Over 90% of our sample of childhood and young adult cancer survivors smoked within the past seven days, but overall the smoking rate of the sample was light, with the vast majority smoking less than one pack per week. However, almost half of the participants were classified as nicotine dependent. Participants in the current study smoked fewer cigarettes than has been found in other studies of childhood cancer survivors [15, 18]. Although we can not definitely determine the reason for this difference, there are several possible explanations. First, several years have passed since the last report of childhood cancer survivors' smoking behavior was published. National data suggests that the prevalence of tobacco use has decreased during this period for the United States population and for cancer survivors as a collective group [31, 32]. Smoking rates for childhood and young adult cancer survivors could have declined at an accelerated rate, which would be consistent with historical data on tobacco use trends among young adults . Second, data on the late effects of cancer treatment are increasingly becoming available. Childhood and young adult cancer survivors may be responding to these data by limiting their tobacco use to decrease their risk for future health problems. Further, survivors who are nicotine dependent may be more likely to reduce their smoking rate, rather than quit smoking, to minimize their risk for late effects, which explains why we saw high rates of nicotine dependence (65%) despite a low smoking rate. Third, the low smoking rate may reflect survivors' efforts to quit as data suggest that some smokers will decrease the amount of cigarettes they smoke before quitting completely. Finally, the low smoking rate may also reflect selection bias such that lighter smokers may have been more willing to participate in a smoking cessation intervention. Regardless, there is no research on suggesting that even light smoking is safe for childhood cancer survivors; thus, it is important to help this high-risk population refrain from tobacco use completely.
The majority of our sample was motivated to quit smoking--22% were in contemplation and 63% were in preparation. However, less than one half of participants in contemplation and preparation were very or extremely confident that they could quit. Over half of survivors made at least one quit attempt in the previous 12 months and about 20% made four or more quit attempts. Collectively, these data suggest that childhood and young adult cancer survivors in this study are motivated to quit, and may benefit from a targeted and tailored smoking cessation intervention to help them build self-efficacy and quit successfully. We anticipate that the web-based format of PFH-2 will be an accessible delivery channel for this population because of the widespread access and utilization of the Internet.
In our multivariate analysis of the baseline data, stage of change and self-efficacy were associated with smoking behaviors. Cancer survivors in an early stage of change for smoking cessation made fewer attempts to quit smoking. Further, survivors with low self-efficacy to quit smoking smoked more cigarettes per week, and there was evidence that they were more likely to be nicotine dependent. These results are consistent with research finding that a later stage of change and high self-efficacy were consistently associated with smoking cessation  as well as literature on the importance of tailoring smoking cessation interventions to smokers' stage of change .
Aspects of the social environment were also associated with smoking behavior. Survivors who smoked more were more likely to have received social support to quit smoking from their family and oncologist. It is possible that survivors who smoked more may have triggered family and health care providers to encourage them to quit smoking. In fact, encouragement to quit smoking by one's regular doctor or oncologist was associated with a making more quit attempts. These data are consistent with Emmons et al., (2003) who found that social support for quitting smoking was associated with higher smoking rates but more quit attempts .
Other evidence for the importance of the social environment came from our data on household and workplace smoking policies. Survivors with a less stringent smoking policy at work were more likely to smoke more, whereas survivors with a less stringent smoking policy at home were more likely to be nicotine dependent. Workplace smoking bans make it more difficult to smoke during the day and may facilitate social norms around not smoking. Moreover, people employed in workplaces with smoking bans are more likely to smoke fewer cigarettes and think about quitting . Household smoking bans may reflect pressure from the smoker's family for them to quit or the smoker's readiness to quit. Regardless, smokers who enact household smoking bans are more likely to make successful quit attempts and have lower risk of relapse [37, 38].
Survivors' psychosocial characteristics were also associated with their smoking behavior. Survivors who did not fear cancer recurrence were more likely to be nicotine dependent; and survivors who felt more certain that they would experience a serious future health event made fewer quit attempts. These findings appear somewhat contradictory. The literature suggests that survivors with lower risk perception may be less motivated to change their smoking behavior than smokers with a greater perceived risk of future health problems , which is consistent with our finding for cancer recurrence. However, it is possible that survivors who feel relatively certain that they will experience a future health event are demonstrating fatalistic beliefs about their future health potential, which are associated with a increased likelihood of smoking and an earlier stage of change for smoking cessation .
Survivors who had been hospitalized for a medical problem in the last year were less likely to be relatively heavy smokers but more likely to be nicotine dependent. In addition, survivors who lacked a healthcare provider for non-emergency care were more likely to be nicotine dependent. Survivors who do not have access to a primary care provider (PCP) may not receive consistent messages about the importance of making healthy lifestyle choices such as quitting smoking. In addition, survivors who lack a PCP may not have the same level of access to pharmacotherapy and behavioral interventions for smoking cessation as survivors with a PCP.
Our findings provide important information about factors that are associated with smoking behavior among childhood cancer survivors. In addition, our results indicate that smoking cessation interventions developed for childhood cancer survivors need to enhance motivation to quit by providing accurate information about the relationship between smoking and future health problems as well as the fact that smoking is particularly harmful for cancer survivors. Coupled with efforts to increase survivors' perceived susceptibility to tobacco-related health problems, interventions must help survivors build self-efficacy for quitting smoking by addressing their psychological barriers to quitting, increasing social support for smoking cessation, and targeting aspects of the social environment that can constrain smoking behavior (e.g., household smoking policy). In addition, even though study participants tended to be light smokers, there are high rates of nicotine dependence in this group, suggesting that pharmacotherapy may facilitate quitting.
This study had several limitations. First, we were unable to contact 24% of survivors who were identified by participating sites and did not opt out, and we do not know whether these individuals were eligible for the study. Our study population is young adults who tend to be very mobile and likely living in a different location from the time of their initial treatment. In many cases, the participating sites did not have accurate addresses on file for their patients. We made a concerted effort to locate these individuals by using internet searches and other telephone and mail-based tracing approaches. However, despite this effort, in some cases, our strategies were unsuccessful. Second, 48% of eligible survivors declined to participate, which introduces selection bias into the study; however, a 52% recruitment rate among a sample of smokers who have not sought out smoking cessation treatment is reasonable. Third, our sample may not be representative of all childhood and young adult cancer survivors who smoke because lighter smokers may have been more willing to participate in this study than heavier smokers. Results should be interpreted accordingly. Third, our sample was predominately white, which limits our ability to generalize to other racial and ethnic groups, although it reflects the racial/ethnic characteristics of the participating treatment sites.
This study also had several important strengths. First, study participants were recruited from five pediatric and young adult cancer treatment facilities in the United States and Canada. Thus, our results have higher external validity than if recruited from a single site. Second, we collected data on participants' level of motivation to quit smoking and confidence in their ability to quit smoking in order to tailor intervention content. This will help us evaluate whether the intervention affected theoretical mediators of smoking cessation and better understand the pathways through which the PFH -2 content affected smoking behavior.