Modeling the determinants of cigarette consumption in Iranian households with children under 5 years of age using the Income and Expenditure Survey 2021

Abstract

Background

Cigarette smoking remains a significant public health concern, with detrimental effects on both smokers and those exposed to secondhand smoke. This study investigates the factors influencing smoking behaviors in Iranian households, focusing on households with children under five years old.

Methods

We conducted a cross-sectional analysis of 8751 Iranian households using data from the Iranian Household Income and Expenditure Survey (HIES) collected by the Statistical Center of Iran (SCI) in 2021. A variety of count regression models were examined, including Poisson (P), Negative Binomial (NB), Generalized Poisson (GP) and their zero-inflated counterparts. The best fitting model was selected based on goodness-of-fit indices.

Results

Approximately 87% of households were non-smokers, with a mean daily cigarette consumption of 14.29 (SD = 8.64) among smokers. The three-level Zero-Inflated Generalized Poisson (TL-ZIGP) model was considered a suitable model for the interpretation of the data. Key predictors of smoking included older age, male gender, lower education, higher income, and larger homes. Homeownership was negatively associated with smoking. Considerable geographic variation in smoking prevalence was observed.

Conclusions

Socioeconomic and demographic factors, including lifestyle and geographic regions, are associated with smoking behavior among Iranian households. A three-level ZIGP regression model is appropriate for analyzing these complex relationships. This study reveals different patterns of smoking prevalence in different population subgroups and emphasizes the need for targeted interventions to effectively reduce smoking rates.

Text box 1. Contributions to the literature
• Accurate data modeling on cigarette consumption in Iranian households, especially with frequent zero counts and skewness, requires special count regression methods.
• The effectiveness of the Three-Level Zero-Inflated Generalized Poisson (TL-ZIGP) model in dealing with the complexity of data with a multilevel structure from a national survey.
• The results can serve as a basis for targeted anti-smoking campaigns, tax measures, tobacco advertising regulations, smoke-free environment initiatives and measures to reduce smoking in households with young children to protect vulnerable populations from second-hand smoke.

