Study design

This study was descriptive cross-sectional following a quantitative approach. For reporting observational research, this cross-sectional study adhered to the guidelines of the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement (Supplementary file Material 1).

Study participants, sample size, and sampling

This study included a total of 288 tobacco using children living in a slum area located in Mirpur Thana of Dhaka district, Bangladesh. The respondents indicated that the children aged 10–17 years, residing in the selected slum of Dhaka city, provided consent to participate in this study.

Considering the highest number of people in Bangladesh living in slum settlements of Dhaka city, slum children of Dhaka were taken into account as the study population⁹. In addition, the slum of Mirpur Thana was chosen as the study site due to its overpopulation compared to other areas within Dhaka city¹⁰.

A potential standard sample size was assumed to be 312, calculated using the formula:

where Z (standard normal deviate) was taken as 1.96, p is the proportion of smoker adolescent boys in Dhaka city, q=1-p, and d is the desired margin of error. The proportion of smoker adolescent boys in Dhaka city was 35.33%¹¹; among them, the proportion living in slums was 80%¹¹. Therefore, the proportion of smoker slum adolescent boys in Dhaka city was estimated at 28.26% or 0.28, with a margin of error of 0.05. However, the final sample size was adjusted downward to account for a 5% non-response rate, as these individuals tend to be transient and hard to reach. After data cleaning and initial processing, the final sample size was set at 288. Since youth who use tobacco in urban slum areas are a hidden or hard-to-reach population, a non-discriminative snowball sampling technique was used to select participants. Due to their high mobility, lack of registration, and stigma, these individuals are often difficult to access using traditional probability sampling techniques. Researchers were able to find participants through peer referrals by using snowball sampling, which is a useful method for reaching underrepresented or socially connected groups. The study’s goal of examining behaviors within a particular, underrepresented group justifies this approach, even though it restricts the findings’ applicability to the general population.

Data collection

A pre-tested, semi-structured questionnaire was used to collect quantitative data from the slum children through interviewer-administered interviews. Respondents were interviewed between July and December 2022 at their convenience. The survey took only 10 to 15 minutes to complete. All authors had access to the data collection process and to safeguarding participants’ information during and after data collection. The survey was conducted in Bengali with the full support and cooperation of the slum residents.

Ethical considerations

This study was approved by the Ethical Review Committee of the Department of Public Health at Northern University Bangladesh (NUB/DPH/EC/2022/21) and adhered to the Declaration of Helsinki. Participants’ anonymity and voluntary participation were maintained. Verbal informed consent was obtained from the respondents at the start of the survey; they could withdraw at any time, and their autonomy was protected.

Questionnaire design

The questionnaire (Supplementary file Material 2) was pre-validated by two independent reviewers and pre-tested with 10 respondents. Expert review was used to assess content validity, ensuring that the items were relevant to the study’s goals and appropriate. The questionnaire was updated based on feedback from pre-test participants, focusing on clarity and relevance, although formal statistical reliability measures were not calculated due to the small pre-test sample. The key components of the questionnaire included: 1) Sociodemographic information such as age of tobacco user, gender, education level, respondent’s occupation, parental occupation; 2) Types of tobacco use such as cigarette smoking, other tobacco products such as bidi, gul, jarda, etc.; 3) Age at first tobacco use including before teenage years, during teenage years, or after.

Data analysis

Study characteristics were summarized with descriptive statistics (frequency and percentage). Age (≤15 years and >15 years) of the respondents and the first initiation age of tobacco use (<13 years and ≥13 years) were categorized based on their respective mean values as cutoff points. Odds ratios with 95% confidence intervals were calculated to examine the associations between sociodemographic characteristics (independent variables) and the first initiation of tobacco use (dependent variable). A multiple logistic regression analysis was performed, followed by a modeling process using backward elimination, including pre-specified covariates such as age of tobacco use, gender, education level, respondent’s occupation, and parental occupation. To ensure statistical significance, gender (p>0.05) was excluded. The collected data were checked and analyzed using the Statistical Package for the Social Sciences (SPSS) software.

Distribution of respondents’ general characteristics by type of tobacco used and first initiation of tobacco use

The analysis uncovered several key findings about tobacco users. Most notably, 66.7% of respondents (192 out of 288) started using tobacco at the age of ≤15 years. Regarding gender, the vast majority were male (95.8%), showing a significant gender gap in tobacco use. Education level indicated that 65.6% had only non-formal or primary education. Occupational data showed that 60.1% of respondents were unemployed, while 56.9% of fathers were day laborers. Additionally, nearly half of the mothers (49%) were housewives.

The analysis also revealed several statistically significant associations between participant characteristics and both the type and initiation of tobacco use. Among those who started using tobacco before the age of 15 years, 64.1% used cigarettes, while among those who initiated after 15 years, 81.2% used cigarettes and only 18.8% used other forms (p=0.00*). Males significantly reported the use of cigarettes (72.1%), while females predominantly used other tobacco forms (83.3%) (p=0.01*). Additionally, 83.3% of females initiated tobacco use before the age of 13 years compared to 55.8% of males, though this difference did not reach significance (p=0.06). Education level showed a strong influence: 88.0% of participants with secondary or higher education used cigarettes, whereas 53.1% of those with non-formal education used other tobacco (p=0.00*). First initiation was also more common among the non-formally educated (69.1%) compared to those with higher education (37.0%) (p=0.01*). Regarding occupation, unemployed individuals were more likely to use cigarettes (63.0%) and to initiate tobacco use before the age of 13 years (69.9%). These associations were significant (p=0.01*). While father’s occupation was not significantly related to the type of tobacco used (p=0.984), it was linked to the first initiation of tobacco (p=0.04*). Mother’s occupation, particularly being a day wager, was significantly associated with first initiation of tobacco (p=0.01*) (Table 1).

Predictors associated with the first initiation of tobacco use

The study’s regression analysis identified key factors linked to respondents’ first tobacco use. Those with no formal education had nearly four times higher odds (AOR=3.84; 95% CI: 1.94–7.62) of initiating tobacco before the age of 13 years compared to those with secondary or higher education. The study also showed that respondents who were unemployed had a greater likelihood of starting tobacco use before the age of 13 years than those in other occupations. Additionally, respondents whose fathers were day workers had more than three times the odds (AOR=3.47; 95% CI: 1.17–10.24) of initiating tobacco before the age of 13 years. Similarly, those with mothers who were business women had over three times the odds (AOR=3.86; 95% CI: 1.12–13.36) of first using tobacco before the age of 13 years compared to those whose mothers were housewives (Table 2).

Limitations

The current study has some limitations. Being cross-sectional, this study does not permit a causal association of the explanatory variables with the outcome variable. Moreover, as all respondents were living in slum areas, and this limits the generalizability of the current findings among the youth of residential or other settlements in Bangladesh. Additionally, the study’s external validity may be impacted by the single-area sampling and comparatively small sample size. Potential sources of bias include measurement inaccuracy from self-reported tobacco use, which may be impacted by recall or social desirability bias, and sampling bias brought on by purposive slum area selection. Despite the limitations, as a pilot study, it demonstrated a detailed analysis of the various determinants of smoking among children living in slums, especially parents’ occupations, which are unique determinants identified by our study.