Methods and materials

The study covers Rajshahi City, which is situated in northern Bangladesh. There are 30 wards in the city area; wards are less-developed parts of the entire city infrastructure, where there is a low literacy rate and there is more accessibility to smoking items such as like bidis, marijuana (local name ganja), cigarettes, and various drugs, in particular phensedyl. The total population of the city area is 449,756, of whom 232,974 are male and 216,782 female, with a sex ratio (M/F) of 107.469 and a population density per square kilometer of 4318 (Bangladesh Bureau of Statistics 2013).

To choose survey questionnaires, relevant studies were reviewed. After the literature reviewed in the Introduction section above was taken into account, we prepares open-ended questionnaires. Pre-testing interviews were conducted to check the adequacy of the responses among the respondents compared to the data of relevant studies. Finally, we prepared a closed-ended questionnaire based on pre-testing interviews. The exact number of the smoker population is unknown because there is no existing study conducted on smokers in the area. The sample size was calculated by the following formula (Survey system 2020) for unknown population.

$$\mathrm{The}\ \mathrm{required}\ \mathrm{sample}\ \mathrm{size},n=\frac{Z^2\times P\times \left(1-P\right)}{C^2}=160$$
(1)

Where, Z = Z value (1.96 for 95% confidence level), P =  percentage picking a choice (0.5 used for maximum sample taken), and C =  margin of error (0.0775 = ± 7.75). Among 30 wards, three wards, namely ward-7, ward-10, and ward-20, were selected randomly using simple random sampling procedure. The study was based on cross-sectional design study, and the required data were collected through face-to-face interview with a pretested questionnaire. The total 160 respondents (male 130 and female 30) who are addicted to smoking were selected randomly from the three selected wards. Statistical analyses were performed using IBM SPSS version 24. Frequency distribution was used to compute the characteristics of smokers and their opinion of smoking on social environment and public health (Marmot and Wilkinson 2006). The main null hypothesis — there is no significant association between smoking attributes and general health or social attributes (bivariate association) — was checked using χ2 test between the following pairs of variables: frequency of heartbeat rate and class of smokers, suffering from diseases and started smoking at age level in years, category of smoking articles and suffering from diseases, gender and class of smokers, class of smokers and wards of the smokers, causes for smoking and class of smokers, from which person smoking was learned and started smoking at age level in year, and profession of the smokers and started smoking at age level in year. The null hypothesis was tested by chi-square test to test the association between two variables, and the null hypothesis was rejected if the p value of the test statistic was less than 0.01 (p < 0.01) at 1% level of significance and 0.05 (p < 0.05) at 5% level of significance (Diener 2008; Mosteller 1968). Binary logistic regression (Islam et al. 2020; Coleman et al. 2003) was estimated considering 'suffering from diseases during smoking' as the dependent variable, and the independent variables to be educational qualification, ward, age group, gender, profession, frequency of smoking related to meal time, BMI class, and from which person did you learn smoking? Multinomial logistic regression was also applied to determine which factors best explain and predict health characteristics of smokers, general health, and their perception of smoking outcome. The significant odds ratios (OR) are found if the p value of the test statistic was less than 0.01 (p < 0.01) at 1% level of significance and 0.05 (p < 0.05) at 5% level of significance (Chakraborty et al. 2003; Wahed and Hassan 2017).

Results and discussion

The demographic characteristics of smokers are shown in Table 1. There were 130 male respondents (81.3%) and 30 female respondents (18.8%). The profession of the respondents was student 18 (11.3%), labor 73 (45.6%), and service 69 (43.1%). The age level of the respondents in years was seven below 15 (4.4%), 25 in between 15 to below 25 (15.6%), 73 in between 25 to below 35 (45.6%), 49 in between 35 to below 45 (30.6%), and six above 45 (3.80%). The mean and SD age for the respondents were 41.797 and 0.511 years respectively.

Table 2 shows that the great majority (93.8%) of the smokers believe that smoking has an impact on the social environment, while 6.3% believe that it does not: 48.8% of smokers feel that smoking cause air pollution, and 51.3% also feel that smoking has an impact on the breathing of non-smokers (passive smoking).