Background

Smoking and its secondhand smoke (SHS) are recognized as health hazards globally [1]. Cigarette smoke is a representative source of exposure to toxic chemicals for humans [2]. In addition to nicotine, cigarettes contain more than 4,000 other substances that are responsible for a 50% premature mortality rate among smokers, and exposure to these factors reduces life expectancy by almost 10 years on average [3]. Smoking is also, a well-established risk factor for cardiovascular disease [4], and chronic obstructive pulmonary disease, which can lead to chronic respiratory problems and increased mortality [5, 6]. Even healthy non-smoking adults are at risk for health issues including coronary heart disease, stroke, and lung cancer when exposed to SHS [7]. Numerous studies have investigated the consequences of smoking in the family, particularly in children under the age of five [8, 9]. These studies have shown a link between children's exposure to tobacco smoke products (TSPs) and the occurrence of certain diseases. Cigarette smoke has been found to significantly increase the proportion of acute respiratory infections in children under the age of five [10]. Children exposed to TSPs or SHS also are at an increased risk for sudden infant death syndrome, middle ear disease, more frequent and severe asthma, respiratory symptoms, and slowed lung growth [11, 12]. Despite worldwide initiatives to reduce tobacco smoking, it is estimated that up to 40% of children are still exposed to tobacco smoke [13]. Mortality due to environmental tobacco smoke has also been estimated at 28% for children [14]. Although prevalence rates of smoking began to decline in most countries during recent decades, it is still high in many parts of the globe [15]. The World Health Organization (WHO) estimates that there will be one and a half billion smokers worldwide by 2050 [16]. It seems the prevention of smoking as a modifiable risk factor has not yet received sufficient attention. [17]. According to the studies carried out, Iran is one of the countries where the smoking rate is still high [18]. The smoking patterns and trends in Iranian households have been assessed in recent years through several national surveys [17, 19]. According to a 2010 study, the overall prevalence of daily cigarette smoking in Iran was 11.9% and this figure was 21.4% for men and 1.5% for women in 2007 [20]. In 2012, smoking was the direct cause of death of 4623 people in Iran [21]. Another study based on the 2016 STEPS survey data reported considerable heterogeneity in smoking prevalence at the district level in Iran. The results showed that the average daily cigarette consumption ranged from 0 to 4.5 for women and from 4.6 to 40.9 for men [18]. A study based on data from the 2021 STEPS survey also found that the prevalence of tobacco use in Iran was 14.01% overall, 4.44% for women and 25.88% for men. In addition, the prevalence of cigarette use was 9.33% overall, 0.77% for women and 19.95% for men [22]. Many socio-cultural and demographic risk factors may affect pattern of cigarette smoking in households. Based on the 2021 study, Iranian women with less education or lower income and women living in rural areas were more likely to smoke or be exposed to SHS. In contrast, men with higher wealth or education tended to smoke more [18]. In the 2023 study, the results also indicated that a lack of education, a male head of household and a large number of adult members in the household were associated with increased smoking in the household. Conversely, smoking in the household decreased with increasing age of the household head [23]. However, as there are some challenges in developing an effective tobacco policy in Iran, measures to reduce tobacco consumption have not been sufficiently considered or implemented [24, 25]. Therefore, it is of interest to study the smoking behavior of Iranian households to achieve a reasonable reduction or cessation of smoking. However, before taking measures to achieve this goal, it is important to identify the factors that influence the trend of smoking in households. Various studies emphasize the importance of reducing exposure to cigarette smoke, especially in young children, to reduce the risk of health problems [26]. Since children's exposure to SHS is one of the major problems in households with children under 5 years of age, it is necessary to pay attention to smoking habits in these households. Therefore, the present study examines smoking behavior by modeling daily cigarette consumption and the factors that influence it in these Iranian households. We treat daily cigarette consumption as a discrete response variable and use count models including Poisson (P), Negative Binomial (NB) and Generalized Poisson (GP). Given the high number of non-smoking households, we also use Zero-Inflated (ZI) models of these distributions to better account for excess zeros [23]. Zero-inflated models treat excess zeros as a separate process from the count, which allows for a more precise estimation of the effects [27]. In addition, the analysis of smoking behavior requires consideration of cultural and ethnic differences, which can be effectively captured by nested models for more accurate results. Therefore, due to the heterogeneity of cigarette consumption in Iranian households, multilevel models are required for in-depth analysis [28, 29]. Since no comprehensive study on smoking consumption and factors influencing smoking behavior in Iranian households using a multilevel count regression model has been conducted so far, the present study aims to address this issue. It is expected by conducting studies like this research and identifying the pattern of smoking in the country as well as the factors affecting it among the households with children under 5 ages, it is possible to plan to reduce or quit smoking by providing practical solutions.

Methods

Source of data

The data was extracted from the Household Income-Expenditure Survey (HIES) in Iran, conducted by the Statistical Center of Iran (SCI) in 2021. The survey was cross-sectional and included a sample of 37,988 urban and rural Iranian households from 449 cities and 31 different provinces. In the SCI, a three-stage cluster sampling method was used to select households: First, census tracts were classified and selected, second, urban and rural blocks were selected, and third, sample households were identified. Data was also collected through face-to-face interviews using a comprehensive and standardized questionnaire [30].

