The construct of cognitive susceptibility to smoking, defined as lacking a firm commitment not to smoke in the future or if offered a cigarette by a friend, integrates behavioral intentions and expectations for future behavior [1]. Over the past decade, the construct has consistently demonstrated strong predictive abilities and has become widely accepted. In prospective studies, conducted in various geographic areas of the US among adolescents from differing ethnic backgrounds, adolescents identified as susceptible to smoking were two to three times more likely to have experimented with cigarettes than their non-susceptible peers at follow-up [2–5]. Moreover, results from a community-based tobacco prevention study indicated that susceptibility to smoking is amenable to interventions [6]. In this Texas-based study, an intervention reduced susceptibility to smoking among teens by 24.6%. Collectively, these studies suggest that identifying and characterizing susceptible adolescents who have never smoked will be critical to optimizing primary smoking prevention efforts among youth.

Mexican origin youth, aged 12 to 17 years, report higher rates of susceptibility to smoking (28.8%) compared to their non-Hispanic white (20.8%) and non-Hispanic black (23.0%) counterparts [7]. Moreover, results from two large population-based surveys of middle and high school students lend support to the predictive validity of the susceptibility construct [8, 9]. In Texas, Latino youth exhibit the highest rates of experimentation with cigarettes [8] and of current smoking [9], among all ethnic groups. In 2001 in Houston, Latino middle and high school students had the highest rate of experimenting with cigarettes (68.2%; 95% confidence interval [CI]: 64.2–72.2%) followed by whites (63.8%; 95% CI: 52.3–75.3%) and African Americans (53.4%; 95% CI: 48.4–58.4%). In addition, statewide in 2001, 14.5% (95% CI: 12.3–16.7%) of all Latino middle school students reported currently smoking compared to only 8.2% (95% CI: 6.3–10.2%) of white and 6.5% (95% CI: 4.5–8.4%) of African Americans middle school students. By high school, smoking rates among Latinos and whites were similar and significantly higher than smoking rates among African American and Asian students. Given that individuals of Mexican descent represent the most rapidly growing segment of the United States (US) population and account for almost 60% of the Hispanic population in the US [10], a close examination of risk factors associated with susceptibility to smoking among Mexican origin youth is timely and warranted.

Our goal in this analysis was to examine the associations between susceptibility to smoking and well-established sociodemographic, social, contextual, and behavioral risk factors associated with smoking initiation among a cohort of 11 to 13 year old Mexican origin youth in Houston, Texas. A better understanding of the relationship between susceptibility and these risk factors may facilitate the development of more effective primary smoking prevention programs targeted to this rapidly growing, understudied, and underserved population.

A total of 1,328, participants had completed the baseline cross-sectional survey. Of these, 129 participants (9.7%) were identified as ever-smokers and were excluded from the current analysis. The sample size of never-smokers available for analysis was 1,199. However, because another 12 participants had missing data on the temptations measure the sample size available for both logistic regression models was 1,187. Therefore to be consistent throughout, descriptive data tables are also based on the 1,187 participants. All analyses were conducted using SPSS.

All three aspects of peer influence were associated with susceptibility in the expected direction: Susceptibles were more likely to report that their peers strongly approve of their smoking than were non-susceptibles, to report that some of their friends smoke, and to report that at least one of their three best friends smoke. Similarly, the majority of the family influence variables were also associated with susceptibility. Susceptibles were more likely than non-susceptibles to believe that their parents strongly approve of their smoking and to have a father, mother, brother, or sister who currently smokes.

Among the school characteristics, only three were not associated with susceptibility: having taken a health class in which smoking was discussed, participating in extracurricular activities, and participating in school sports. Susceptibles were more likely to believe that students could smoke on campus and to endorse the idea that "smoking is cool." Susceptibles were less likely to endorse the idea that "smoking is stupid," and to believe their parents would be told if they were caught smoking at school. Susceptibles also reported lower subjective social status at school, and more detentions but reported higher grades than non-susceptibles. Three neighborhood characteristics were also associated with susceptibility: susceptibles were less likely to believe that if neighbors saw them smoking the neighbors would report their smoking to their parents, to have seen pro-messages, and a greater proportion of susceptibles considered it easy to buy cigarettes in their neighborhood or near school.

