|Home | About | Journals | Submit | Contact Us | Français|
Smoking cessation among cancer patients is critical for improving outcomes. Understanding factors associated with smoking abstinence after the diagnosis of cancer can provide direction to develop and test interventions to enhance cessation rates. The purpose of this study was to identify determinants of smoking outcomes among cancer patients.
Standardized questionnaires were used to collect data from 163 smokers or recent-quitters (quit ≤ 6 mo) at study entry of which 132 and 121 had data collected at 3 and 6-months. Biochemical verification was conducted with urinary cotinine and carbon monoxide. Descriptive statistics, Cronbach alpha coefficients, Pearson correlations, Fisher’s exact test, and multivariable logistic regression were used for analyses.
Seven-day-point-prevalence-abstinence (PPA) rates were 90/132 (68%) at 3-months; 46/71 (65%) among lung and 44/61 (72%) among head and neck cancer patients, whereas 7-day-PPA rates were 74/121 (61%) at 6-months; 31/58 (53%) among lung and 43/63 (68%) among head and neck cancer patients. Continuous abstinence rates were 63/89 (71%) at 3-months; 32/45 (71%) among lung and 31/44 (70%) among head and neck cancer patients, whereas continuous abstinence rates were 46/89 (52%) at 6-months; 18/45 (40%) among lung and 28/44 (64%) among head and neck cancer patients. Lower cancer-related, psychological and nicotine withdrawal symptoms were associated with increased 7-D-PPA abstinence rates at 3 and 6 months in univariate models. In multivariable models, however, decreased craving was significantly related with 7-day-PPA at 3-months and decreased craving and increased self-efficacy were associated with 7-D-PPA at 6-months. Decreased craving was the only factor associated with continuous abstinence at 6-months.
Smoking outcomes among lung and head and neck cancer patients appear to have remained the same over the last two decades despite the availability of an increased number of pharmacotherapy options to treat tobacco dependence. Decreased craving and increased self-efficacy were the most consistent factors associated with improved smoking outcomes but symptom control may also play a role in optimal management. Use of combined, and/or higher doses of pharmacotherapy along with behavioral interventions that increase self-efficacy and manage symptoms may promote enhanced cessation rates.
In order to optimize outcomes after a diagnosis of cancer, smoking cessation is an essential component of cancer control. Continued smoking after the diagnosis is associated with inferior outcomes such as increased recurrence and shortened survival [1–3].
Smoking abstinence rates ranged from 40 to 79% in lung cancer patients, whereas abstinence rates ranged from 65 to 79% for head and neck cancer patients in previous studies[4–11]. A notable proportion of smokers with cancer continue to smoke despite the potential adverse impact on their health. Many of the studies conducted have been retrospective and focused on medical and demographic factors. Symptom-related, behavioral, and cognitive factors that may influence smoking behaviors in adults with cancer have received less attention [5, 9–12].
Cancer patients often experience symptoms that may influence their ability to maintain positive behavioral change . Uncontrolled symptoms, pain, fatigue, nausea, depression, anxiety and nicotine withdrawal, were identified as triggers for smoking relapse in lung and head and neck cancer patients [11, 14–16].
Behavioral and cognitive variables may also influence the ability to maintain cessation among cancer patients. Two studies examined whether self-efficacy influenced abstinence and found that lung and head and neck cancer patients who relapsed were less confident in their ability to quit [9, 14]. Studies examining the relationship between alcohol use and smoking cessation among cancer patients have had mixed results. Two studies found a positive relationship between continued alcohol use and smoking whereas another study found no relationship between smoking cessation and moderate drinking [17–19]. Smokers who believe their health will be improved from stopping smoking are often more motivated to quit. One study found that head and neck cancer patients who attributed the cause of their cancer to tobacco use exhibited a lower likelihood of continued smoking if they indicated that their current smoking behavior influenced their future health .
