Journal of Adolescent Health 52 (2013) 724e730
www.jahonline.org Original article
Disposable Income With Tobacco Smoking Among Young Adolescents: A Multilevel Analysis Chuan-Yu Chen, Ph.D. a, b, *, I-Feng Lin, Ph.D. a, Song Lih Huang, M.D., Ph.D. a, Tzu-I. Tsai, Ph.D. c, and Ying-Ying Chen a a
Institute of Public Health, National Yang-Ming University, Taipei, Taiwan Center of Neuropsychiatric Research, National Health Research Institutes, Miaoli, Taiwan c School of Nursing, National Yang-Ming University, Taipei, Taiwan b
Article history: Received July 10, 2012; Accepted December 13, 2012 Keywords: Smoking; Adolescence; Allowance; School
A B S T R A C T
Purpose: Prior studies examining the connection between disposable income and adolescent smoking often yielded mixed results, partly due to the lack of consideration for contextual variables. In the present study, we sought to broaden understanding of disposable income on adolescent smoking behaviors via both absolute and relative perspectives in the school context. Methods: We obtained data from the 2010 Global Youth Tobacco Survey (GYTS) in Taiwan. Information concerning sociodemographics, disposable income, smoking history, and contextual smoking exposure (e.g., school) were assessed via self-report. Recent-onset smokers were defined as those who had their first cigarette within two years of the survey. Complex survey and multilevel analyses were carried out to estimate association. Results: Adolescents with higher monthly disposable income were 2w5 times more likely to start smoking and become regular smokers. Having the least disposable income in a class appeared linked with increased risk of tobacco initiation by 40% (95% confidence interval [CI]: 2%e91%). Pupils’ odds to start smoking were lowered to .70 when the majority of schoolmates had low disposable income (95% CI: .51e.99). Conclusions: Adolescent risk of smoking initiation may be differentially affected by individualand contextual-level absolute and relative disposable income. Future research is needed to delineate possible mechanisms underlying unfavorable health behaviors associated with disposable incomes in early adolescence. Ó 2013 Society for Adolescent Health and Medicine. All rights reserved.
Background Earlier-onset age of smoking is associated with not only the increased risk of dependence but also the lower likelihood of cessation in adulthood [1e3]. For that reason, adolescence is a critical developmental period for smoking interventions, such as * Address correspondence to: Chuan-Yu Chen, Ph.D., Associate Professor, National Yang-Ming University, Institute of Public Health, No.155, Sec.2, Linong Street, Taipei, 112 Taiwan, Medical Building II, Rm 204. E-mail address:
[email protected] (C.-Y. Chen).
IMPLICATIONS AND CONTRIBUTION
Adolescent risk of smoking initiation may be differentially affected by individual- and contextuallevel absolute and relative disposable income. Our evidence indicates that tobacco-control efforts should pay attention to the broader effects of disposable income in adolescence in order to reduce experimental tobacco use and to delay smoking onset.
measures aimed at reducing tobacco initiation or delaying onset age. Despite massive efforts to stamp out tobacco use and related health problems, progress in the underage population has stalled in recent years [2], which poses a challenge for tobacco-control programs. In Taiwan, as in much of the rest of the world, underage smoking has emerged as a serious public health problem. In 1997, Taiwan implemented its first antismoking policy, the “Tobacco Hazards Prevention Act (THPA).” The act restricted advertising and promotion of tobacco products and set a minimum age to purchase tobacco products, as well as regulated nonsmoking areas, required warning labels for
1054-139X/$ e see front matter Ó 2013 Society for Adolescent Health and Medicine. All rights reserved. http://dx.doi.org/10.1016/j.jadohealth.2012.12.009
C.-Y. Chen et al. / Journal of Adolescent Health 52 (2013) 724e730
cigarettes, and tar- and nicotine-content labeling. The THPA was revised in 2007, with major changes involving expanding nonsmoking areas to indoor public places and banning advertisement of tobacco products in print media. Following the implementation of a series of antitobacco policies, smoking rates have shown a gradual drop for the adult population (e.g., 27% in 2002 to 19.8% in 2010) [4]; however, the rate for adolescents remains relatively stable and may even be rising. The latest available Global Youth Tobacco Survey (GYTS) in Taiwan indicated that approximately 7%e8% of youth 13e15 years of age are current smokers, almost half the rate for those 16e18 years of age [5]. The connection between adolescent spending money and smoking consumption has received substantial scientific interest. By and large, young people’s disposable income comes from two main sources: allowance given by parents or guardians, and earnings from paid employment outside the home [6,7]. Although inconclusive, allowance amount and work payment have both been indicated for increased risk of tobacco smoking [8e11]. Several potential explanations have been proposed. First, since youngsters are especially sensitive to price changes in tobacco products [12,13], having more spending money may reflect the ability to afford tobacco purchases or to maintain smoking behaviors. Second, those with higher amounts of disposable income are more likely to engage in leisure activities wherein cigarettes or tobacco products are often easily available or smoking behaviors are commonly observed (e.g., Internet cafés or karaoke) [8]; therefore, their opportunities for exposure and access to tobacco products are elevated. Similar processes may also contribute to the higher smoking rate among employed adolescents who, in certain workplaces, may often have higher exposure to adults’ favorable norms toward smoking [8,14]. One research study also indicated the positive association between spending money with smoking behaviors among adolescents may be mediated through engaging in