Modeling the Underlying Predicting Factors of Tobacco Smoking among Adolescents.

BACKGROUND: With regard to the willing and starting tobacco smoking among young people in Iran. The aim of the study was to model the underlying factors in predicting the behavior of tobacco smoking among employed youth and students in Iran.
METHODS: In this analytical cross-sectional study, based on a random cluster sampling were recruited 850 high school students, employed and unemployed youth age ranged between 14 and 19 yr from Iran. The data of demographic and tobacco smoking related variables were acquired via a self-administered questionnaire. A series of univariate and multivariate logistic regressions were performed respectively for computing un-adjusted and adjusted Odds Ratios utilizing SPSS 17 software.
RESULTS: A number of 189 persons (25.6%) were smoker in the study and the mean smoking initiation age was 13.93 (SD= 2.21). In addition, smoker friend, peer persistence, leaving home, and smoking in one and six month ago were obtained as independent predictors of tobacco smoking.
CONCLUSIONS: The education programs on resistance skills against the persistence of the peers, improvement in health programs by governmental interference and policy should be implemented.

Introduction

Willingness and starting tobacco smoking among young people accompanied by its morbidity and mortality is one of the most worrisome aspects of smoking. Based on statistics from CDC, more than 80% of established adult smokers begin smoking before age 18 years. Based on CDC in 2009, 8.2% of middle school students and 23.9% of high school students reported current use of any tobacco product, and 5.2% of middle school students and 17.2% of high school students reported current use of cigarettes (1). If the trend in early initiation of cigarette smoking continues, approximately 5 million children aged <18 years who are living today will die prematurely because they began to smoke cigarettes during adolescence (2).

In Iran, the prevalence of the smoking was reported as 19.4% for youth in their lifetime (3) and the prevalence was reported as 28.9% among students who somehow smoked (4). The rate of smoking was estimated as 92.5% among addict youth in Iran (5). Various figures from 14.2% to 39% of smoking prevalence have been reported from different areas of the world (6–21).

Tobacco smoking in adolescence and young is one of the gateways and pathways as well as familiarity to the consumption of substances such as heroin, opium, hashish, cocaine, and stimulant drugs. It has a very strong role in changing and predicting the behaviors associated with drug in future (22, 23). Tobacco smoking in adolescence and young and continuing it, also declines in physical and educational performance in students, in addition to the attenuation and increasing the chance of tobacco related disease in adulthood (4). In addition, teens that start smoking at age 16 would likely to continue for 16 and 20 years of smoking in boys and girls respectively (24).

Employed and unemployed young people are of the major risky population groups, exposed to high-risk behaviors which are unfortunately neglected in health related researches and preventive interventions. The employment rate of school-age youth is different in various communities. Mitchell (1921) pointed to the estimates about the employment of school-aged youth; at the age of 16 years, 75 percent of young people are doing a work type activity. This rate at the age between 14 and 15 varies between 1 and 6 to between 1 and 2 in young population (25). Half of people in age between 14 and 17 years have expressed the incidence of at least one damage caused by the work (26).

A study showed higher rate of drug consumption among employed students, compared to un-employed students (27). One third of young people have started smoking in the workplace (28). Although drug abuse in the workplaces can lead in threatening consequences for the health of this age group (29), they have been studied less with regard to drug abuse in young people. Hence, it seems necessary to study the underlying effective factors cause smoking with the aim of executing the prevention and control programs.

The aim of this study was to model the underlying factors in predicting the tobacco smoking among employed and unemployed youth and students in Iran.

Material and Methods

Study participants

This analytical study was performed cross-sectionally on the 14 to 18 years age high school students and employed and unemployed youth in the Hamadan City, Iran in 2008.

For sample size determination, the primary information for the prevalence of tobacco smoking in the youth estimated in a pilot study by 100 subjects randomly selected form the study population. By a primary analysis, the least value of the Odds Ratio (OR) for investigating the relationship was 1.35. Considering a 95% confidence level, 80% Power, and utilizing G-Power software, the sample size resulted to 375 cases, which multiplied, by design effect of 1.75 and considering 20% for dropout, finally the sample size of the study was estimated 850 cases form, which 760 questionnaires entered in the analysis.

