INTRODUCTION
Cigarette smoking is a major contributing factor of lung cancer, cancers of other sites, chronic obstructive pulmonary dis- ease (COPD) and heart diseases (World Health Organization 2005). In spite of this, smoking prevalence among Malay- sian adults remains high. In 2006, ap- proximately half of male adults aged 18 years and above were smokers (Institute for Public Health 2008). Ten thousand deaths from diseases related to smoking have been reported annually in Malaysia since 1980 (Chua 2003). It is estimated that the mortality rate may rise to 30 000 annually by 2020 if there is no drastic change in the current trend of smoking among Malaysians (Disease Control Division 2003). The Gateway Drug theory suggests that the use of ‘soft’ drugs such as cigarettes can lead to the use of other ‘harder’ drugs (Fleming et al. 1989). Pre- vious studies have also shown that smokers are more likely to risk harming themselves further by resorting to other abusive substances compared to non- smokers (Fleming et al. 1989; Kandel et al. 1992). Therefore, at this critical mo- ment, it comes as no surprise that re- ducing the prevalence and incidence of smoking is a priority of the Ministry of Health. Studies have shown that smoking is a learned behavior which generally starts during adolescence. It has been reported that 80-90% of current smokers start smoking as adolescents, and be- come regular smokers before the age of 18 (USDHHS 1994; Health Canada 2005). Adolescents who smoke are either unwilling to quit smoking, have low moti- vation to quit or are uninterested to quit because they do not think quitting is im- portant (Balch 1998; Pallonen 1998; Mermelstein 2003). These factors have driven researchers to focus on strategies, at the earliest possible opportunity, at identifying those adolescents who face the greatest risk of becoming habitual smokers in the future. Such a move will certainly help smoking-prevention pro- grammes to be targeted to this group and hopefully, reduce smoking initiation among them.
A ‘susceptibility-to-smoking’ measure has been developed to identify adoles- cents who are at risk of initiating smoking (Pierce et al. 1996). It measures the cog- nitive commitment of adolescents not to smoke and adolescents’ capability to resist offers from friends to smoke. Unger et al. (1997) and Difranza et al. (2006)
reported that respondents who were sus- ceptible to smoking were two to three times at higher risk of becoming experi- mental smokers after a period of two years. Jackson (1998) revealed that cog- nitive susceptibility was the most signifi- cant predictor of smoking initiation among elementary school students after one year. Those who were susceptible to smoking were 80% more likely to initiate smoking compared to non-susceptible respondents. Cognitive susceptibility has been reported to be a more significant independent predictor of experimental smoking compared to parental smoking and peer smoking (Pierce et al. 1996). Huang et al (2005) in their longitudinal study among adolescents aged from 14 to 17 years old found that susceptible adolescents were also two to three times more likely to initiate smoking after two years. The simplicity of the measure lends itself as a useful and efficient pre- dictor of smoking initiation. Although the “susceptibility” measure has been shown to be an effective tool for identifying and predicting adolescent smoking initiation, it has not, hitherto, been applied in Ma- laysia. The aims of this study were to determine the validity of the measure and to identify the independent factors for smoking initiation among school children in Kota Tinggi district, Johor, Malaysia.
MATERIALS AND METHODS
Study Design
This was a longitudinal study conducted from 2008 to 2009. Baseline data was collected in May 2008 and follow up in June 2009. Respondents were second- ary school students in Forms 1, 2 and 4 in the district of Kota Tinggi in Johor. The study was approved by the Ministry of Education Malaysia and the Johor State Education Department while ethics ap- proval was given by the Ethics commit- tee, Ministry of Health, Malaysia. This project was a collaborative effort between the Institute for Medical Research (IMR) and the Kota Tinggi District Health Office.
Study protocol
Passive consent was obtained from stu- dents’ parents prior to their participation in the study. Respondents were informed that participation was voluntary. Students who agreed to participate were asked to put down their signatures on the ques- tionnaire and were given assurance of anonymity. In addition, teachers were not allowed to observe the students while they were completing the questionnaires. The questionnaires were self-adminis- tered by the students. A detailed expla- nation on the questionnaire was given to the respondents. Further clarification on any of the items was given when needed. At the end of the session, completed questionnaires were packed into enve- lopes, and the envelopes sealed in the presence of the respondents.