alcohol consumption or deviant behaviors [11]. Other than disposable income, research dating back at least two decades has shown that initiation and progress of smoking behaviors may be influenced by one’s socioeconomic status and surrounding social context. In adolescence, epidemiological evidence generally supported that an inverse relationship exists between parental socioeconomic status and adolescent smoking behaviors [15], and some researchers noted that lower socioeconomic status may partially exert effects through other social context factors, such as parental smoking [10]. Identified socialcontext-level factors affecting smoking in the underage population often cover three major domains (i.e., family, peers, and school) [9,16e18]. Although prior studies have often focused on just one social context, some have attempted to examine several social contexts simultaneously. For instance, using a social perspective, Ennett and colleagues detailed the pathways underlying the development of cigarette smoking in adolescence through the lens of modeling, closeness, social regulation, and strain [17]. The longitudinal analyses found that modelingdas manifested by the number of smoking families, smoking friends, and smoking schoolmatesdappeared to be the most salient actors in all microsystem social contexts under study, and the association estimate of modeling in the context of school emerged as the greatest. A major limitation of prior research on the effects of school context concerns the individual-level analytic approach. To the extent that students attending the same school are often similar in many ways (e.g., parental education or socioeconomic
725
status and neighborhood), as a result of selection and socialization, the lack of consideration of school-level interrelatedness and embeddedness may create spurious association estimates for school-related factors and lead to inappropriate inferences [17,19]. Although some earlier works have conceptualized young people’s disposable income as the socioeconomic status of their family, empirical data have accumulated to support the existence of a nonlinear or even reverse relationship [6,10,20]. At this point, most evidence on the connection linking disposable income and adolescent smoking behaviors comes from studies of European and North American youths [10,11,21]. Much remains to be learned about how disposable income is associated with youths’ smoking progression in other parts of the world [14]. We also note that a great number of published studies on disposable income have approached its effects solely from the standpoint of “absolute value,” under the hidden assumption that the disposable incomeesmoking relationship is homogeneous across different contexts (e.g., school district or community). As might be expected, each school may have its culture or norms toward adolescent allowance, and even the market for young consumers may vary across school districts [21,22]; the lack of consideration for possible intercontext variation and intracontext correlation may therefore lead to bias in interpretation. Also, like adults, adolescents have concerns for their social status compared with that of peers within social worlds in terms of academic achievement, material possessions, or popularity (i.e., positional concerns). Considering that greater amounts of spending money have been linked with more resources to participate in the consumption-oriented youth culture, young people are very likely to judge their status in relation to their peers within their social network based upon pocket money, just as income serves as an indicator for some in adult society [23]. Some studies have pointed to lower relative income (e.g., family and individual income) having negative effects on self-reported health [24e26]; however, few studies have probed this issue among adolescents via the perspective of “relative disposable income.” Utilizing a large survey with a nationally representative sample of youths in Taiwan, the present study sought to comprehensively examine the potential relationship between disposable income and smoking behaviors on the basis of school context. To broaden understanding of disposable income in the relative perspectives at both individual and school levels, we tested whether youth with lower disposable income relative to their classmates were more likely to engage in tobacco use, taking into account absolute allowance amount. Additionally, we tested whether pupils’ risk of smoking initiation was affected by school norms towards allowance. Methods Study design and population The Global Youth Tobacco Survey (GYTS) is a worldwide collaborative surveillance initiative that developed to enhance countries’ capacity to design, implement, and evaluate tobaccocontrol and prevention programs among youth [27]. This analysis makes use of data from the 2010 Global Youth Tobacco Survey (GYTS) in Taiwan, a continuing cross-sectional survey with nationally representative school-attending youth in Taiwan. Taiwan has nine years of compulsory education, comprised of sixyear elementary schools and three-year middle schools. In 2010, approximately 90.1% of middle-school students attended public
726
C.-Y. Chen et al. / Journal of Adolescent Health 52 (2013) 724e730