Overall, we found that 129 (9.7%) of the study participants reported that they had experimented with cigarettes. Of the remaining 1,187 never-smokers, 262 (22.1%) were susceptible to smoking. It is difficult to directly compare this percentage with those in other studies that have assessed susceptibility to smoking among Latinos because of between-study differences in the age and geographic location of the participants, as well as when and how the data were collected. For example, data collected from in-person household interviews from Mexican origin youth aged 12 to 17 in 2002 through 2004 reported a 28.8% susceptibility rate [7]. However, among a cohort of migrant Latino youth with a mean age of 13 years (standard deviation, ± 1.11), Elder et al. [15] reported a susceptibility rate of 35.6%, which is comparable to the rate of 31.6% we observed among the 13-year-old participants in our study (Table 2).

The principal aim of our study was to investigate the relationship between risk factors for smoking initiation and susceptibility to smoking. Most risk factors studied were associated with susceptibility to smoking, and all associations were in the expected direction. Therefore, first we focus on the results from the logistic regression models and second discuss associations that were not significant. The strongest independent risk factor for susceptibility to smoking in our study was holding positive expectations about smoking (OR = 4.85) (Table 4). Simmons-Morton et al. [16] too reported that youth are more likely to smoke if they think that smoking will yield socially beneficial outcomes, such as gaining more friends and gaining in popularity. Dalton et al. [17] found that both negative and positive outcome expectations are associated with susceptibility to smoking; however, in our study, negative expectations were not significantly associated with susceptibility to smoking. We used the outcome expectations measure developed by Dalton et al., but adjusted for a wider range of variables and still found the summary measure of positive expectations to be the strongest predictor of susceptibility to smoking. Our results, therefore, are consistent with previous findings [16, 17] and combined with the previous studies' results underscore the importance of variables, such as outcome expectations, as predictors of smoking initiation.

In general there is a robust association among sibling's smoking status [18]; in particular smoking by older siblings predicts smoking among younger siblings [19]. We found that having an older brother who currently smokes is associated with a two fold increased odds of being susceptibility to smoking (OR = 2.65). Recent studies demonstrated that having a parent who smokes or smoked is another strong and consistent predictor of smoking initiation [20–26], while parental expectations not to smoke are protective [27]. We examined the role of the fathers' smoking status independent from that of the mothers' smoking status. Although a higher percentage of fathers than mothers were current smokers, only the mothers' smoking status was associated with their children's susceptibility (OR = 1.92). Familismo, which refers to the belief that the needs of the family outweigh the needs of the individual, plays a central role in how Mexican and Mexican American families operate. One component of familismo is the obligation to provide material and emotional support to family members [28]. In many families, this is realized by the father working long hours while the mother stays at home to raise the children and take care of the home. As a result, children spend many more hours with their mothers than fathers and form a stronger emotional bond with their mothers. It is possible, therefore, that the mothers' behaviors, including smoking behaviors, exert a stronger influence on children than the fathers' behaviors. The findings that both mother and brother smoking are associated with susceptibility underscores the need to focus on the family context when developing primary prevention messages targeted to Mexican origin youth.

The roles of both peer smoking and perceived peer norms have been examined extensively in studies of adolescent smoking. Research has typically found a strong association between participants' smoking status and close friends' smoking status [29–31]. We also found that friends' smoking was a risk factor for susceptibility to smoking (OR = 2.19). Because people tend to choose their friends based on shared characteristics [32], one of which could be smoking status, having close friends who smoke does not mean that it was the friends who caused the participant to smoke. However, one aspect of the susceptibility to smoking construct is lacking a firm commitment not to smoke if offered a cigarette by a friend. In addition, our finding that peer approval of smoking is a risk factor for susceptibility to smoking (OR = 1.76) is consistent with previous studies [33, 34]. Together, these results suggest that the peer social context in which youths find themselves plays an important role in determining susceptibility to smoking and warrants further research in this population.

Previous research has demonstrated that older adolescents are significantly more likely to be susceptible to smoking [2, 15], have experimented with cigarettes [35], and smoke than their younger peers [36]. Our findings are consistent with this well-established risk factor.

Temptations to smoke are typically examined among current smokers and recent quitters [37, 38], though temptations to try smoking have been examined as predictors of smoking initiation [39]. To the best of our knowledge, no studies have examined the relationship between temptations to try cigarettes and susceptibility to smoking. The temptations measure assesses how tempted the participant is to try cigarettes in positive social situations, as a means of coping with negative affect, and to satisfy curiosity. We found that participants who reported more temptations to try cigarettes had a higher risk of susceptibility to smoking (OR = 1.36). This finding is consistent with previous studies that reported curiosity [40] and socializing with peers [41, 42] as two of the major reasons why adolescents start smoking.

