Smoking Behaviors in Survivors of Smoking-Related and Non–Smoking-Related Cancers

Key Points Question Among adult cancer survivors, do cigarette smoking behaviors vary according to whether the cancer was smoking related or not smoking related? Findings This cross-sectional study found that, in the 2017 US National Health Interview Survey, which included 26 742 respondents aged 18 years or older, current smoking prevalence was higher among smoking-related cancer survivors compared with non–smoking-related cancer survivors (19.78% vs 10.63%). After cancer diagnosis, the odds of continued cigarette smoking were twice as high among those with smoking-related cancers compared with those with non–smoking-related cancers. Meaning Compared with non–smoking-related cancer survivors, smoking-related cancer survivors have a higher risk of being cigarette smokers and of continuing smoking.


Introduction
The cancer survivor population in the US has increased over the past half century to an estimated 18.1 million in 2020 and is projected to increase to more than 20 million by 2026. 1,2 This phenomenon is largely due to more cancer diagnoses as a result of an aging population and improvements in early detection and treatment of cancer. 3,4 Healthy lifestyle behaviors (eg, smoking cessation, physical activity, maintaining healthy weight, and consuming a healthy diet) may help improve quality of life among cancer survivors and prevent recurrent and subsequent cancers. 4 Smoking remains the leading preventable cause of disease and death in the US. Although numerous antitobacco efforts have made substantial contributions to public health, cigarette smoking continues to claim half a million lives annually as a result of lung cancer and other tobaccorelated diseases. [5][6][7] Compared with individuals who have never smoked, cigarette smokers are at increased risk of having detrimental health conditions such as cancer (eg, leukemia and lung cancer), cardiovascular disorders (eg, atherosclerosis and hypertension), respiratory diseases (eg, chronic obstructive pulmonary disease, pneumonia, chronic bronchitis, and emphysema), diabetes, and oral conditions (eg, periodontal disease, oral leukoplakia, and oral cancer), among others. [8][9][10][11][12] Consequently, smokers tend to be admitted to the hospital more often than nonsmokers and, ultimately, have a higher risk of premature death, 13 which carries an associated economic cost to the nation of nearly $300 billion a year. 6,14,15 Many cancer survivors continue to smoke, despite the knowledge that continued smoking leads to poor clinical outcomes and shorter survival times. [16][17][18] It is established that individuals who continue to smoke after a cancer diagnosis are at increased risk of death from smoking-related cardiovascular and respiratory complications, as well as higher risk of cancer recurrence, the development of second primary cancers, and complications from treatment. [19][20][21] Continuing to smoke after a cancer diagnosis has also been associated with overall poorer physical, social, and emotional functioning. 22,23 Previous studies 24-28 have examined the smoking practices among cancer survivors in the US. However, those studies did not include indicators of mental health, nor did they measure the impact on smoking behaviors of other tobacco-related nonmalignant conditions among cancer survivors who smoke. 24-26 Furthermore, it remains unclear whether cigarette smoking behaviors vary between survivors of smoking-related cancers (SRCs) and non-smoking-related cancers (NSRCs).
In this cross-sectional study, we investigated the prevalence and patterns of continuing or quitting smoking after cancer diagnosis by taking into account key potential confounders, while comparing smoking behaviors between survivors of SRCs and NSRCs. Understanding the factors that contribute to smoking behaviors among cancer survivors will assist in the development of smoking cessation interventions targeted to and tailored for this high-risk group.

Data and Sampling Design
NHIS study provided written informed consent before participation, and the survey was approved by the research ethics review board of the National Center for Health Statistics. Our cross-sectional study was limited to adult participants (ie, respondents aged Ն18 years). This secondary data analysis of the 2017 NHIS data was exempt from institutional review board approval because all data sets are publicly available, in accordance with 45 CFR §46. This study follows the American Association for Public Opinion Research (AAPOR) reporting guideline.
A detailed description of the NHIS sampling design is reported by Parsons et al. 29 Briefly, the data collection process used a multistage stratified area probability design to recruit households and implemented the survey in face-to-face interviews. In the first stage of the sampling design, 319 primary sampling units were sampled from approximately 1700 geographically defined primary sampling units from all 50 states and the District of Columbia. In the second stage of the sampling design, area segments (geographically defined areas within a primary sampling unit) and permit segments (defined using housing units built after the 2000 US Census) were used to sample households within a primary sampling unit.