schools, and 9.9% attended private schools. For the 2010 GYTS, a total of 222 middle schools, both public and private, were selected via multistage sampling design, with urbanity level as sampling stratum (i.e., rural township, urban township, small city, and big city). The survey team randomly drew 1w2 classes from 7th to 9th grade (approximate ages 13e15 years), in which all the students were eligible to participate in the study (n ¼ 22,552). An introductory letter was sent to students/parents at least three days prior to the survey day, and both parental permission and youth assent were required. Taiwan’s GYTS was conducted during regular class hours; in most cases, 1w2 research assistants were present to deliver standardized instructions pertaining to research purpose, process, and anonymity assurance, as well as to maintain classroom order while teachers were absent from classrooms. The self-administered questionnaire consisted of 75 items inquiring after information such as smoking behaviors, access to cigarettes, exposure to environmental tobacco smoke, cessation of cigarette smoking, and sociodemographic background [28]. The responses were entered into a separate computer-scannable answer sheet. The survey took on average 20e30 minutes to complete; and the data collection was completed within a day for students in the same school. In 2010, a total of 21,282 students completed the survey questionnaires, an overall response rate of 94.37%. In the publicly released dataset, the only released schoollevel variable is encrypted school identification number. This study has been approved by the institutional review board of Taipei City Hospital. Measures Lifetime experience of cigarette smoking was assessed by two items “Including a puff or two, have you ever smoked a cigarette?” and “How old were you when you first tried smoking a cigarette?” Through the responses on the age of smoking initiation and age at time of survey, we classified participating youth into three subgroups: “nonsmokers (n ¼ 15,563);” “recent-onset smokers, if their ages at first smoking are not two years earlier than the ages at survey (n ¼ 1,824);” and “non-recent-onset smokers, if their ages at first smoking were three years or more than the ages at survey (n ¼ 3,037).”[29] Among non-recent-onset smokers, we defined current smokers as those who had smoked a cigarette at least 1 day during the 30 days prior to the survey (n ¼ 843), and quitters as those who reported no use of cigarettes in the 30 days preceding the survey (n ¼ 2,064). Monthly disposable income was assessed in a seven-category response: none, 1e499 New Taiwan Dollars (NTD, one USD is nearly 30 NTD), 500e1,499 NTD, 1,500e2,499 NTD, 2,500e3,499 NTD, 3,500e4,499 NTD, and 4,500 NTD or above. Considering that Taiwan’s Labor Standards Act prohibits employers from hiring individuals under 15, the main source of disposable income for this study population should not include workrelated pay. At the individual level, self-reported monthly disposable income was further recoded into the following collapsed categories: none, 1e499 NTD, 500e1,499 NTD, and 1,500 or above NTD. In addition, we ordered youngsters according to self-reported monthly disposable income (from high to low) within each class, and assigned the code of “1” for the variable “having the least disposable income at class” for those at the fifth quintile to reflect relative income position. For school-level characteristics, we calculated the proportion of students with low monthly disposable income (i.e., none or less than 500 NTD) to indicate general socioeconomic background for
each school. Finally, we sorted the 222 schools on the basis of percentage of students with low disposable income (from low to high), and used the fifth quintile ( 76%) as a cut point to create a binary variable to reflect whether the majority of students had low disposable income in schools. The present study also took into account an array of covariates thought to influence the occurrence of smoking initiation in early adolescence [30]. At the individual level, recent exposure experiences to paternal and maternal smoking, close friends’ smoking, and teachers’ smoking behaviors were assessed in a binary response. The present study utilized two proxy measures to evaluate school climate toward tobacco: prevalence of non-recentonset smokers and prevalence of smoking teachers observed by students’ self-report. Statistical analysis Because the GYTS adopted multistage probability sampling procedures, we used standard survey analysis to take into account the sampling weights (e.g., to compensate for variation in sample selection probability). We first examined the distribution of demographic and social factors among three subgroups defined by their smoking history. Next, we adopted polytomous regression to estimate the association linking allowance amount and allowance ranking with smoking status, while simultaneously adjusting for gender, grade, parental smoking, close friends’ smoking, and teacher smoking. Given that students in the same school may share the same contextual climates for tobacco use and norms toward disposable income, we turned to the multilevel random intercept logistic regression models to evaluate independent effects of monthly allowance amount, allowance ranking in class, and a school’s lower allowance composition on youngsters’ smoking initiation. In this series of analyses, students were analyzed as the first level (n ¼ 17,387) and schools as the second level (n ¼ 222). For the models accounting for the occurrence of recent onset smoking (reference outcome: nonsmokers), we first included only individual-level characteristics in model I (including parental, peers’ and teachers’ smoking), followed by the inclusion of all school-level characteristics simultaneously (Model II). All statistical analyses were carried out by STATA 10.0 with the commands of svymlogit and xtmelogit (Stata Corp. Inc., College Station, TX, United States). Results More than one-fifth had tried smoking on at least one occasion in their lifetimes, with 8.4% being recent-onset smokers (n ¼ 1,824). Table 1 summarizes the distribution of sociodemographics and smoking behaviors of parents, close friends, and teachers among students across groups defined by smoking history. As compared with nonsmokers, recent-onset and non-recent-onset smokers were more likely to be male, have a higher weekly disposable income, have smoking parents, have smoking friends, and have ever observed teachers smoking in schools (all p < .001). Furthermore, a higher proportion of nonsmoking youth were the ones with the least disposable income at class as compared with that of recent-onset smokers (13%) and non-recent-onset smokers (15%). Almost half of recent-onset smokers had used cigarettes in the past 30 days; the corresponding estimate was 27% for the nonrecent-onset smokers.