Table 3 shows that the majority of respondents (113, 70.6%) smoke bidis, 38 (23.8%) smoke low-price cigarettes, seven (4.4%) smoke high-price cigarettes, and two (1.3%) smoke other things (including marijuana). The smoker respondents suffer from various diseases such as gastric problems (110, 68.8%), fever (11, 6.9%), and headache (39, 24.4%). The smoker respondents learnt to smoke at below 10 years old (51.3%). Mainly, they learnt 'second-hand smoking' by collecting bidis or cigarettes thrown away by smokers, and they sometimes learnt smoking from their older friends aged 10 to 15 (25%) or 15 plus (23.8%). The mean and SD age of learning smoking among the respondents were 11.894 and 3.442 years respectively. The frequency of smoking in relation to meal time was as follows: 15 of respondent smokers (9.4%) smoked before meals, 129 (80.6%) smoked after meals, 12 (7.5%) smoked at any time and four (2.5%) smoked at any other convenient time. At the time of smoking, 160 (100%) smokers felt an effect on heartbeats, such as low heartbeat rate (40, 25%), high heartbeat rate (113, 70.6%), and other effects (seven, 4.4%). Out of 160 smokers, 82 (51.3%) were chain smokers (chain smoking is the practice of smoking several bidis or cigarettes in succession, sometimes using the ember of a finished cigarette to light the next). The term chain smoker often also refers to a person who smokes relatively constantly, though not necessarily chaining each bidi or cigarette. Of the rest, 38 (23.8%) smoked sometimes and 40 (25%) smoked more than three times a day. The smokers smoke for various reasons such as being addicted (78, 48.8%), depressed (41, 25.6%), influence of friends (38, 23.8%) and other reasons (three, 1.9%). The smokers were asked "from which person did you learn smoking?": 86.3% of smokers learnt smoking from friends, 2.5% from fathers, 8.1% from brothers, and 3.1% from others, as shown in Fig. 1.

The testing of the null hypothesis — that there is no association between smoking and a series of variables — was tested using the following pairs of variables: frequency of heartbeat rate and class of smokers, suffering from diseases and started smoking at age level in years, category of smoking articles and suffering from diseases, gender and class of smokers, class of smokers and wards of the smokers, causes of smoking and class of smokers, from which person smoking was learned and started smoking at age level in year, and profession of the smokers and started smoking at age level in year. The cross tables of smoking attributes with χ2 test statistics and p value are presented in Table 4.

As seen from Table 4, the results are described in the following manner. Class of smokers are significantly associated with effects on heartbeats (χ2 = 160, p < 0.01). The smokers who smoked more than three times (100%) were more likely to have a low heartbeat than chain smokers and occasional smokers; on the other hand, chain smokers (68.3%) and occasional smokers (31.7%) were more likely to have high/other heartbeats. Started smoking at age level in year was significantly associated with suffering from diseases (χ2 = 104.603, p < 0.01). The smokers who were below 10 years old (70.0%) or above 15 years (30.0%) were likely to have gastric problems; on the other hand, 80.0% of smokers who were at age 10-15 were likely to have fever/headache/other diseases. Category of smoking articles was significantly associated with suffering from diseases (χ2 = 104.603, p < 0.01). The smokers who smoked bidis (95.5%) were more likely to suffer from gastric problems; on the other hand, smokers who smoked low-price cigarettes/high price cigarettes/others (84.0%) were more likely suffering from fever/headache/other diseases. Class of smokers was not significantly associated with gender of smokers (χ2 = 1.396, p > 0.01). Class of smokers was not significantly associated with ward of smokers (χ2 = 0.107, p > 0.01). Class of smokers was significantly associated with causes for smoking (χ2 = 289.356, p < 0.01). The smokers who were chain smokers (100%) were more likely to have addiction to smoking; the smokers who smoked more than three times in a day (95.1%) were more likely to have depression; and the smokers who smoked sometimes (92.7%) were more likely to have been influenced by friends/others. Started smoking at age level in year was not significantly associated with from which person did you learn smoking (χ2 = 2.092, p > 0.01). Started smoking at age level in year was significantly associated with profession of smokers (χ2 = 40.553, p < 0.01). The smokers who started smoking at age 10–15 years (44.0%) were more likely to be from the profession student/labour; the smokers who started smoking at age below 10 (37.4%) were more likely to be from the profession student/labour; and the smokers who started smoking at age level below 10 (69.6%) were more likely to be service/other profession holders.

In the above section, bivariate analysis of smokers' characteristics, general health, and perception of smoking in the social environment, categorized as predisposing and enabling factors, was performed to examine the nature of the association between these factors and smoking exposure status. Numerous associations were found to be significant in the bivariate analysis. However, bivariate association between two variables using χ2 test does not necessarily imply a significant causal relationship between them. Therefore, binary and multinomial logistic regression was applied to determine which factors best explain and predict smokers' health characteristics, general health, and perception of smoking outcome. For binary logistic regression, suffering from diseases during smoking was considered as the dependent variable, with outcomes gastric problems (68%) and fever/headache/other (31.3%), and independent variables such as education qualification, ward, age group, gender, profession, frequency of smoking related to meal time, BMI class and 'from which person did you learn smoking?' were considered as categorical covariates. Table 5 shows the frequency distribution of required variables for binary logistic regression estimation, and Table 6 shows the estimated results of binary logistic regression of suffering from diseases during smoking on categorical covariates including p value, odds ratio (OR) and 95% CI for OR.