Sampling
Multi-stage sampling was used in this study. Schools were stratified by urban, rural and FELDA settlement areas. Six schools were randomly selected from FELDA settlement areas, three from town areas and one from a rural area. Stu- dents in the selected schools were then stratified by Forms: Form 1, Form 2 and Form 4. The sampling frame was then obtained from the schools’ administra- tors, and simple random sampling was used to select respondents using random numbers generated by Epi Info version 6.04d.
A sample size of 2700 was calculated based on smoking incidence of 3.5% for Forms 1 and 2, and 6% for Form 4 stu- dents, maximum error of 3%, design ef- fect of 0.67, intra-class correlation coeffi- cient of 0.5, average proportion of stu- dents per strata of 0.33 (prevalence of smoking was assumed to be equal among all three strata), an additional 30% extra to cater for non response and another 30% for exclusion of smokers. (Kalton 1983; Scheaffer et al. 1990; United Nation 2005). The number of stu- dents selected from each school was de- rived as a proportion to the total student population in the respective schools. Students who were non-smokers at baseline were followed up after one year. Each student was given a unique identifi- cation (ID) number at the beginning of the study, and the same ID was allocated to the student during the follow-up study.
Study instrument
The same self-administered question- naire was used at baseline and at follow- up. The instrument used in this study was adapted from previous studies (Hanjeet et al. 2001; Lim et al. 2006) and the questionnaire included demographic and psychosocial factors. These were: per- centage of peers who smoke, family members who smoke, percentage of friends who smoke, and perceptions of both parental and social acceptance of smoking. Perceptions of parental and social acceptance of smoking were as- sessed using a 7-point Likert type scale in which lower scores indicated lower acceptance by parents and society to- wards smoking behavior among adoles- cents. Perceived benefits of smoking and perceived negative effects of smoking were assessed using 12 questions com- prising six questions for each. Low scores in the “Perceived benefits of smoking” domain indicate positive per- ception toward smoking, while low scores in the “Perceived negative effects of smoking” domain indicate highly negative perception of respondents towards smoking. The ‘Susceptibility-to-Smoking’ measure was adapted from Pierce et al. (1996). It consisted of two questions:
a) Do you think you will smoke a ciga- rette in the next year?
b) If one of your best friends were to offer you a cigarette, would you smoke?
The choice of answers was: (a) Yes, (b) Probably yes, (c) Probably no, and (d) Not at all. If the respondent answered ‘Not at all’ to both questions, they were categorised as ‘Not susceptible to smok- ing’ while those who gave other answers to both questions were categorised as ‘Susceptible to smoking’.
The dependent variable in this study was smoking initiation, with ‘Yes’ or ‘No’ responses, which was measured at one- year follow-up. Smoking initiation was defined as ‘Have smoked at least once in the last 30 days’.
Statistical analysis
Data were entered and analysed using SPSS version 16 (SPSS, 2007). Chi square and independent t test were em- ployed in attrition analysis. The attrition analysis was done to determine the ex- tent to which respondents who dropped out of the study differed from the respon- dents who continued to participate at fol- low up. These tests were also used for associations between categorical variables and continuous data with the de- pendent variable that is initiation of smoking after one year. Independent va- riables are gender, susceptible to smoking, percentage of friends who smoked, form of study, schooling area, father smoking, parents’ reaction toward smoking, perceived benefits of smoking and perceived negative effects of smoking. Variables with p value equal or less than 0.25 from the univariate analysis were included into binary logistic regression to determine the independence of susceptibility-to-smoking variable from other predictor variables. Analysis for two way interactions reveals a non-significant inte- raction between all the variables in the model. Model fit was checked using Hosmer-Lemeshow goodness of fit test. The p value was not significant indicating the model fitted the data (p = 0.640). The level of significance was set at p- value of less than 0.05 (2-sided).
RESULTS
Table 1 shows that, out of the 1763 non smokers at baseline in 2008, 1288 (73.1%) responded at follow -up in 2009. Attrition analysis revealed that three in- dependent variables, i.e. form, gender and locality, were significantly different between those who responded at follow- up and those who dropped out. Attrition was higher among female (29.1%), Form Four students (30.7%) and FELDA set- tlement areas (29.5%)
From the univariate analysis, these va- riables were associated with the initiation of smoking: percentages of friends who smoke, father’s smoking status, suscep- tibility-to smoking, and gender. Gender and susceptibility-to-smoking were the most significant independent variables associated with smoking initiation, that is, at 21.7% for male respondents, com- pared to females at 2.4%. Respondents who were susceptible were approx- imately four times more likely to initiate smoking compared to their counterparts who were not susceptible, that is, 31.9% as compared to 8.2 %. It also shows that initiators, firstly, believed more strongly in the perceived positive consequences of smoking, however misguided this may be, secondly, perceived more accep- tance of parents towards their smoking behavior, and thirdly, perceived less negative consequences of smoking. The two groups were not significantly different on opinions of (societal) perceptions to- wards adolescents smoking (Table 2). Table 3 shows the results of multiple lo- gistic regression analysis. Susceptibility to smoking remained a significant risk factor for smoking initiation after a period of one year, after controlling for the po- tential confounding effects of gender, school locality, percentage of friends who smoke, father smoking, parental accep- tance of smoking, and belief in the posi- tive and negative consequences of smoking. The odds ratio was, however, attenuated by the inclusion of the other independent variables.