Studies are needed to identify which factors are associated with maintaining abstinence in patients surviving lung and head and neck cancer to guide development of smoking cessation programs. Our study extends previous studies by conducting a prospective, longitudinal study and using biochemical measurement to verify smoking status. The specific aims for this study were to identify smoking outcomes at 3 and 6-month follow-up, and identify symptom-related, behavioral and cognitive factors associated with smoking outcomes at 3 and 6-months.
Non-small cell lung cancer and head and neck cancer patients who were diagnosed within the last 120-days and were current smokers or recent-quitters (defined as those who quit smoking ≤ 6 months) were eligible for this study. Recent-quitters were included in this study since they are at a nine-fold higher risk for smoking relapse as compared to those who quit for longer periods of time .
The study was approved by the Institutional Review Board. Written informed consent was obtained from those who agreed to participate in the study. Data were collected at baseline, 3 and 6-months later .
Demographic, tobacco and type of smoking cessation interventions received from health care providers [HCP] (defined as advice to quit smoking and assistance provided through recommendations for pharmacotherapy) and patient use of pharmacotherapy variables were collected through a self-report questionnaire[15, 22].
Clinical variables were collected through a medical chart review by a research coordinator. Stage of disease was classified according to the American Joint Cancer Committee guidelines for staging cancer.
Smoking: Current smoking was measured through 7-day-point-prevalence as well as continuous abstinence. Seven –day-point-prevalence-abstinence (7-D-PPA) was defined as whether the participant smoked within the last 7-days. Continuous abstinence was defined among non-smokers entering the study who remained abstinent at both 3 and 6 months after entry to the study .
Current smoking was determined through self-report and verification with urinary cotinine and carbon monoxide measurement using the vitalograph breath monitor . Current smokers were defined as participants who responded yes to smoking now, or responded no to smoking now but had a positive cotinine level of 3 or higher on the cotinine urinary dipstick . Patients who responded no to smoking now and reported current use of nicotine replacement treatment [NRT] were classified as non-smokers if their carbon monoxide was less than 10ppm .
Cancer-related symptoms were measured with the Symptom Distress Scale (SDS). Higher scores have been associated with decreased function, hospital re-admissions, and decreased survival among cancer patients [27, 28]. Cronbach alpha coefficient was .80 at baseline.
Psychological symptoms were measured with the Hospital Anxiety and Depression Scale  . Scores of 8 or higher correlate with clinical interviews which indicate moderate distress scores and scores of 11 or more indicate more severe distress. The Cronbach alpha coefficient for the anxiety subscale was 0.83 and 0.81 for the depression subscale at baseline.
Nicotine withdrawal symptoms were measured with the Minnesota Nicotine Withdrawal Scale . The items used in this study were irritability, increased appetite, sleep disturbance, difficulty concentrating, nervousness/tension, restlessness, and feeling blue or sad . The symptom of craving was analyzed separately. The Cronbach alpha coefficient was .79 at baseline.
Self efficacy was measured with the Self-Efficacy Short-Form . Higher scores indicate lower levels of self-efficacy. This questionnaire was used in previous research and was found to be predictive of smoking relapse . The Cronbach alpha coefficient was .93 at baseline.
Alcohol use was measured by the Alcohol Use Disorders Identification Test. Items are summed and a score of 8 or greater is associated with problem drinking . The Cronbach alpha coefficient was .86 at baseline.
Perceived control was measured by a single item that asked patients about their perceived level of control related to their future course of cancer. This question was found to be predictive of smoking relapse in patients with head and neck cancer .
Descriptive statistics were used to characterize patient characteristics. Cronbach alpha coefficient was used to assess the reliability of the Likert-type scales. Fisher’s-exact-test was used to examine whether there were any significant differences in abstinence rates and differences in the provision of smoking cessation intervention among the lung and head and neck cancer patients.