C.-Y. Chen et al. / Journal of Adolescent Health 52 (2013) 724e730 Table 1 Selected characteristics of middle-school-attending adolescents in Taiwan, by smoking experience Variablesa
Gender Female Male Grade 7th 8th 9th Paternal smoking No Yes Maternal smoking No Yes Close friends’ smoking No or few Most or all Observing teachers smoking in school No Yes Disposable income (NTD) 0 1e499 500e1,499 1,500 or above Having the least disposable income at class No Yes Smoking in past 30 days No Yes
Nonsmokers
Smokers
n ¼ 15,563 (74.2%wt)
Recent-onset n ¼ 1,824 (8.4%wt)
Non-recentonset n ¼ 3,037 (13.5%wt)
8,228 (51.8) 7,149 (47)
652 (34.9) 1,152 (64.1)
1,006 (32.7) 1,956 (64.8)
5,858 (35.6) 5,246 (33.2) 4,459 (31.3)
535 (26.9) 717 (38.2) 572 (34.9)
832 (25.8) 997 (30.7) 1,208 (43.5)
8,250 (53.2) 7,313 (46.8)
593 (33.2) 1,231 (66.8)
1,042 (35.1) 1,995 (64.9)
14,132 (91.4) 1,431 (9)
1,413 (78.7) 411 (21.3)
2,387 (78.9) 650 (21.9)
15,052 (96.8) 342 (2.1)
1,382 (75.6) 384 (21.6)
2,531 (83.5) 411 (13.1)
14,144 (91.1) 1,419 (8.9)
1,523 (83) 301 (17)
2,629 (86.4) 408 (13.6)
4,644 6,207 3,151 1,543
(29.8) (39.4) (20.3) (10.5)
12,215 (77.9) 3,330 (22) e
278 660 459 426
(15) (34.4) (25.1) (25.4)
573 1,058 775 637
(19.7) (32.8) (25.4) (22)
1,598 (87.3) 225 (12.6)
2,605 (84.7) 428 (15.3)
882 (49.3) 854 (46.4)
2,064 (68.7) 843 (26.9)
NTD ¼ New Taiwan Dollar (1 USD ¼ 30 NTDs). a The columns may not add up to 100% due to missing data.
With statistical adjustment for listed variables simultaneously by means of polytomous logistic regression analyses, we then assessed the associations linking individual characteristics with four groups of smoking experiences (i.e., nonsmokers as reference group, recent-onset smokers, former smokers, and current smokers) (Table 2). Youths whose friends were mostly or all smokers generally had 10 times and 18 times greater odds, respectively, to initiate and continue smoking. Having a monthly disposable income of 1,500 NTD or above was associated with 5-fold increased odds of smoking initiation and continuation, whereas having the least disposable income at class was significantly associated with smoking initiation only (aOR ¼ 1.58, 95% CI ¼ 1.11e2.24). The intraclass correlations of recent-onset smoking at the class and school levels were estimated at .15 and .09, respectively. With a focus on recent-onset smokers and nonsmokers, Table 3 shows the association estimates linking individual and contextual characteristics with the risk of smoking initiation. By and large, having higher disposable income was associated with a 2w4-fold risk of smoking initiation; whereas having the least disposable income at class increased the risk by 37% (Model I). With school-level characteristics statistically adjusted, the individual-level association estimates for disposable income amount and disposable income rank at class were slightly increased (Model II). As to school characteristics, for each 1%
727
increase in students’ and teachers’ smoking rates, youngsters’ odds of smoking initiation were elevated by 4% and 2%, respectively. Meanwhile, the odds were lowered by 30% when the majority of attending students had none or lower allowance (95% CI ¼ .51e.99, p < .05). Discussion Building upon the recent Global Youth Tobacco Survey (GYTS) conducted in Taiwan, our results demonstrated three key findings regarding disposable income and smoking behaviors among adolescents in Taiwan. First, a higher absolute amount of disposable income was positively associated with the occurrence of smoking behaviors, regardless of onset age and current smoking status. The association estimates appeared diminished for non-recent-onset smokers who had quit. Second, having the least disposable income at class may increase a youth’s risk of smoking initiation by 58%; yet this relationship did not exist for non-recent-onset smoking. Finally, with individual-level demographics, disposable income, and social modeling taken into account, youths’ excessive risk of smoking initiation can be moderately explained by their school’s compositional