.53
Not in workforce	41.2 (39.3 to 43.2)	1.20 (1.03 to 1.39)	.02
Metropolitan area
Yes	34.4 (33.0 to 35.8)	Referent
No	34.2 (31.5 to 37.1)	0.91 (0.78 to 1.06)	.21

a

Analytical sample included current smokers with ≥1 past-year quit attempt or former smokers who quit during the past year and reported cessation treatment outcomes (n = 9038). The analytical sample included 2018-2019 respondents only because limited cessation treatments were asked in the 2014-2015 survey timepoint. aOR = adjusted odds ratios; CI = confidence interval; H.S. = high school; N/A = not applicable.

b

Smoking cessation treatments included 1) nicotine patch, gum, lozenge, nasal spray, or inhaler (n = 9031); 2) prescription pill, called Chantix, Varenicline, Zyban, Bupropion, or Wellbutrin (n = 9028); 3) telephone helpline or quitline (n = 9030); 4) one-on-one in-person counseling by a health professional (n = 9025); 5) stop smoking clinic, class, or support group (n = 9025); and 6) internet or a web-based program or tool including smartphone apps and text messaging programs (n = 9024). Results from analyses for each of these individual 6 outcomes are reported in Supplementary Table 3 (available onaline).

c

Nicotine patch, gum, lozenge, nasal spray or inhaler or prescription pill, called Chantix, Varenicline, Zyban, Bupropion, or Wellbutrin, in addition to one-on-one in-person counseling by a health professional or a stop smoking clinic, class, or support group.

d

Weighted estimates within row.

e

Multivariable logistic regression with age, sex, and all variables in table as simultaneous regressors. Age and sex-specific results are reported in Supplementary Table 4 (available online).

f

P value from multivariable logistic regression model. All statistical tests were 2-sided.

Discussion

More than three-quarters of US adult current smokers in the TUS-CPS during 2014-2019 expressed interest in quitting smoking, and approximately one-half made past-year quit attempts. Yet, only 7.5% made a quit attempt resulting in sustained cessation of smoking in the past year. Low cessation rates might, in part, be explained by this study’s finding that most smokers making quit attempts did not report using an FDA-approved cessation pharmacotherapy or behavioral treatment.

Timepoint analyses showed no change in cessation interest or success and slight reductions in past-year quit attempt prevalence from 2014-2015 to 2018-2019. Despite recent attempts to increase smoking cessation treatment access by advertising availability of telephone quitlines that offer free counseling, government cost subsidization of nicotine replacement therapy, efforts to increase health-care coverage for quit-smoking medications, and proliferation of digital quit-smoking treatment applications (21), use of each of these treatment modalities remained modest in 2018-2019 and not markedly different from previously published national estimates from 2015 NHIS (8) or 2010-2011 TUS-CPS (12). Although observational research on the association of e-cigarette use and smoking cession is inconclusive (21,22), results from a recent randomized clinical trial support the cessation-promoting efficacy of e-cigarettes (23). Some evidence is suggestive that previous upticks in US adult smoking cessation prevalence leading up to 2015 might have been due in part to more smokers switching to e-cigarettes (24). This study found attempting to quit smoking by switching to e-cigarettes declined from 2014-2015 to 2018-2019. The reasons for this trend are unknown. Targeted industry marketing of e-cigarettes to youth and widespread adolescent vaping could dissuade adults from perceiving vaping as a possible quit aid for adult smokers (25). Media reports of new research on the toxicity of e-cigarettes could also create perceptions that overestimate the harms of e-cigarettes (26).

This study elucidates the role of cessation inequalities in the recently widening disparities in smoking prevalence facing certain racial and ethnic minorities, socioeconomically disadvantaged, disabled, and rural populations (27). With the exception of disabled smokers, many subgroupings within each of these populations had disproportionately low rates of sustained cessation during this study but not necessarily lower quit motivation. For example, in this study, Black vs White smokers had higher prevalence of interest in quitting smoking and making quit attempts, but lower odds of sustained cessation and lower reported use of treatments. Similar findings were also found in 2015 estimates (8). Here we show the persistence of this disparity from 2014 to 2019 and offer new evidence that Black (vs White) smokers are less likely to report using pharmacological or behavioral treatments or e-cigarettes when attempting to quit. Structural and interpersonal racism may contribute to these disparities (17). Inequalities in access to quality health care can reduce Black smokers’ contact with physicians offering advice to quit smoking and cessation treatments (7). Targeted industry marketing results in an overrepresentation of menthol-flavored cigarette use among Black vs White smokers (28). Smokers of menthol-flavored vs unflavored cigarettes report lower odds of using tobacco cessation treatments and quitting smoking (29,30). Prior studies also show fewer tobacco retailers selling e-cigarettes in neighborhoods with a higher proportion of Black residents (31), which might account for fewer Black smokers attempting to switch to e-cigarettes. These results indicate the need to identify methods for improving the accessibility and appeal of smoking cessation aids for Black smokers to capitalize on high motivation to quit smoking.