As seen in Table 6, the results are interpreted in the following ways. The OR of smoking learnt from friends is (OR = 2.630, 95% CI 0.739–9.355, p > 0.01). The smokers who learnt smoking from friends were more likely to suffer from diseases compared to smokers who learnt smoking from father/brothers/others, although this is not statistically significant. It is evident that smokers who learnt smoking from friends were 2.630 times more likely to suffer from diseases compared to smokers who learnt smoking from father/brothers/others. The ORs of age at below 15/15 to below 25, and 25 to below 35 are OR = 2.630, 95% CI 0.739–9.355, p > 0.01 and OR = 1.739, 95% CI 0.460–6.576, p > 0.01 respectively. Younger smokers were more likely to suffer from diseases compared to older smokers, although ORs were not statistically significant. It is evident that smokers below 25 years of age were 2.275 times more likely to suffer from disease compared to the smokers of age 35 and above, and smokers below 35 years of age were 1.739 times more likely to suffer from disease compared to smokers of age 35 and above. The OR for the male gender was OR = 0.804, 95% CI 0.286–2.266, p > 0.01, which is not statistically significant. The OR of profession of students/labours was OR = 6.363, 95% CI 1.918–21.104, p < 0.01. Student/labour smokers were more likely to suffer from diseases compared to service/other smokers; the difference was statistically significant. It was shown that student/labour smokers are 6.363 times more likely to suffer from diseases compared to service/other smokers. The OR of smoking before meal time was OR = 1.280, 95% CI 0.329–4.973, p > 0.01. The smokers who smoked before meal time were more likely to suffer from diseases compared to smokers who smoked after meal/any time/others, although this was not statistically significant. It was shown that smokers who smoked before meal time were 6.363 times more likely to suffer from diseases compared to smokers who smoked after meal/any time/other. The OR of overweight/obese smokers was OR = 1.130, 95% CI 0.517–2.471, p > 0.01. Overweight/obese smokers were more likely to suffer from diseases compared to underweight/normal weight smokers, although this was not statistically significant. It was shown that overweight/obese smokers were 1.130 times more likely to suffer from diseases compared to underweight/normal weight smokers. The ORs of ward-7 and ward-10 were OR = 0.335, 95% CI 0.088–1.279, p > 0.01 and OR = 0.936, 95% CI 0.223–3.937, p > 0.01 respectively; this was not statistically significant. The ORs of Illiterate and primary qualified smokers were OR = 0.821, 95% CI 0.212–3.180, p > 0.01 and OR = 1.665, 95% CI 0.477–5.808, p > 0.01 respectively. Smokers with a primary educational qualification were more likely to suffer from diseases compared to SSC/HSC/graduate/Masters qualified smokers, although this was not statistically significant. It was shown that primary qualified smokers were 1.665 times more likely to suffer from diseases compared to SSC/HSC/graduate/Masters qualified smokers.

Table 7 represents a case processing summary for multinomial logistic regression of cause for smoking with a view to corresponding categorical factors such as ward, age, gender, 'from which person did you learn smoking?', BMI class and profession. Table 8 represents estimation results of multinomial logistic regression (trichotomous) of cause for smoking (ref: addiction) with a view to corresponding categorical factors including p value, Odds Ratio (OR) and 95% CI for OR.

In Table 8, the ORs are described in the following manner. The OR of underweight/normal weight smokers was OR = 1.060, 95% CI 0.462–2.434, p > 0.01. Underweight/normal weight smokers were more likely to suffer from depression compared to overweight/obese smokers, although this was not statistically significant. It was shown that underweight/normal weight smokers were 1.060 times more likely to suffer from depression compared to overweight/obese smokers. The OR of learnt smoking from father/brothers/others was OR = 2.649, 95% CI 0.698-10.060, p > 0.01. Smokers who learnt smoking from father/brothers/others were more likely to suffer from depression compared to smokers who learnt smoking from friends, although this was not statistically significant. It was shown that smokers who learnt smoking from father/brothers/others were 2.649 times more likely to suffer from depression compared to smokers who learnt smoking from friends. The OR of smokers at age below 15/15 to below 25 was OR = 1.008, 95% CI 0.260–3.907, p > 0.01. Smokers who were at age below 15/15 to below 25 were more likely to suffer from depression compared to smokers of ages 35 and above, although this was not statistically significant. It was shown that smokers who were at age below 15/15 to below 25 were 1.008 times more likely to suffer from depression compared to smokers of ages 35 and above. The OR of smokers at age 25 to below 35 was OR = 0.848, 95% CI 0.209–3.436, p > 0.01. Smokers who were at age 25 to below 35 were less likely to suffer from depression compared to smokers of ages 35 and above, although this was not statistically significant. It was shown that smokers who were at age 25 to below 35 were 0.848 times less likely to suffer from depression compared to smokers of ages 35 and above. The OR of smokers from ward-7 was OR = 1.259, 95% CI 0.339–4.670, p > 0.01. Smokers from ward-7 were more likely to suffer from depression compared to smokers from ward-20, although this was not statistically significant. It was shown that smokers from ward-7 were 1.259 times more likely to suffer from depression compared to smokers from ward-20. The OR of smokers from ward-10 was OR = 1.095, 95% CI 0.253–4.748, p > 0.01. Smokers from ward-10 were more likely to suffer from depression compared to smokers from ward-20, although this was not statistically significant. It was shown that smokers from ward-10 were 1.095 times more likely to suffer from depression compared to smokers from ward-20.