factors (e.g., rates of schoolmates with non-recent-onset smoking and with low disposable income). Youths’ risk of smoking initiation was generally elevated as the percentage of schoolmates with low disposable income increased; yet some protective effects seem to emerge when having low disposable income was a norm in school. Although to our knowledge this is one of few studies to investigate the association linking disposable income with smoking behaviors in early adolescence via the absolute/relative perspective, there are important comparisons with other studies [10,11,20,21]. Several prior studies found that absolute amount of disposable income had a dose-response relationship with regular smoking or smoking amount; some researchers have attributed this connection to the purchasing ability for tobacco products or exposure to smoking workplace, particularly among the experienced smokers [8,11,14]. However, for the study population another plausible explanation is that having higher disposable income may reflect less parental engagement (e.g., giving pocket money instead of having dinner at home or buying goods together), poorer parental supervision (e.g., knowing how children spend their money and time), or even possibly lower parental socioeconomic status [20]. We found support for our hypothesis in that youngsters who had lower disposable income position in class were more likely to initiate cigarette smoking. Given the important effects of pocket money on perceived adolescent popularity or peer status [23], it is possible that youngsters may seek to ameliorate status-related anxiety or stress or to gain the acceptance of their classmates via cigarette smoking [31e33]. This piece of evidence, if replicated by other studies, may reinforce the need to integrate social asset enhancement while delivering tobacco control and health promotion programs targeting early-to-middle adolescents [34]. The results derived from multilevel analyses were generally in accordance with prior research [17,18,30,35e37], suggesting that pupils’ risk of smoking initiation is affected by school-level compositional factors (i.e., smoking teachers and schoolmates). Possible mechanisms accounting for these findings may include (1) collective socialization: on campuses wherein smoking behaviors of teachers and schoolmates are observable, social norms toward smoking may be more favorable, or informal social
728
C.-Y. Chen et al. / Journal of Adolescent Health 52 (2013) 724e730
Table 2 Estimated association linking individual-level characteristics with smoking status among middle school students Variablesa
Gender (ref: female) Male Grade (ref: 7th) 8th 9th Paternal smoking (ref: none) Yes Maternal smoking (ref: none) Yes Close friends’ smoking (ref: none or few) Most or all Observing teachers smoking in schools (ref: none) Yes Monthly disposable income (ref: 0 NTD) 1e499 500e1,499 1,500 or above Having the least disposable income at class (ref: no) Yes
Recent-onset smokers
Non-recent-onset smokers
(n ¼ 1,824) Adjusted ORs
Quitters (n ¼ 2,064) Adjusted ORs
Current smokers (n ¼ 843) Adjusted ORs
2 (1.73e2.31)***
1.81 (1.60e2.05)***
3.79 (2.95e4.82)***
1.45 (1.22e1.72)*** 1.35 (1.11e1.64)**
1.24 (1.07e1.43)** 1.89 (1.63e2.21)***
1.20 (.94e1.54) 1.68 (1.29e2.18)***
1.85 (1.63e2.11)***
1.66 (1.46e1.89)***
2.10 (1.74e2.54)***
1.96 (1.64e2.36)***
1.95 (1.64e2.31)***
2.25 (1.80e2.81)***
10.3 (8.25e12.9)***
2.24 (1.72e2.92)***
17.8 (14.2e22.2)***
1.93 (1.60e2.32)***
1.41 (1.20e1.66)***
1.84 (1.41e2.40)***
2.34 (1.69e3.23)*** 2.98 (2.11e4)*** 4.87 (3.39e7)***
1.19 (.96e1.49) 1.65 (1.29e2.11)*** 1.98 (1.52e2.56)***
1.86 (1.17e2.96)** 2.38 (1.46e3.86)*** 4.89 (2.97e8.04)***
1.58 (1.11e2.24)**
1.13 (.89e1.45)
.89 (.51e1.57)
aOR ¼ adjusted odds ratio; CI ¼ confidence interval; NTD ¼ New Taiwan Dollar; OR ¼ odds ratio. a Association estimates were obtained from multinominal logistic regression (weighted data with Taylor series linearization), with all listed variables in the models and nonsmokers as the reference outcome. ** p < .01. *** p < .001.