Identification and Classification of Cancer Survivors
In the 2017 NHIS survey, questions about cancer survivorship were administered by all US states and territories and the District of Columbia. Respondents were asked whether they had ever been told by a physician, nurse, or other health care professional that they had cancer. If they answered yes, they were asked how many different types of cancer they had had, their age at first diagnosis, and the type(s) of cancer. If respondents reported more than 1 cancer, they were included in the analyses separately for each cancer reported. Respondents who had an unknown history of cancer or who did not answer the question were excluded from the analyses.
Cancer survivors were classified into 2 categories depending on whether tobacco was positively associated with the type of cancer they were diagnosed with, per the 2014 Surgeon General Report 6 : survivors of SRCs (eg, cancers of the bladder, blood, cervix, colon, esophagus, kidney, larynx or windpipe, leukemia, lung, liver, mouth, tongue, lip, pancreas, rectum, stomach, throat or pharynx, and uterus) and survivors of NSRCs (eg, cancers of the bone, brain, breast, gallbladder, lymphoma, melanoma, ovary, prostate, skin, soft tissue, testis, and thyroid). Respondents who reported a prior cancer diagnosis at any other site were included in the population of other NSRC survivors.

Smoking Behavior
Using NHIS definitions, individuals who smoked 100 cigarettes or more in their lifetime and who smoked every day or on some days at the time of the survey were defined as current smokers.
Individuals who smoked fewer than 100 cigarettes in their lifetime were defined as never smokers.
Individuals who smoked 100 cigarettes or more in their lifetime but did not smoke at the time of the survey were defined as former smokers. The information on current smokers included the number of cigarettes smoked per day, whether they were daily or smoked only some days, and whether they had tried to quit smoking in the last 12 months. Those respondents who were current smokers and stopped smoking for more than 24 hours in the last 12 months because they were trying to quit were classified as having attempted to quit smoking in the past 12 months but still smoking.
hopelessness. In the 2017 NHIS, survey respondents were asked, "The next questions ask about difficulties you may have doing certain activities because of a HEALTH PROBLEM. By 'health problem' we mean any physical, mental, or emotional problem or illness (not including pregnancy)." Those who mentioned "depression/anxiety/emotional problem" as a health problem were considered as having this condition. "Fatigue/tiredness/weakness" was defined in the same way.
Hopelessness was ascertained using the following questions: "During the past 30 days, how often do you feel hopeless?" and the possible answers included all of the time, most of the time, some of the time, a little of the time, and none of the time. Items assessing smokeless tobacco or electronic nicotine delivery device use were not included in this study.

Statistical Analysis
Because the NHIS data are based on a multistage sampling design, analyses were performed using the sample weights that take into account nonresponse and post hoc observed variations in age, sex, and race compared with the US census data. To account for these weights, we used the R statistical software package survey 30 version 3.6.1 (R Project for Statistical Computing) to analyze the data. We referred to the 2010 US census data to calculate the age-adjusted and cancer type-specific prevalence of smoking by stratifying the survey data into 4 age groups: 18 to 24 years, 25 to 44 years, 45 to 64 years, and 65 years or older.
We compared demographic characteristics and tobacco use reported by all adult NHIS respondents, respondents with SRCs, those with NSRCs, and individuals without cancer. We performed descriptive analyses in which we calculated the age-adjusted prevalence of smokers who quit smoking after cancer diagnosis. Two questions from the NHIS survey were used to assess whether a participant quit or continued smoking after cancer diagnosis: (1) their age at cancer diagnosis, and (2) the age at which they quit smoking. To identify factors associated with quitting smoking compared with continuing smoking among cancer survivors, we performed survey logistic regression analysis with the R command svyglm using the quasibinomial link function to adjust for sociodemographic factors (eg, age, sex, education level, marital status, and race), mental health indicators (depression, anxiety, emotional problem, fatigue, tiredness, weakness, and hopelessness), and other tobacco-related comorbidities (eg, hypertension, asthma, emphysema, diabetes, and arthritis). The logistic regression model used was based on the complex survey design, with inverseprobability sample weighting and design-based standard errors. To avoid warnings regarding noninteger number of success in the estimation process, we used the quasibinomial link function per the package survey's instructions. Because variables considered to be potential confounders of smoking behaviors were all included in the final model, we did not conduct any variable selection approaches (eg, stepwise regression analysis). In this analysis, respondents' age (in years), as well as the years survived after cancer diagnosis, were included in the model as continuous variables. The percentages presented in this report are age-adjusted and weighted. For all statistical analyses, the significance level was calculated with 2-sided t tests and defined as P Յ .05. Data analysis was performed from June to October 2019.  Abbreviation: NA, not applicable. a The overlap observed here is because patients with more than 1 cancer were included separately in the analyses for each cancer diagnosed.