The OR of underweight/normal weight smokers was OR = 0.556, 95% CI 0.251–1.235, p > 0.01. Underweight/normal weight smokers were less likely to smoke upon the impact of effects of friends/others compared to overweight/obese smokers, although it is not statistically significant. It was shown that underweight/normal weight smokers were 0.556 times less likely to smoke upon the impact of effects of friends/others compared to overweight/obese smokers. The OR of learnt smoking from father/brothers/others was OR = 1.119, 95% CI 0.379–3.309, p > 0.01. The smokers who learnt smoking from father/brothers/others were more likely to smoke as a result of the influence of friends/others compared to smokers who learnt smoking from friends, although this was not statistically significant. It was shown that smokers who learnt smoking from father/brothers/others were 1.119 times more likely to smoke as a result of the influence of friends/others compared to smokers who learnt smoking from friends. The OR of smokers at age below 15/15 to below 25 was OR = 0.423, 95% CI 0.102–1.758, p > 0.01. The smokers who were at age below 15/15 to below 25 were less likely to smoke as a result of the influence of friends/others compared to smokers of ages 35 and above, although this was not statistically significant. It was shown that smokers who were at age below 15/15 to below 25 were 0.423 times less likely to smoke as a result of the influence of friends/others compared to smokers of ages 35 and above. The OR of smokers at age 25 to below 35 was OR = 0.473, 95% CI 0.122–1.841, p > 0.01. Smokers who were at age 25 to below 35 were less likely to smoke upon the impact of effects of friends/others compared to smokers of ages 35 and above, although it is not statistically significant. It was shown that smokers who were at age 25 to below 35 were 0.473 times less likely to smoke as a result of the influence of friends/others compared to smokers of ages 35 and above. The OR of smokers from ward-7 was OR = 2.286, 95% CI 0.629–8.312, p > 0.01. Smokers from ward-7 were more likely to smoke as a result of the influence of friends/others compared to smokers from ward-20, although this was not statistically significant. It was shown that smokers from Ward-7 are 2.286 times more likely to smoke as a result of the influence of friends/others compared to smokers from ward-20. The OR of smokers from ward-10 was OR = 2.013, 95% CI 0.441–9.180, p > 0.01). Smokers from ward-10 were more likely to smoke as a result of the influence of friends/others compared to smokers from ward-20, although this was not statistically significant. It was shown that smokers from ward-10 were 2.013 times more likely to smoke as a result of the influence of friends/others compared to smokers from ward-20.

Conclusions

Smoking in public places should be restricted because non-smokers cannot breathe freely, and it is not healthy for them to inhale smoke indirectly (passive smoking), which has many adverse effects on public health. In Bangladesh, smoking-related materials are easily available to children and adolescents, which might encourage experimentation and the subsequent development of a regular smoking habit and ultimately addiction. The majority of respondents learnt smoking from their friends when they were less than 10 years old (51.3%). Parents should take the initiative so that their children cannot mix with smoker friends. A significant association exists between smoking creating adverse effects on the social environment and human health and various characteristics of the smoker respondents in the city area. Therefore, further laboratory-based research should be conducte,d and research should also include non-smoker respondents as a control group for better comparison. Emphasis has to be placed on the legislative authorities making new laws to ban illegal sales of smoking materials to children and also active enforcement of the existing law with regard to smoking in public places. Moreover, in order to curb tobacco usage the authority should take measures such as increasing tobacco taxation, encouraging e-cigarette use, placing a total ban on tobacco advertising and promotion, and also establishing health education campaigns. Finally, an environmental approach needs to be developed to reduce risk factors causing health hazards and to promote comprehensive multi-dimensional protective factors.