controls against smoking may be less tangible; (2) social contagion: youngsters’ frequent exposure to smoking schoolmates may foster the transition of definitions favorable to cigarette smoking and even to start smoking upon positive reinforcement
(e.g., peer acceptance); and (3) institutional process: when smoking students are present or smoke-free campus policies are loosely enforced, pupils’ ease of access to cigarettes can be elevated [17,35e38]. Additionally, although our findings
Table 3 Individual and school characteristics in relation to recent-onset smoking among middle school students (n ¼ 17,387) Variablesa Individual characteristics Gender (ref: female) Male Grade (ref: 7th) 8th 9th Paternal smoking (ref: none) Yes Maternal smoking (ref: none) Yes Close friends’ smoking (ref: none or few) Most or all Observing teachers smoking in schools (ref: none) Yes Monthly disposable income (ref: 0 NTD) 1e499 500e1,499 1,500 or above Having the least disposable income at class (ref: no) Yes School characteristics % Of non-recent-onset smoking schoolmates % Of teacher smoking % Of low disposable income Majority with low disposable income (Y/N)
Univariate OR (95% CI)
Model Ib aOR (95% CI)
Model IIc aOR (95% CI)
2.03 (1.83e2.25)***
2.03 (1.82e2.27)***
2.01 (1.80e2.25)***
1.53 (1.36e1.73)*** 1.43 (1.26e1.62)***
1.43 (1.26e1.63)*** 1.34 (1.17e1.54)***
1.43 (1.25e1.63)*** 1.34 (1.16e1.53)***
2.23 (2e2.47)***
1.80 (1.61e2.02)***
1.75 (1.56e1.97)***
2 (1.74e2.30)***
1.94 (1.69e2.24)***
11.8 (10e13.9)***
9.43 (7.94e11.2)***
9.22 (7.77e10.9)***
1.83 (1.59e2.10)***
1.65 (1.42e1.92)***
1.60 (1.37e1.86)***
1.75 (1.51e2.03)*** 2.57 (2.19e3.01)*** 4.98 (4.22e5.89)***
2.03 (1.52e2.71)*** 2.71 (2.01e3.65)*** 4.44 (3.27e6.03)***
2.05 (1.53e2.74)*** 2.80 (2.07e3.78)*** 4.60 (3.38e6.24)***
1.37 (1.01e1.88)*
1.40 (1.02e1.91)*
2.77 (2.44e3.14)***
.49 (.42e.56)*** 1.06 1.05 1 .73
(1.04e1.07)*** (1.03e1.06)*** (.99e1.02) (.51e1.03)
aOR ¼ adjusted odds ratio; CI ¼ confidence interval; NTD ¼ New Taiwan Dollar; OR ¼ odds ratio. a The odds ratios were obtained via the random intercept logistic regression, with nonsmokers as reference group in the outcome. b Model I: adjusted for all individual-level characteristics. c Model II: adjusted for all individual-level and school-characteristics. * p < .05. *** p < .001.
1.04 1.02 1.02 .70
(1.03e1.05)*** (1e1.03)* (1.01e1.04)*** (.51e.99)*
C.-Y. Chen et al. / Journal of Adolescent Health 52 (2013) 724e730
concerning the relationships between percentage of lowdisposable-income students and pupils’ smoking initiation seem less intuitive, this observation is also likely a reflection of relative disposable income position at the macro-contextual level. That is, when youngsters’ anxiety or stress toward having no or little pocket money might be diminished when the majority of schoolmates are in the same disposable income condition. These results highlight the importance of understanding school attributes (e.g., smoking behaviors of compositional members) and should be considered when devising school-based smoking curricula or tobacco-control programs. Finally, our study found that initiation and maintenance of cigarette smoking in early adolescence are strongly associated with the smoking behaviors of parents and peers, which are also regarded as two of the most important social contexts shaping healthy development in early life [16e18,21,23]. Also, the association estimates unanimously follow the gradient of nonrecent-onset current smoking, recent-onset smoking, and smoking quitters, highlighting the salient effects of smoking parents, peers, and teachers in maintaining and initiating smoking. These notable differences may demonstrate that risk factors for tobacco and other substance-use problems should be examined via a stage-dependent approach and suggest the possibility that reciprocity may operate between social context and smoking behaviors [38]. The present study has several potential limitations. First, since the sample recruitment was based on a school-based survey, the generalizability of the results to nonschool samples (e.g., dropouts or nonattendees) in Taiwan may be limited, although middle school education is compulsory in Taiwan. Similarly, the generalizability across societal or country boundary should be warranted considering possible variation in educational system, antitobacco policies, and social norms toward youth’s disposable income. Second, all measures concerning smoking experiences relied solely on self-report. Even though great efforts were made to encourage completeness and validity (e.g., ensuring anonymity and filling-in answer in separate scannable card), there is a possibility that young participants may skip questions or underreport their smoking behaviors [39]. Finally, perhaps the most serious methodological issue can be traced to the cross-sectional and retrospective nature of our study design, which creates opportunities for errors in specification of