Study design, participants, and setting
We conducted a cross-sectional school-based survey with high-school students aged 13 to 19 years. Adolescents from all 14 public high schools of the Federal Institute of Education Science and Technology of Goiás, from 13 municipalities in the Midwest region of Brazil, including one of the Brazilian capital cities, Goiânia, were selected as a convenience sample to be part of this survey. The study's response rate was 99.7%. Detailed descriptions about the setting and sample size calculations were reported in a previous publication [15].
Data collection, instrument, and procedures
Data were collected at the beginning or end of the students’ regular classes. The research instrument, a self-administered printed questionnaire specifically designed for this survey, consisted of 67 closed questions concerning an array of sociodemographic, psychosocial, smoking behaviours and other health-related factors. The methodological steps of the questionnaire design and testing were previously described [15]. It included original questions and others compiled from previous studies, among which is the SOC questionnaire [12].
Students took approximately 20 minutes to complete the questionnaire. To improve response rate, minimize missing data and prevent information bias, the researcher who collected the data provided information about the study and gave prior instructions to the adolescents, using printed banners, on how to fill out the questionnaire. For example, the banners were used to explain which questions were aimed at smokers and should be left blank by non-smokers, and vice-versa.
Smoking status
The adolescents answered the following yes-or-no questions: (i) ‘Have you ever tried smoking cigarettes, even one or two puffs?’, (ii) ‘Currently, do you smoke cigarettes? (Select yes if you smoked at least one cigarette in the past 30 days)’.
Based on the responses, adolescents were classified as (i) never-smokers (negative answers to the two questions), (ii) former smokers (answered yes to the first but not to the second question), (iii) and smokers (two positive answers). Former smokers were not eligible for the present analysis, therefore a status of a non-smoker in this study refers to adolescents who had never tried smoking.
Motivation to changing the smoking behaviour
We relied on the Prime Theory of Motivation [7] to conceptualize and measure the motivation of adolescents to change their smoking behaviours. This comprehensive theory of motivation poses that the main elements of motivation to change a behaviour include, in addition to beliefs about what to do and what is good or bad, the feelings of desire and the self-conscious intentions to act in a specific way [7].
Concepts are at the psychological level, and the theory was designed to perform as a pegboard into which other theories can be plugged [7]. The pictorial representation of the Prime Theory was adapted to illustrate our assumption of an association between the adolescents’ SOC and their motivational system (Fig. 1).
Outcome 1: Smokers’ motivation to stop smoking
We assessed MStop using the Motivation to Stop Scale (MTSS), a one-item instrument based on Prime Theory that was validated in English [16] and Dutch [17]. For the present study, we have translated and adapted the instrument to the Brazilian Portuguese language [15]. The MTSS scale consists of the following question: ‘Which of the following describes you?’ Response options are: (i) ‘I don’t want to stop smoking’, (ii) ‘I think I should stop smoking but don’t really want to’, (iii) ‘I want to stop smoking but haven’t thought about when’, (iv) ‘I REALLY want to stop smoking but I don’t know when I will’, (v) ‘I want to stop smoking and hope to soon’, (vi) ‘I REALLY want to stop smoking and intend to in the next 3 months’ and (vii) ‘I REALLY want to stop smoking and intend to in the next month’. The ordering of the categories indicates an ascending level of MStop: (i) absence of any belief, desire or intention, (ii) belief only, (iii) moderate desire but no intention, (iv) strong desire but no intention, (v) moderate desire and intention, (vi) strong desire and medium-term intention and (vii) strong desire and short-term intention [16, 17].
Outcome 2: Non-smokers’ motivation to start smoking
The measurement of MStart was based on a three-item questionnaire proposed by one of the senior authors of Prime Theory [7] to assess desire, beliefs, and intentions to start smoking: 1) Does the idea of smoking feel good to you? (desire); 2) Do you think the benefits of smoking for you would make up for the risks? (belief); and 3) Do you plan to try smoking in the next year? (intention). Each question has five response categories, where 1 indicates strong agreement (Definitely yes) and 5 strong disagreement (Definitely not).
To construct the outcome variable MStart, we first evaluated the internal consistency of the questions, and it was satisfactory (Cronbach’s α = 0.72; 95% CI = 0.70–0.74). Then, we dichotomized the response categories into (0) Definitely not (a strong and firm absence of desire, belief, or intention to smoke) and (1) Probably not + Not sure + Probably + Definitely yes. Subsequently, we computed the combination of the responses to the three questions and created an eight-category level MSS ordinal variable which was ordered from (1) absence of any belief, desire and intention (no MStart) to (8) simultaneous presence of any level of belief, desire, and intention (highest level of MStart) (Table 1). The categories were ordered according to previous studies that assessed motivation to change smoking behaviours and were based on Prime Theory (7,8,16,17). In addition, we analysed each of the three dimensions of motivation investigated (desire, belief, and intention) as a single ordinal outcome.
Table 1
Levels of motivation to start smoking identified through the combination of responses to the Prime Theory-based questionnaire.
Levels of Motivation | Combination of responses |
| Desire | Belief | Intention |
1 | No motivation | 0 | 0 | 0 |
2 | Belief, only | 0 | 1 | 0 |
3 | Desire, only | 1 | 0 | 0 |
4 | Desire and belief, no intention | 1 | 1 | 0 |
5 | Intention, only | 0 | 0 | 1 |
6 | Belief and intention | 0 | 1 | 1 |
7 | Desire and intention | 1 | 0 | 1 |
8 | Desire, belief, and intention | 1 | 1 | 1 |
0 = Absence; 1 = Presence |
Sense of Coherence
The adolescent’s SOC was measured with the short version of the SOC Orientation to Life Questionnaire [12], which was adapted to the Brazilian context [18]. The scale consists of 13 items on a 7-point Likert-type scale with descriptive endpoints. The total SOC score is obtained by the sum of the 13 items single scores, ranging from 13 to 91, with the higher scores indicating stronger SOC [12]. Before computing the SOC score, some of the items (1, 2, 3, 7 and 10) are recoded in reverse to provide an ascendant measurement to the scale. The scale’s Cronbach’s Alpha in the present study was 0.79 (95% CI = 0.77–0.80) indicating a good internal consistency.
Background variables
The potential confounders were: 1) Self-reported race/skin colour (official categories in Brazil [19]: white, black, brown, yellow (Asian descendants) or indigenous; 2) Sex: (i) male and (ii) female; 3) Age: 13 to 19 years; 4) Socioeconomic status, based on the level of education of the adolescents’ mother, with categories ascending from no study (illiterate) to university degree (Table 1); 5) Exposure to smoking parents: (i) yes (at least one of their parents was a smoker) and (ii) no; 6) Exposure to smoking friends: (i) yes (at least one of their friends was a smoker) and (ii) no.
Statistical analysis
We describe our data using absolute and relative frequencies (n; %), mean values and standard deviations (SD). Ordinal Logistic Regression was used to estimate the odds ratio (OR) and 95% confidence intervals (CI) for the associations between SOC and adolescents’ motivation to start and stop smoking. We used a Direct Acyclic Graph (DAG) [20, 21] (Fig. 2) to guide the selection of variables to be adjusted for in the multivariable models, which were age, sex, mothers’ educational level, parents’ smoking status, and friends’ smoking status. The IBM SPSS (v. 24) software was used in the statistical analyses and the DAG was structured using the DAGitty tool [22].