The sampling scheme was cluster sampling which was performed in the first to fourth grade high school boy and girl students. For employed youth, the working places were chosen randomly and the data of the youth willing to incorporate in the study were gathered. In addition, for unemployed youth, in the same time of the presence of the students in the school, the data of the subjects who willing to incorporate in the study were gathered in the streets and parks. The adolescents had to have permission by their parents to incorporate in the study.

A proportion to size sample selection scheme was used; based on the 65% (60% boys), 20% and 15% distribution of the study population for students, employed and unemployed youth respectively, the samples was chosen. A number of 8 high schools were randomly chosen from the fourfold educational district of the Hamadan city, and in these schools students were selected based on stratified random sampling schedule, so that in each of four educational grades, 20 boy student and 15 girl students were selected randomly and complete the checklist of the information. In addition, for employed and unemployed youth the sampling scheme was the convenience sampling procedure.

This study was conducted with approval from Hamadan University of Medical Sciences’ Institutional Review Board. Informed assent and consent were obtained from participants and to increase the validity of the responses, efforts were made to guarantee complete anonymity.

The data of demographic and tobacco smoking related variables were gathered through a self-administered questionnaire.

Demographics

Background collected data included: age, gender, employment status (students; employee; unemployment), educational course (natural sciences; mathematics; human sciences; technical and occupational; work and knowledge; first grade high school student; pre university), living status (living with both parents; one parent; alone), number of siblings, father’s and mother’s age, father’s and mother’s job, history of failing in school (yes/no), history of truancy (yes/no), history of leaving home (yes/no).

Tobacco smoking Related Factors

In addition to demographic characteristics, survey instruments included several items that were specifically designed to capture the tobacco smoking related behaviors. History of tobacco smoking in the past month (yes/no), and six months (yes/no), Pattern of tobacco smoking (daily or occasionally), having parents who smoked (never; occasionally; always), having friends who smoked (never; occasionally; always), having friends who had experienced substance (never; occasionally; always), history of smoking ceasing (yes/no), peer pressure to smoke (yes/no), persuasion enticement from friends to tobacco smoke (yes/no). In addition, the participants were asked to report their motivation to begin tobacco smoking (sensation seeking, having smoker friends, to take pleasure in, and sense of need to smoking, not to recall problems). Smokeless tobacco is not common among Iranians, as a result, cigarettes smoking is the only method of tobacco use in Iran.

Statistical Analysis

All statistical analyses were performed by SPSS 17 Statistical Analysis (SPSS Inc., Chicago, IL). Data were presented by mean (SD) and frequency (percent) for quantitative and qualitative variables respectively. Tobacco smoking status considered as the dependent outcome of interest by coding one for smokers and coding zero for non-smokers. For investigating the relationship of the demographic and tobacco smoking related variables with this outcome a series of simple and multiple logistic regressions were performed in the context of univariate and multivariate analyses respectively for computing un-adjusted and adjusted Odds Ratios (ORs) and their 95% confidence intervals. In the univariate analyses, each (demographic or tobacco smoking related) variable were entered separately and in the next step for multivariate analyses, those variable entered which were significant in the univariate analyses. In addition in the final step a backward elimination multiple logistic regression was performed to find the set of best predictors of the tobacco smoking. P-values <0.05 considered to be as significant.

Results

Study participant characteristics

From all the subjects in the study a number of 537 persons (70.7%) were student, 161 persons (21.2%) were employed and the rest of them (8.2%) were Un-Employed. also from all participants recruited in the study 388 (51.1%) were boys (Table 1, in addition for other background characteristics) .