temporal sequences that may open up opportunities for reciprocal influences. For example, the amount of disposable income was measured in terms of recent status (e.g., the past year), whereas the assessment of smoking experiences can date back to middle childhood. With efforts to restrict the outcome to recent-onset smoking, some potential problems from temporal relationship have been overcome. Still, the observed results should be interpreted with caution for causal directions. Longitudinal studies with multiple assessments of disposable income and smoking behaviors from late childhood into adolescence would help clarify the directionality of the observed findings. Our findings with respect to absolute/relative disposable income at both the individual and school levels may be worthy of future attention in relation to public health implications and research directions. Beyond well-established family and peer determinants (e.g., parental and peer smoking), our evidence indicates that those engaged in tobacco-control efforts should pay attention to the broader effects of disposable income in adolescence in order to reduce experimental tobacco use or to delay the onset of smoking. Importantly, such efforts should not be
729
restricted to individual-level risk factors without taking into account the compositional and structural characteristics of social context surrounding developing youth. Finally, no one-size-fits-all approach exists for antitobacco policies or prevention programs. Successful solutions should be based on individual- (e.g., developmental stage and disposable income), family- (e.g., parental smoking and engagement), and school-related characteristics (e.g., smoking-free policies and smoking norms) [40]. Human participation protection This study has been approved by the institutional review board of Taipei City Hospital. Acknowledgments This work was supported by the Bureau of Health Promotion (BHP), Department of Health, Executive Yuan, Taiwan (grant CCY 10009-1). We thank those who collected and managed the Global Youth Tobacco Survey dataset in the BHP. Dr. Chen was supported by a grant from the Ministry of Education, Aim for the Top University Plan. The funder had no role in the design or conduct of the study; collection, management, analysis, or interpretation of the data; or preparation, review, or approval of the manuscript. References [1] Dierker L, Swendsen J, Rose J, et al. Transitions to regular smoking and nicotine dependence in the Adolescent National Comorbidity Survey (NCSA). Ann Behav Med 2012;43:394e401. [2] U.S. Department of Health and Human Services. Preventing Tobacco Use Among Youth and Young Adults: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 2012. [3] Kandel DB, Hu MC, Griesler PC, Schaffran C. On the development of nicotine dependence in adolescence. Drug Alcohol Depend 2007;91:26e39. [4] Department of Health. Adult Smoking Behavior Surveillance System (ASBS). Available at: http://tobacco.bhp.doh.gov.tw/Show.aspx?MenuId¼581. [5] Department of Health. Global Youth Tobacco Survey. Available at: http://tobacco.bhp.doh.gov.tw/Show.aspx?MenuId¼582. [6] Darling H, Reeder AI, McGee R, Williams S. Brief report: Disposable income, and spending on fast food, alcohol, cigarettes, and gambling by New Zealand secondary school students. J Adolesc 2006;29:837e43. [7] Miller J, Yung S. The role of allowances in adolescent socialization. Youth Soc 1990;22:137e59. [8] Chen CY, Chen WC, Lew-Ting CY, et al. Employment experience in relation to alcohol, tobacco, and betel nut use among youth in Taiwan. Drug Alcohol Depend 2006;84:273e80. [9] Gilreath TD, Chaix B, King G, et al. Multi-level influence of school norms on tobacco use in South Africa: An ecometric consideration of group differences. Cancer Causes Control 2012;23:S27e36. [10] Soteriades ES, DiFranza JR. Parent’s socioeconomic status, adolescents’ disposable income, and adolescents’ smoking status in Massachusetts. Am J Public Health 2003;93:1155e60. [11] Zhang B, Cartmill C, Ferrence R. The role of spending money and drinking alcohol in adolescent smoking. Addiction 2008;103:310e9. [12] Ross H, Chaloupka FJ. The effect of cigarette prices on youth smoking. Health Econ 2003;12:217e30. [13] Kostova D, Ross H, Blecher E, Markowitz S. Is youth smoking responsive to cigarette prices? Evidence from low- and middle-income countries. Tob Control 2011;20:419e24. [14] Do YK, Finkelstein EA. Youth employment, income, and smoking initiation: Results from Korean panel data. J Adolesc Health 2012;51:226e32. [15] Goodman E, Huang B. Socioeconomic status, depressive symptoms, and adolescent substance use. Arch Pediatr Adolesc Med 2002;156:448e53. [16] Frohlich KL, Potvin L, Gauvin L, Chabot P. Youth smoking initiation: Disentangling context from composition. Health Place 2002;8:155e66. [17] Ennett ST, Foshee VA, Bauman KE, et al. A social contextual analysis of youth cigarette smoking development. Nicotine Tob Res 2010;12:950e62.