DISCUSSION
To our knowledge, this is the first publi- cation towards establishing ‘Susceptibil- ity-to-Smoking’ as a reliable predictor for identifying adolescents who have the potential to initiate smoking in Malaysia. The response rate was almost similar to previous longitudinal studies (Conrad et al. 1992). The study revealed that this measure is a reliable predictor for identi- fying adolescents who are at risk of smoking initiation, as well as, for predict- ing smoking initiation among non- smokers at baseline. It shows a substan- tial effect size and it predicts smoking initiation independently of other predic- tors such as, peer smoking, perceived benefits and negative effects of smoking, family members (parents and sibling) smoking, social norms of smoking, gender and age, and factors that have been shown to contribute to smoking in previous studies (Kalton 1983; Kremers et al. 2001; Maxwell 2002; Wen et al. 2005; Wilkinson et al. 2007; Otten et al. 2009) Almost a third of the adolescents who did not smoke but were susceptible to smoking at baseline initiated smoking after a period of one year.
Susceptible adolescents were 3.7 times more likely to be smokers at follow-up. This finding is similar to that reported by Pierce et al. (1996), Unger et al. (1997) and Huang et al. (2005). The effect size of the ‘Susceptibility-to-Smoking’ meas- ure was attenuated after other in dependent variables were included in the multivariate analysis model. This sug- gests that a better model may be devel- oped from a combination of this construct with other predictor variables such as gender and smoking status of fathers.
These findings show that the measure can be used in adolescent prevention programmes to reduce smoking initiation among adolescents in the long term. Needless to say, co-operation between school and health authorities is essential, if not vital, to the successful implementa- tion of an effective long term strategy to reduce smoking initiation among our youths. School authorities can carry out surveys using the construct to determine the susceptibility level among those who are not smoking and at the same time, health departments can play a role as training providers. The activities that are to be carried out should be based on the susceptibility level of the adolescent. Dis- semination of information on the health hazards of smoking and training to incul- cate the ability to resist smoking offers from peers should be delivered to non- susceptible groups (Johnson 1990). The formation of peer support groups in order to develop the required attitude towards the smoking habit among those who are susceptible is recommended. Care should be taken to prevent stigmatisation of those who have been identified as susceptible to initiate smoking and also to prevent self-fulfilling prophecies that trigger adolescents to initiate smoking (Unger et al. 1997).
The measure can also be used by health workers at adolescent health clin-ics to identify susceptible adolescents, where appropriate advice and proactive counseling can be given to reduce their risk of initiating smoking in the future. Adolescents who are not susceptible to smoking should be given advice and support against smoking, including tips and methods on how to improve their refusal skills. A study showed that advice given by health professionals regarding a health subject is more acceptable and effective for attitudinal behavioural changes (Morgan et al. 1996).
There was a limitation in this study. The high attrition of respondents after a pe- riod of one year may reduce the internal and external consistency of the findings. Differential attrition (more females, Form Four students and FELDA settlement areas dropped out) indicated that the follow-up sample may not represent the population of interest which is all non-cur- rent smoking secondary school students in the district.
The study reveals that the susceptibility measure was a significant independent predictor to smoking initiation among adolescents. It may also be used as a screening instrument to identify those adolescents who do not smoke but who are at increased risk to smoking initiation.
School personnel especially those in the secondary sector can use this measure to identify adolescents who are at risk of initiating smoking as the very first pre- ventive step towards battling the smoking scourge. The use of the measure might reduce the incidence of smoking among our adolescents, and ultimately contri- bute to lowering morbidity and mortality resulting from smoking-related diseases.
ACKNOWLEDGEMENTS
We would like to thank the Director-Gen- eral of Health Malaysia for his permission to publish this paper. We would also like to thank the Ministry of Education, the schools involved in the study and all those who assisted in data collection and management for their support and cooperation.