Table 1:

Summary statistics for characteristics of study participants

Variables	Frequency	Percent
Employee Status
Student	537	70.7
Employed	161	21.2
Un-Employed	62	8.2
Sex
Boy	388	51.1
Girl	372	48.9
Major
natural sciences	82	15.3
mathematics	81	15.1
human sciences	63	11.8
technical and occupational	71	13.2
work and knowledge	100	18.7
first grade high school student	99	18.5
other	40	7.5
Father’s Job
Worker	248	36.1
Employee	131	19.1
Free job	246	35.8
Retired	48	7.0
Unemployed	14	2.0
Mother’s Job
Housewife	630	89.0
Employed	78	11.0
Living Status
Both parents	695	92.1
Father	9	1.2
Mother	44	5.8
Alone	7	.9
Age #	16.72 (1.15)
Father Age #	47.90 (7.38)
Mother Age #	41.79 (6.25)
Sister No #	1.59 (1.22)
Brother No #	1.61 (1.26)

For These variables, Mean (Std. Deviation) is reported

Tobacco smoking related variables

A number of 189 persons (25.6%) were smoker in the study and the mean smoking initiation age was 13.93 (SD= 2.21) in them. One hundred seventy four (23.8%) of subjects had fathers who were always smoker, 174 (23.8%) subjects had fathers who were occasionally smoker.

A number of 52 subjects (6.9%) had always smoker friends and 237 (31.3%) of subjects had occasionally smoker friends and the seven (0.9%) and 73 (9.8%) of the subjects had always and occasionally substance- user friends respectively. A total of 234 persons (31.0%) and 100 persons (14.8%) had persuasion and persistence to smoke by their friends. One hundred forty five (19.3%) and 373 (50.9%) of the subjects had the habit of leaving home and truancy, respectively. In the smoker subjects, 67 (35.3%) and 49 (25.8%) of the subjects enjoyed and somehow enjoyed from smoking respectively and in these subjects the consumption rate was 12.3% in daily smoking and the rest somehow smoking. The most frequent Motivation (31.2%) for smoking was Reject-Inability and also, Sense of Need, Taking Pleasure, Decreasing Pressures, Smoker Friends, and Sensation seeking was the reason for 12.1%, 20.1%, 21.6%, 5.0%, and 7.5% of the smoker subject and the rest of them (2.5%) have not recall the reason. Since one month ago and six month ago 89 (14.2%) and 130 (20.7%) of the subjects have already smoked, respectively. Of the smokers’ subjects, 72 persons (45.9%) had already tried to leave the smoke (Table 2).

Table 2:

Summary statistics for tobacco smoking related characteristics

Variables	Frequency	Percent
Tobacco smoking
No	549	74.4
Yes	189	25.6
Smoker Father
Always	174	23.8
Occasionally	174	23.8
Never	384	52.5
Smoker Friend
Always	52	6.9
Occasionally	237	31.3
Never	468	61.8
Substance - User Friend
Always	7	.9
Occasionally	73	9.8
Never	668	89.3
Persuasion
No	520	69.0
Yes	234	31.0
Peer pressure
No	574	85.2
Yes	100	14.8
Leaving Home
No	607	80.7
Yes	145	19.3
Truancy
No	360	49.1
Yes	373	50.9
Enjoy Smoking
Yes	67	35.4
No	74	39.2
Somehow	48	25.4
Consumption Rate
Daily	20	12.3
Occasionally	142	87.7
Motivation
Sense of need	24	12.1
Take Pleasure	40	20.1
Decreasing Pressures	43	21.6
Smoker Friends	10	5.0
Reject-Inability	62	31.2
Sensation seeking	15	7.5
not recall	5	2.5
Smoking during one month ago
No	539	85.8
Yes	89	14.2
Smoking during six month ago
No	498	79.3
Yes	130	20.7
Leaving Smoking
No	85	54.1
Yes	72	45.9
Smoking Initiation Age #	13.93 (2.21)

For this variable, Mean (Std. Deviation) is reported

Relationship of participant characteristics with tobacco smoking

Based on the results of simple logistic regression (Un-adjusted OR’s) for participant characteristics, employee status, age and sex were significantly related to tobacco smoking (All P<0.05); Unemployed and employed youth respectively had 55% more and 2.43 times the odds of smoking than students, with an increase of age in one year the odds of smoking increased by 27% and Boys had 2.45 times more than girls the odds of smoking. These variables were candidate as to enter in the multivariate analysis. However, none of the above-mentioned variables were significant in the multivariate analysis (All P>0.05) (Table 3).