730
C.-Y. Chen et al. / Journal of Adolescent Health 52 (2013) 724e730
[18] Lakon CM, Hipp JR, Timberlake DS. The social context of adolescent smoking: A systems perspective. Am J Public Health 2010;100:1218e28. [19] Duncan C, Jones K, Moon G. Context, composition and heterogeneity: Using multilevel models in health research. Soc Sci Med 1998;46:97e117. [20] West P, Sweeting J, Young R, Robins M. A material paradox: Socioeconomic status, young people’s disposable income and consumer culture. J Youth Stud 2006;9:437e62. [21] Hughes SK, Hughes K, Atkinson AM, et al. Smoking behaviours, access to cigarettes and relationships with alcohol in 15- and 16-year-old schoolchildren. Eur J Public Health 2011;21:8e14. [22] Mistry R, McCarthy WJ, de Vogli R, et al. Adolescent smoking risk increases with wider income gaps between rich and poor. Health Place 2011;17: 222e9. [23] Sweeting H, West P, Young R, Kelly S. Dimensions of adolescent subjective social status within the school community: Description and correlates. J Adolesc 2011;34:493e504. [24] Kawachi I, Kennedy BP. Income inequality and health: Pathways and mechanisms. Health Serv Res 1999;34:215e27. [25] Subramanyam M, Kawachi I, Berkman L, Subramanian SV. Relative deprivation in income and self-rated health in the United States. Soc Sci Med 2009;69:327e34. [26] Sun P, Unger JB, Palmer P, et al. Relative income inequality and selected health outcomes in urban Chinese youth. Soc Sci Med 2012; 74:84e91. [27] Global Youth Tobacco Survey Collaborative Group. Tobacco use among youth: A cross country comparison. Tob Control 2002;11:252e70. [28] Chen PL, Chiou HY, Chen YH. Chinese version of the Global Youth Tobacco Survey: Cross-cultural instrument adaptation. BMC Public Health 2008;8:144. [29] Chen CY, Dormitzer CM, Bejarano J, Anthony JC. Religiosity and the earliest stages of adolescent drug involvement in seven countries of Latin America. Am J Epidemiol 2004;159:1180e8.
[30] Wen M, Van Duker H, Olson LM. Social contexts of regular smoking in adolescence: Towards a multidimensional ecological model. J Adolesc 2009;32:671e92. [31] Barton J, Chassin L, Presson CC, Sherman SJ. Social image factors as motivators of smoking initiation in early and middle adolescence. Child Dev 1982;53:1499e511. [32] Weiss JW, Palmer PH, Chou CP, et al. Association between psychological factors and adolescent smoking in seven cities in China. Int J Behav Med 2008;15:149e56. [33] Wiltshire S, Amos A, Haw S, McNeill A. Image, context and transition: Smoking in mid-to-late adolescence. J Adolesc 2005;28:603e17. [34] Eccles J, Gootman JA. Community programs to promote youth development. Board on Children, Youth, and Families, Division of Behavioral and Social Sciences and Education, National Research Council & Institute of Medicine. Washington, DC: National Academies Press; 2002. [35] Mrug S, Gaines J, Su W, Windle M. School-level substance use: Effects on early adolescents’ alcohol, tobacco, and marijuana use. J Stud Alcohol Drugs 2010;71:488e95. [36] Chen YH, Chen PL, Huang WG, et al. Association between social climate for smoking and youth smoking behaviors in Taiwan: An ecological study. Int J Nurs Stud 2010;47:1253e61. [37] Leatherdale ST, Cameron R, Brown KS, et al. The influence of friends, family, and older peers on smoking among elementary school students: Low-risk students in high-risk schools. Prev Med 2006;42:218e22. [38] Mercken L, Candel M, Willems P, de Vries H. Disentangling social selection and social influence effects on adolescent smoking: The importance of reciprocity in friendships. Addiction 2007;102:1483e92. [39] Brener ND, Billy JO, Grady WR. Assessment of factors affecting the validity of self-reported health-risk behavior among adolescents: Evidence from the scientific literature. J Adolesc Health 2003;33:436e57. [40] Kaestle CE, Wiles BB. Targeting high-risk neighborhoods for tobacco prevention education in schools. Am J Public Health 2010;100:1708e13.