Prevalence and Patterns of Smoking Among Cancer Survivors
The age-adjusted prevalence of current, former, and never smokers among cancer survivors and by cancer type is reported in

Percentage of Smokers Who Quit Smoking or Attempted to Quit After Cancer Diagnosis
The age-adjusted percentage of smokers who quit smoking after cancer diagnosis and that of cancer survivors who attempted to quit within the past 12 months but were still smoking at the time of the survey are reported in Table 3

Factors Associated With Continuing Cigarette Smoking Compared With Quitting After Cancer Diagnosis
Factors associated with continuing smoking compared with quitting smoking after cancer diagnosis are presented in Table 4

Discussion
In this study, we assessed smoking behaviors among US adult cancer survivors using the 2017 NHIS data. We found the prevalence of cigarette smoking in this high-risk group ( Behavioral Risk Factor Surveillance System data, the prevalence of smoking was also found to be higher among cancer survivors with SRCs (27%) compared with NSRC survivors (16%) and individuals without cancer (18%). 26 The lower rates observed in our study could be attributed to the recency of our data, the use of an updated classification of SRCs (the 2014 US Surgeon General report), and the use of a different sampling approach to enroll study participants. A key finding of this study is that individuals with SRCs had lower odds of quitting smoking after cancer diagnosis. However, the existing literature is inconsistent. The odds of quitting smoking after cancer diagnosis were found to be higher among SRC survivors (and not NSRC survivors) in a  36 On the other hand, a recent report 37 from a comprehensive smoking cessation intervention in an oncology setting found that cessation rates did not differ between those with and without SRCs. Our study accounted for the presence of multiple tobacco-related comorbidities. This is reflected in the higher odds of continued smoking among cancer survivors with comorbidities like angina pectoris and chronic bronchitis. Although a large majority of oncology practitioners ask about tobacco use, they may be less likely to be involved in discussing treatment options. [40][41][42] In this study, older people as well as individuals with asthma or coronary artery disease had lower odds of continuing smoking after cancer diagnosis. Consistent with our results, a study on cancer survivorship found that a substantial proportion of cancer survivors also has heart disease, diabetes, asthma, and other health-related illness. 43 Persons with comorbidities are typically older and have more interaction with health care practitioners and, therefore, more opportunity for counseling to quit smoking. Thus, the lower smoking prevalence among the older population of cancer survivors may be partly explained by comorbidities. However, these results should be interpreted with caution, considering the higher survival rates observed among cancer survivors who are nonsmokers, compared with those who are smokers. 6,7 This was notably evidenced among never-smoking lung cancer survivors in whom lung cancer harbors targetable long survival variants like EGFR. 44,45 Interestingly, within the group of SRC survivors, those with lung cancer had a lower rate of smoking (10.19%) compared to individuals with other types of cancer (eg, 30.08% kidney) which could be due to the lethality of lung cancer, especially for continuing smokers. Also, according to selfreported knowledge of SRC among patients visiting a urology clinic, 46 it has been observed that many patients are not aware that smoking is a critical risk factor for kidney cancer as well. This could be another explanation as to why kidney cancer survivors continue to smoke at higher rates.

JAMA Network Open | Public Health
Therefore, immediate actions aimed at raising awareness about the association of smoking with certain cancer types and educating patients about the consequences of smoking after cancer treatment and survival are warranted.

Strengths and Limitations
Our study had strengths. First, we included cancers reported in the NHIS and classified them as SRCs and NSRCs per the 2014 Surgeon General report, unlike previous studies that have focused on state-level data of a limited number of SRCs and NSRCs 25 or used an older classification of SRCs. 26 To the best of our knowledge, this is the first nationwide study to comprehensively examine the prevalence and determinants of smoking behaviors in the US adult population of cancer survivors, since the 2014 Surgeon General report was issued. Second, beyond demographic characteristics, data about other comorbidities were included in our study.
The study limitations included the cross-sectional nature of the data, which could be affected by a number of biases, including the low-response rate bias and social desirability bias. Because mortality is higher among cancer survivors who continue to smoke, 6,7 this group of patients is less

Conclusions
Given what is known about the adverse effects of continued smoking after cancer diagnosis, 6 survivors of any cancer, particularly SRCs, are at elevated risk for developing further disease and should be prime targets for intervention. [47][48][49][50] These findings reinforce the importance of broad smoking cessation efforts among cancer survivors. [51][52][53] In the US adult population of cancer survivors, current smoking prevalence continues to be higher among SRC survivors compared with NSRC survivors. Although smoking cessation interventions are critically important for all cancer survivors, special efforts should target survivors of SRCs.