Table 3:

Results of logistic regression for un-adjusted and adjusted Odds Ratios (ORs) for participant characteristics

Variables	Un-Adjusted				Adjusted
	OR	Lower	Upper	P-Value	OR	Lower	Upper	P- value
Employee Status
Un-Employed	1.55	1.04	2.31	.030	3.13	0.89	11.03	.076
Employed	2.43	1.40	4.20	.002	1.94	0.51	7.34	.331
Student	Referent	----	----	----	Referent	----	----	----
Age (yr)	1.27	1.10	1.48	.002	1.06	0.81	1.38	.668
Sex
Boy	2.45	1.73	3.46	<0.001	1.03	0.53	1.99	.937
Girl	Referent	----	----	----	Referent	----	----	----
Major
natural sciences	0.72	0.31	1.67	.439	----	----	----	----
mathematics	0.50	0.21	1.22	.129	----	----	----	----
human sciences	0.57	0.23	1.44	.236	----	----	----	----
technical and occupational	0.42	0.16	1.10	.078	----	----	----	----
work and knowledge	0.98	0.44	2.19	.962	----	----	----	----
first grade high school student	0.61	0.27	1.42	.253	----	----	----	----
other	Referent	----	----	----	Referent	----	----	----
Father Age	1.03	1.01	1.04	.007	1.01	0.97	1.04	.665
Mother Age	1.03	1.00	1.05	.044	1.00	0.96	1.05	.981
Father’s Job
Worker	.93	.28	3.05	.898	----	----	----	----
Employee	.86	.25	2.93	.808	----	----	----	----
Free job	.87	.26	2.86	.813	----	----	----	----
Retired	.83	.22	3.15	.788	----	----	----	----
Unemployed	Referent	----	----	----	----	----	----	----
Mother’s Job
Housewife	.867	.508	1.481	.602	----	----	----	----
Employed	Referent	----	----	----	----	----	----	----
Sister No	1.12	0.97	1.28	.117	----	----	----	----
Brother No	1.12	0.98	1.28	.098	----	----	----	----
Living Status
Both parents	1.37	0.15	12.37	.777	----	----	----	----
Father	1.14	0.08	16.95	.923	----	----	----	----
Mother	1.60	0.16	15.82	.688	----	----	----	----
Alone	Referent	----	----	----	----	----	----	----

OR: Odds ratio

Lower: Lower Bound for 95% C.I. for OR

Upper: Upper Bound for 95% C.I. for OR

Hosmer and Lemeshow Test showed an acceptable of model fit (Chi-square (8) = 7.454, P-Value = .489)

A total of 86.8% of subjects were correctly classified

Relationship of smoking related characteristics with tobacco smoking

Based on the results of simple logistic regression (Un-adjusted OR’s) for smoking related characteristics, smoker father, smoker friend, substance-user friend, suggestion, persist, leaving home, truancy, smoking in one and six month ago were significantly related to smoking (All P<0.05); These variables were candidate as to enter in the multivariate analysis.

However, of the above mentioned variables, smoker friend, substance-user friend, persist and smoking in six month ago, were significant in the multivariate analysis (All P<0.05). Based on these results, subjects with always and occasionally smoker friends had smoking odds of 5.28 and 2.92 times compared to never smoker friends respectively. Subjects with peer pressure had smoking odds of 2.64 times versus without pressure. Persons who had smoked during six month ago, had smoking odds of 19.48 times compared with those who had not smoking during six month ago (Table 4).