Logistic regression was used to model smoking outcomes. Potential predictors included demographic (age, gender, type of cancer, and stage of disease), symptom-related (cancer-related, psychological, nicotine withdrawal), behavioral (self efficacy and alcohol use), and cognitive variables (perceived control). Demographic variables were measured at baseline; symptom-related, behavioral, and cognitive variables were measured at baseline, 3, and 6-months later. Models for 7-D-PPA and continuous abstinence at 3-months were fit using variables (demographic, symptom-related, behavioral, and cognitive) measured at baseline, and models for 7-D-PPA and continuous abstinence at 6-months were fit using demographic variables measured at baseline and all other variables measured at 3-months after study entry. For 7-D-PPA, models were fit among all patients, and for continuous abstinence, models were fit among non-smokers at study entry.
Interactions between type of cancer and all other variables were examined, and no significant interactions were found; and the regression coefficients were numerically consistent for both cancers. Therefore, the cancer types were combined in the final analysis. Pearson correlation coefficients were used to examine predictor variables for evidence of multicollnearity. None of the variables were found to have correlations greater than .80. Backward model selection was used to identify significant predictors after examining univariate associations. The significance level used for entering effects was 0.20, and the significance level of the Wald chi-square for an effect to stay in the model was 0.10. All p-values were two-sided, and a p-value < 0.05 was considered significant. Hosmer and Lemeshow tests did not provide evidence of lack of fit to the data for all final multivariable models at the 0.05 significance level. All analyses were conducted using SAS (version 9.2).
One-thousand-seven- hundred-and-eighty-three participants were screened for eligibility. Of these, 282 were eligible, 180 consented, and 163 had complete data at entry and were included in this analysis. Reasons for non-eligibility were; never smoked (n=382), quit smoking > 1 year (n=662), quit smoking 7–12 months ago (n=45), ineligible diagnosis (n=217), non-English speaking (n=27), no follow up at cancer center (n=126), and too sick (n=42). Reasons for non-enrollment were; health limitations (n=4), no time (n=19), not interested (n=58), other (n=9), and no response (n=12). Prior to data collection 17 participants who signed consent cancelled due to health limitations (n=5), died (n=1), changed mind (n=5), limited time (n=4), or other reasons (n=2). Among the participants at baseline, 132 (81%) and 121 (74%) completed the survey at 3- and 6-month follow-ups. Reasons for attrition were died (n=21), health limitations (n=8), changed mind (n=7), lost to follow-up (n=3), or other reasons (n=3).
Table 1 describes participant characteristics. At entry to the study, 60 (37%) participants were smokers by self-report, whereas 74 (45%) were smokers by biochemical verification.
As Figure 1 illustrates, 7-D-PPA smoking rates peak at 3-months among the entire sample (n= 90/132, 68%) and for both lung (n= 46/71, 65%) and head and neck cancer patients (n=44/61, 72%). At 6-months, 7-D-PPA smoking rates decreased among lung cancer patients (n=31/58, 53%), whereas they decreased only slightly among head and neck cancer patients (n=43/63, 68%).
Among 45 non-smoking lung cancer patients at baseline, 32 (71%) patients remained continuously abstinent at 3-months, and 18/45 (40%) patients remained continuously abstinent at 6-months; among 44 non-smoking head and neck cancer patients, 31 (70%) were continuously abstinent at 3-months and 28(64%) at 6-months. No significant statistical differences in abstinence rates were noted across time or cancer type.
No significant differences in provision of cessation interventions were noted between the lung and head and neck cancer patients at baseline or 3-months. The majority of patients (78%, n=127) received advice to quit smoking and no significant differences were noted among the lung (82%, n=75) versus the head and neck cancer (73%, n=52) patients. Fifty-two percent (n=85) of patients received recommendations for assistance with quitting smoking and no significant differences were noted among the lung (58%, n=53) versus the head and neck cancer (45%, n= 32) patients. No differences were noted in provision of advice to quit smoking among the lung (25%, n=18) versus the head and neck cancer (31%, n=19) patients or in provision of recommendations for assistance with quitting smoking among the lung (18%, n=13) versus head and neck cancer (18%, n=11) patients at 3-months. However, only 26% (n=43/163) of patients were using pharmacotherapy at baseline and 29% (38/132) were using pharmacotherapy 3-months later and no differences were noted between the lung and head and neck cancer patients.