To date, one study has examined the relationship between subjective social status and smoking, and none have used susceptibility to smoking as an outcome variable. Finkelstein et al. [43] found that among adolescents in grades 7 through 12, those with lower social status were at increased risk of smoking at baseline and initiating smoking during the subsequent year. Using the same measure of subjective social status [44], similarly we found that participants who reported lower perceived social status were at greater risk for being susceptible to smoking (OR = 1.13).

Finally, we found that having had one or more detentions at school was significantly associated with being susceptible to smoking (OR = 1.08). Our finding is consistent with previous research that demonstrates detentions are associated with being susceptible [2, 45] and smoking [46].

Some factors were not associated with susceptibility. Neither the child's country of birth nor the language spoken at home was related to susceptibility status. This is consistent with the finding of Gritz et al. [2] that level of cultural identification was not associated with smoking susceptibility among Latino youth. Parental educational attainment also was not related to children's susceptibility status, although previous studies have noted an inverse relationship between parental educational attainment and smoking [47, 48].

The majority of participants (81%) reported having taken a class in school in which the bad effects of smoking on health were discussed; however, this variable did not impact susceptibility. Evaluations of school-based interventions designed to prevent smoking have demonstrated that knowledge-based interventions alone do not impact behavior [49], while those that teach resistance skills do [50]. In the current study, we do not know what content was presented and discussed in the classes or whether resistance skills were taught.

Studies investigating the association between smoking and participation in school sports have yielded mixed results. While most have found participation in school sports to be protective against smoking [51–53], others have found school sports to be associated with higher rates of smoking [54]. We found no association. It is possible that we did not observe an association because according to local school district policy, only students in the 7^th grade and above participate in school sports. Most of our study participants were in the 5^th or 6^th grade when they enrolled in the study, and those who did answer the question about participation in school sports were likely answering about their participation in physical education classes rather than school sports.

Previous research has demonstrated that messages perceived as pro-tobacco and those perceived as anti-tobacco influence susceptibility to smoking [55] and that exposure to pro-tobacco media and advertising does increase susceptibility to smoking over time [56]. In the current study, these associations were not statistically significant.

Our study has both strengths and limitations. A strength of our study pertains to the large sample size, which allowed us to ascertain the number of co-occurring risk factors that tip the balance from non-susceptible to susceptible. This analysis, based on a cumulative risk score, revealed no differences in susceptibility among participants with zero or one risk factors. However, among the 72% of participants with two or more risk factors, the chances of being susceptible to smoking increased with the number of risk factors. Indeed, it was striking that the 169 participants (14.2% of participants) with five or more risk factors were over 20 times more likely to be susceptible compared to their peers with no risk factors.

Our study also generated some novel findings. To our knowledge, our study is the first to report that low subjective social status and temptations to smoke are associated with susceptibility to smoking. Future research will need to confirm these findings and determine if either or both risk factors generalize to other populations. Other strengths of the study include the fact that our participants were from a population-based cohort and included roughly equal numbers of girls and boys. In addition, we used validated measures and collected the data in the participants' homes using PDAs to ensure their privacy. A final strength of the study was our ability to recruit a large sample of Mexican origin participants, which is an understudied population.

One limitation of this study was that we were not able to examine the relationship between depression and susceptibility to smoking. We also did not have biochemical validation of the participants' smoking status (e.g., cotinine levels in saliva). However, we informed participants during the consent process that they might be selected to provide a saliva sample to check their smoking status; this "bogus pipeline" procedure has been shown to increase the validity of self-reported smoking status [57].

In summary, compared to non-susceptible participants, smoking-susceptible participants were more likely to hold more positive expectations about smoking, have a brother who smoked, report that their friends smoked, have a mother who smoked, believe their peers approve of their smoking, be older, report more temptations to try smoking, report lower subjective social status at school, and have had a detention during the school year. Overall, the strongest risk factor we identified was holding positive expectations about smoking, although both family- and school-based characteristics were important, too. These findings can be used to inform the development of culturally sensitive primary prevention programs.

School-based interventions that target positive expectations about smoking, the role that peers may play in promoting positive expectations, and potential differences in positive expectations among students with differing levels of social status, may be useful in this population. In addition, the findings that having a mother and a brother who smoke increase the risk of being susceptibility underscore the continued need to develop family- and community-based primary prevention programs.