Table 4:

Results of logistic regression for un-adjusted and adjusted Odds Ratios (ORs) for tobacco smoking related variables

Variables	Un-Adjusted				Adjusted
	OR	Lower	Upper	P-Value	OR	Lower	Upper	P-Value
Smoker Father
Always	1.94	1.29	2.93	.001	1.11	0.55	2.26	.773
Occasionally	1.91	1.27	2.87	.002	1.86	0.91	3.78	.087
Never	Referent	----	----	----	Referent	----	----	----
Smoker Friend
Always	7.79	4.18	14.52	<0.001	5.28	1.62	17.18	.006
Occasionally	7.96	5.38	11.77	<0.001	2.92	1.43	5.97	.003
Never	Referent	----	----	----	Referent	----	----	----
Substance - User Friend
Always	2.85	0.63	12.89	.174	0.05	0.00	0.65	.022
Occasionally	7.60	4.49	12.85	<0.001	1.12	0.43	2.89	.814
Never	Referent	----	----	----	Referent	----	----	----
Persuasion
Yes	9.85	6.76	14.37	<0.001	1.93	0.90	4.13	.089
No	Referent	----	----	----	Referent	----	----	----
Peer Pressure
Yes	8.01	5.03	12.76	<0.001	2.64	1.13	6.18	.025
No	Referent	----	----	----	Referent	----	----	----
Smoking Initiation Age	1.05	0.82	1.33	.708	----	----	----	----
Leaving Home
Yes	4.21	2.87	6.19	<0.001	1.82	0.92	3.58	.084
No	Referent	----	----	----	Referent	----	----	----
Truancy
Yes	2.69	1.88	3.85	<0.001	1.26	.68	2.34	.459
No	Referent	----	----	----	Referent	----	----	----
Motivation
Sense of need	2.75	0.20	38.01	.450	----	----	----	----
Take Pleasure	1.75	0.16	18.97	.645	----	----	----	----
Decreasing Pressures	1.09	0.11	11.15	.940	----	----	----	----
Smoker Friends	-----	-----	-----	.999	----	----	----	----
Reject-Inability	1.96	0.19	20.15	.570	----	----	----	----
Sensation seeking	1.62	0.11	22.98	.719	----	----	----	----
not recall	Referent	----	----	----	----	----	----	----
Smoking during one month ago
Yes	40.31	18.90	85.95	<0.001	3.13	0.98	10.00	.055
No	Referent	----	----	----	Referent	----	----	----
Smoking during six month ago
Yes	37.85	21.45	66.80	<0.001	19.48	7.89	48.10	<0.001
No	Referent	----	----	----	Referent	----	----	----
Leaving Smoking
Yes	1.15	0.46	2.91	.767	----	----	----	----
No	Referent	----	----	----	----	----	----	----

OR: Odds ratio

Lower: Lower Bound for 95% C.I. for OR

Upper: Upper Bound for 95% C.I. for OR

Hosmer and Lemeshow Test showed a acceptable of model fit (Chi-square (8) = 7.454, P-Value = .489)

A total of 86.8% of subjects were correctly classified

In addition, a backward elimination modeling leads to a model with smoker friend, persist, leaving home and smoking in one and six month ago variables which were significant in the analysis (All P<0.05).

Discussions

This study demonstrated the prevalence and predictive ability of some demographic and tobacco smoking related characteristics with smoking among employed youth and students in Hamadan in 2008. In this study about 25.6% smokers were found which is compatible with some studies (6, 10, 15, 19), higher than that of other studies (8, 11, 13, 14, 16), and lower than reported by some other studies (7, 9, 12, 17, 18, 20, 21).

The results showed that smoker friend, persist, leaving home and smoking in one and six month ago, variables were obtained as independent predictors of smoking in this study. Subjects with always and occasionally smoker friends had smoking odds of 2.46 and 3.83 times compared to never smoker friends, respectively. Subjects with peer pressure had smoking odds of 3.47 times versus without pressure. Those youth who had leaving home, had 2.34 times odds of smoking than those without leaving home. Persons who had smoked during one and six month ago, had smoking odds of 16.79 and 7.06 times compared with those who had not smoking during one and six month ago (Table 5). These results have been supported by other studies (9–15, 17, 19, 21, 30–32).