The large sample of TUS-CPS also permitted some of the first nationally representative epidemiologic estimates of smoking cessation indicators in understudied groups, including American Indians and Alaska Natives who have disproportionately high smoking prevalence (32). This study shows notable increases in interest in quitting smoking among American Indian and Alaska Native smokers from 2014-2015 to 2018-2019. We also found a higher prevalence of quit attempts in American Indian and Alaska Native vs White smokers averaged across the 2 timepoints, indicating that efforts to increase motivation to quit smoking commercially manufactured cigarettes in American Indian and Alaska Native communities might have been successful (33). Although American Indian and Alaska Native and White smokers did not differ in reported treatment use, this study provides new evidence of lower prevalence of switching to e-cigarettes when trying to quit smoking in American Indian and Alaska Natives.

Rural vs urban and educational disparities in smoking prevalence are some of the most robust disparities in smoking and contribute substantially to geographic and socioeconomic differences in life expectancy (27). We provide insight into these disparities by demonstrating that high school nongraduates, rural area residents, and (to some extent) lower income and unemployed groups exhibited disproportionately low prevalence on several cessation-related outcomes that persisted across 2014-2015 to 2018-2019. Similar to some racial and ethnic disparities, this study found that lower income and rural residents had disproportionately lower odds of attempting to quit smoking with e-cigarettes but were not underrepresented among users of FDA-approved smoking cessation medications or behavioral programs. The current findings suggest comprehensive approaches for rural or socioeconomically disadvantaged communities to increase multiple points of the smoking-cessation process, including motivating quit attempts and improving successful quit-smoking outcomes, merit consideration.

This study has limitations. First, self-reported data from the TUS-CPS are subject to potential recall errors and other self-report measurement biases, although the CPS takes multiple steps to minimize measurement error (34). Second, the TUS-CPS data lack sexual and gender minority status, behavioral health problems, pregnancy, and active-duty military, which precluded examination of inequities for these important subgroups. Third, this study focused on inequalities associated with each sociodemographic variable separately and addressed the correlation of demographic statuses with multivariable models by statistical adjustment, which showed their incremental associations over and above co-occurrence. The possibility of complex intersections of multiple demographic statuses in altering smoking cessation warrants future investigation with other modeling strategies, although prior research on current smoking status shows additive, incremental instead of synergistic associations of multiple demographic statuses (3,35). Fourth, data on several smoking cessation treatments were not collected in the 2014-2015 TUS-CPS, precluding direct comparison between 2 timepoints.

Although more than one-half of 2014-2019 TUS-CPS US adult smokers made past-year quit attempts, only 7.5% reported sustained cessation and most who made quit attempts did not report using a cessation treatment. Sociodemographic inequalities in smoking cessation outcomes were pervasive and not entirely correspondent with sociodemographic variation in motivation to quit. Given these findings and that smoking cessation prevalence and inequalities did not improve from 2014 to 2019, encouraging quit attempts and access to smoking cessation aids equitably across demographic subgroups merits priority in clinical and public health practice.

Funding

Research reported in this publication was supported by the National Cancer Institute under Award Number U54CA180905 (Leventhal/Pentz), the National Institute on Drug Abuse under Award Number K24DA048160 (Leventhal), and the National Institute on General Medical Sciences under Award Number P20GM103644 (Higgins). Research reported in this publication was supported by the National Cancer Institute under Award Number U54CA180905 (Leventhal/Pentz), the National Institute on Drug Abuse under Award Number K24DA048160 (Leventhal), the National Institute on General Medical Sciences under Award Number P20GM103644 (Higgins) and National Institute on Drug Abuse under Award Number (U54DA036114) (Higgins).

Notes

Role of funders: The funding agencies had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclosures: The authors report no potential conflicts of interests.

Author contributions: Conceptualization: Leventhal, Dai, Higgins; Data curation: Leventhal, Dai; Formal analysis: Dai; Funding Acquisition: Leventhal, Higgins; Investigation: Leventhal, Dai, Higgins; Methodology: Leventhal, Dai; Project Administration: Leventhal, Dai; Resources: Leventhal, Dai; Software: Dai; Supervision: Leventhal, Dai; Validation: Leventhal, Dai; Writing-Original Draft: Leventhal, Dai, Higgins; Writing—Review & Editing: Leventhal, Dai, Higgins.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Data Availability

The CPS-TUS data are publicly available and can be downloaded here: https://cancercontrol.cancer.gov/brp/tcrb/tus-cps/.