Factors were examined in logistic regression models for 7-D-PPA (see Table 2) and for continuous abstinence. In the univariate models, factors associated with 7-D-PPA at 3-months were higher self-efficacy (OR=.95 [95% CI, .91–.98]), lower craving (OR=.56 [95% CI, .43–.75]), lower anxiety (OR=.92 [95% CI, .85–1.0]), and lower depression (OR=.91 [95% CI, .82–1.0]). Factors associated with 7-D-PPA at 6-months included male gender (OR= .46 [95% CI, .21–.99]), absence of withdrawal symptoms (OR=.93 [95% CI, .86–1.0]), lower craving (OR=.44 [95% CI, .31–.62]), lower cancer-related symptom distress (OR=.91 [95% CI, .85–.97]), higher self-efficacy (OR=.90 [95% CI, .86–.95]), and lower craving (OR=.44 [95% CI, .31–.62]).
In the multivariable models, however, lower craving was the only factor that was associated with increased 7-D-PPA at 3-months (OR =.56 [95% CI, .43–.75]). Factors associated with 7-D PPA at 6-months were lower craving (OR = .54 [95% CI, .37–.80]) and higher self-efficacy (OR = .94 [95% CI, .89–1.0]).
In the univariate analysis, craving (OR = .62 [95% CI, .36–1.08]) and self-efficacy (OR = .95 [95% CI, .89–1.01]) were the only factors that approached significance for continuous abstinence at 3-months, whereas craving (OR = .52 [95% CI, .31–.88]) and self-efficacy (OR = .94 [95% CI, .94, .89–1.0]) were significant factors associated with continuous abstinence at 6-months. No multivariable analysis is needed for continuous abstinence at 3-months. In the multivariate analysis, factors associated with continuous at 6-months were lower craving (OR = .52 [95% CI, .31–.88]).
Smoking abstinence rates were similar for lung and head and neck cancer patients’ at 3-months but decreased among the lung cancer patients at 6-months. Although the differences in abstinence rates at 6-months were not statistically significant, the 24% difference in cessation rates was clinically significant. It is unclear why there was a decrease in abstinence in the lung cancer as compared to the head and neck cancer patients between 3 and 6-months. Both groups received similar levels of advice to quit smoking and assistance with quitting at baseline and 3-months later. HCP advice and providing assistance are powerful motivators to assist patients to quit smoking. When clinicians are trained in tobacco treatment and when systemic interventions are in place there is a five-fold likelihood of quitting smoking as compared to having no interventions in place. The high rate of advice to quit smoking in combination with a cancer diagnosis, may account for the high rates of smoking abstinence that were seen in this group of patients as compared to the general population . Despite the fact that a significant number of patients received suggestions for pharmacotherapy, only 26–29% of patients implemented these recommendations. This finding is consistent with other studies that have found that use of evidence-based treatments to assist with smoking cessation among smokers with cancer have not been adequately utilized [8, 22, 36]. Among highly addicted smokers more intensive smoking cessation interventions may be needed to enhance abstinence rates [37, 38].
The smoking abstinence rates from this study were similar to other studies that have examined smoking behaviors among head and neck cancer patients, but are somewhat lower among the lung cancer patients [4–6, 8, 10]. The differences found for smoking rates among lung cancer patients in our study could be related to several factors, which include that previous studies used self-report data alone to measure abstinence, used varied definitions of abstinence (i.e. 48-hours PPA, 7-day PPA), used varied time points (i.e. 6-months, 15-months) and varied samples of smokers (i.e. ever-smokers, former-smokers, recent-quitters, current smokers), which makes comparison across studies difficult. Integrating routine use of biochemical verification as well as coordinating efforts to achieve common definitions of smoking outcome measures among cancer patients may help enhance the ability of researchers to compare findings across studies [25, 39].