Table 5:

Results for multiple logistic regressions Analysis of study variables based on Backward LR procedure

	OR	Lower	Upper	P-Value
Smoker Friend
Always	2.46	.73	8.31	.149
Occasionally	3.83	2.05	7.15	<0.001
Never	Referent	----	----	----
Peer Pressure
Yes	3.47	1.54	7.82	.003
No	Referent	----	----	----
Leaving Home
Yes	2.34	1.18	4.62	.015
No	Referent	----	----	----
Smoking during six month ago
Yes	16.79	6.51	43.28	<0.001
No	Referent	----	----	----
Smoking during one month ago
Yes	7.06	1.70	29.29	.007
No	Referent	----	----	----

OR: Odds ratio

Lower: Lower Bound for 95% C.I. for OR

Upper: Upper Bound for 95% C.I. for OR

Hosmer and Lemeshow Test showed a acceptable of model fit (Chi-square (4) = 9.343, P = .053)

A total of 87.1% of subjects were correctly classified

The persistence of peers and advising smoking to the friends is one of the most important risk factors predisposing to the smoking experience and in the next stages substance abuse among youth. Compared to the other age periods, young people are more and more under the influence and persistence of peers. In this regards, training the skills of “saying no” against the persistence of the peers have an important role in the prevention of substance usage. There are several studies in the field of substance abuse prevention which emphasis on dealing with peer pressure (33–35).

However, at a young age to ensure maximum effectiveness the implementation of the programs of abuse prevention in the transitional period, and especially before the transitional period and continuing it in the transitional period is more essential. Therefore, the design and evaluation of abuse prevention approaches should focused on adolescence and youth as the most vulnerable risk groups. Schools as the first places where their focus is youth should be used for prevention programs; positive teacher-relationships would reduce the risk of daily smoking (32, 36). In addition the program should not be unaware and possibly ignore the population of young people who drop out or are excluded from the study by any reason since they are much higher vulnerable and at risk than students. However, schools should accept their important and critical role in developing and enhancing students’ social (37) and life skills (38) in addition to formal training and with an improvement in health programs by the interference of ministry of education and ministry of health and medical education to prevent the increasing prevalence of the abuse.

On the other hand, some disciplines and suitable regulations should be provided for regular attendance of students in schools to leave little opportunities for them to have unwarrantable absence in the school (37, 39).

In Botevin (2000) theory (33), according to the complex nature of drugs, all abuse prevention programs designed for youth, in addition to the emphasis on individual approaches in school-based programs, should be combined with prevention approaches based on social influences (40–42) environmental factors, demographic factors (10), Life Skill training (38) and policy (43) such as reducing tobacco use in movies (44, 45), should be implemented to achieve the maximum effectiveness.

Obviously, with considering all young people of school age, especially those who are out of school for any reason, the health providers should concentrate on social inoculation strategies (46, 47).

Finally, the data were collected from youths in Hamadan city who might not be representative of all youths in our country. This was one of limitation of our study. Youths who have dropped out of school are more likely to smoke than youths who are in school, which this issue has not been considered in our study. As another limitation in this study, response and recall bias might have been introduced because the data are self-reported. One possible reason for non-significancy of the big ORs, for example in the “Employee Status”, is the decreased sample size in the multivariate analysis because of the simultaneous occurring of missing values in the variables entered in the model. This was another limitation of the study.

In addition, understanding the trends in the prevalence of cigarette smoking among youths enables policy makers to target prevention resources more effectively. Longitudinal studies are needed to encompass this aim.

Willing of youths to experience cigarette use as an important indicator of the effectiveness of tobacco control policies should be evaluated in Iranian population. With this regard, the health promotion plans should be focused on making negative social imaginations of tobacco smoking behaviors (48, 49). In addition, changes in social norms, behavioral intention, behavioral control, perceived severity and susceptibility (50, 51) attitude toward and intention to smoking might help fewer tobacco uses among youths, which can be evaluated and planned in a series of studies.

In addition further efforts are needed to decrease tobacco use among youths; restrictions on advertising, promotion, and availability of tobacco products to youths and tobacco tax increases, graphic health warnings on cigarette packages and in advertisements and restrict access to tobacco by youths, should be combined with full implementation of evidence-based, communitywide, comprehensive tobacco control policies.

Ethical considerations

Ethical issues (Including plagiarism, Informed Consent, misconduct, data fabrication and/or falsification, double publication and/or submission, redundancy, etc) have been completely observed by the authors.