A striking finding of this study is that continuous abstinence rates appear to have remained the same over the last 15–20 years despite the availability of increased pharmacotherapy options to treat tobacco dependence [4, 8]. The most consistent factor associated with smoking abstinence was decreased craving. Although all tobacco dependence medications address baseline craving among quitters, short acting NRT, such as the nicotine lozenge, is the only medication that has a quick onset and can address cue-induced craving and may help quitters maintain abstinence if used with transdermal NRT . Schneider and colleagues  examined the effect of various combinations of NRT on the effects of craving and withdrawal among quitters in the general population and found that the lozenge and transdermal NRT combination was superior in suppressing reports of craving as compared to other combinations.
Increased self-efficacy was an important factor associated with continued abstinence. This finding highlights that inclusion of behavioral interventions along with pharmacotherapy are essential to enhance cessation rates. One potential intervention to enhance self-efficacy is tapering the amount of nicotine in cigarettes before the quit attempt along with pharmacotherapy. Benowitz and colleagues  found that having smokers’ progressively lower nicotine levels in cigarettes smoked was associated with a decrease in nicotine dependence and increased self-efficacy. Another study assessed the effects of a taper along with pre-cessation and post-cessation NRT versus a taper and post-cessation NRT versus a taper and placebo. Results from this study found that those who received the taper along with pre-cessation and post-cessation NRT had higher continuous abstinence rates .
Although symptoms were associated with 7-D-PPA rates at 3 and 6-months in univariate models, craving and self-efficacy were the only factors that entered the multivariate models. This finding underscores that craving and self-efficacy were powerful predictors of abstinence and are important components for future interventions. Given that symptoms were associated with 7-D-PPA abstinence rates in univariate models, incorporation of a symptom-based approach to cessation treatment may provide additive value in enhancing cessation outcomes but further testing is needed. .
The use of a relatively small sample size in a single institution is a weakness of the study. Another limitation was the short length of follow-up. We used 6-month follow-up as the outcome measure as this has been used in other clinical trials assessing short-term outcomes among cancer patients . In epidemiological studies, 1 to 2-years is used as the outcome measure to avoid misclassification of smokers as former-smokers.
Large numbers of patients were screened to identify those who were eligible. The major reason for exclusion in our study was related to smoking status (72%, n=1089/1501). We had a higher number of eligible participants (16%) as compared to the study by Martinez and colleagues who found that < 2% of cancer patients were eligible for their study. Our eligibility rate was most likely higher since our study was conducted in those with smoking-related malignancies and smoking rates tend to be higher as compared to those with non-smoking-related malignancies[48, 49]. Similar to our study, the main reason that participants were not eligible was because of smoking status (82%). This selection bias underscores the challenges researchers face with conducting clinical trials among smokers with cancer. Although our study had several limitations, we used a prospective, longitudinal study design and biochemical verification, which adds a unique perspective to the current literature about smoking behaviors after the diagnosis of cancer.
In summary, it appears that continuous abstinence rates among cancer patients have not changed over the last two-decades [4, 8]. Craving and self-efficacy were the main factors associated with abstinence but symptom control may also play a role in optimal management. Future studies need to evaluate whether combined and/ or higher doses of pharmacotherapy, enhanced symptom management and behavioral treatments enhance cessation rates among cancer patients.
Source of support: National Cancer Institute 1 K07 CA92696-02 and James B. Gillen Thoracic Oncology Research Fund, Dana-Farber Cancer Institute (Mary E. Cooley)
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest: None declared that are relevant.
Mary E. Cooley, Dana Farber Cancer Institute, 450 Brookline Ave, LW-521, Boston, MA, 02115.
Qian Wang, Dana Farber Cancer Institute, 450 Brookline Ave, Boston, MA, 02115.
Bruce E. Johnson, Dana Farber Cancer Institute, 450 Brookline Ave, Boston, MA, 02115.
Paul Catalano, Dana Farber Cancer Institute, 450 Brookline Ave, Boston, MA, 02115.
Robert I. Haddad, Dana Farber Cancer Institute, 450 Brookline Ave, Boston, MA, 02115.
Raphael Bueno, Brigham and Women’s Hospital, 75 Francis Street, Boston, MA, 02115.
Karen M. Emmons, Dana Farber Cancer Institute, 450 Brookline Ave, Boston, MA, 02115.