Due to our objectives, only households with at least one child under the age of 5 were included in this study. The sample used consisted of 8571 households from 425 cities in all provinces with complete information on the selected variables. The number of households for the provinces varies between 109 and 537 in the data included. In the HIES survey, the monthly number of cigarettes consumed in the households was recorded, but in the present study the number of cigarettes consumed daily in the households is considered as the outcome variable, with the month being converted into a day and rounded. Two categories of variables were used to identify the factors influencing cigarette consumption in the country's households: Head-of-household characteristics and household-level characteristics. Information about the household head, including age, gender, education, marital status with two categories, including married and others (Single/divorced/widow), occupation and a range of socio-demographic variables related to the household, such as area of residence, family size, number of educated members, number of student members, members who work, monthly health expenditure, house ownership status, house area, monthly income with four categories, including; less than 150 US$ (low), 150–400 US$ (low to middle), 400–600 US$ (middle) and more than 600 US$ (high), machine/motorcycle, bicycle, internet access, and computer/tablet. All variables were selected based on literature research and previous studies.

Data analysis

The three counting models from generalized linear models used in this study are the Poisson model, the negative binomial model and the generalized Poisson model, as follows:

Poisson distribution (P)

The famous counting distribution is the Poisson distribution as follows:

$$f\left(y;\mu \right)=\frac{{e}^{-\mu }{\mu }^{y}}{y!} ; y=\text{0,1},2,\dots$$

(1)

where $\mu>0$ is a real positive number representing the mean and variance of the distribution [31].

Negative Binomial distribution (NB)

The NB model is obtained by adding another source of variability to the P model, e.g. the dispersion parameter. The added parameter allows the variance to exceed the mean. Therefore, the NB distribution allows the calculation of overdispersion [32]. The probability distribution function (p.d.f) is as follows:

$$f\left(y;r,p\right)=\left(\genfrac{}{}{0pt}{}{r+y-1}{y}\right){p}^{r}{(1-p)}^{y} ; y=0, 1, 2,\dots , r>0 , 0< p <1$$

(2)

where the r is known as the dispersion parameter (overdispersion) [33]. The mean value and the variance of the distribution are $\frac{r(1-p)}{p}$ and $\frac{r(1-p)}{{p}^{2}}$, respectively.

Generalized Poisson distribution (GP)

The p.d.f of GP is as follows;

$$f\left(y;\alpha,\lambda\right)=\left\{\begin{array}{ll}\left(\frac1{y!}\right)\lambda\left(\lambda+\alpha\lambda\right)^{y-1}e^{-\lambda-y\alpha}&;\;y=0,1,2,\dots\\0&;\;for\;\;y>k\;\;if\;\;\alpha<0\end{array}\right.$$

(3)

where $\lambda>0$, $\text{max}\left(-1,-\frac{\lambda }{k}\right)<\alpha \le 1$, and $k\ge 4$. The mean and variance of the distribution are $\frac{\lambda }{1-\alpha }$ and $\frac{\lambda }{{(1-\alpha )}^{3}}$ ,respectively. If $\alpha =0$, this distribution reduces to the Poisson distribution [34].

In statistical modeling, particularly in the context of regression analysis, the relationship between a dependent variable and one or more explanatory variables (${X}_{1},\dots ,{X}_{p}$) is often expressed by mathematical equations that capture the underlying dynamics of the data. A common approach uses a logarithmic transformation of the mean ($\mu$) of the outcome distribution [35], expressed as:

$${log}_{e}\left(\mu \right)=\alpha +{\beta }_{1}{X}_{1}+\cdots +{\beta }_{p}{X}_{p}$$

Zero-inflated model

Zero-inflated count models are a useful approach when dealing with datasets that have an excessive number of zeros that cannot be adequately described by standard count distributions. In other words, the purpose of these models is to account for the excess zeros in the data that cannot be explained by the count model alone. These models are a type of mixture model that combines a binary model (logit, probit, etc.) and a count model (Poisson, Negative Binomial, etc.).

The structure of a Zero-Inflated distribution is as follows:

$$P\left(Y=y;p,\Theta \right)=p+\left(1-p\right)f(y\left|\Theta )\right. ; y=0, 1, 2, \dots$$

(4)

where $f(y\left|\Theta )\right.$ is the p.d.f of a count distribution with $\Theta$ parameters and $p$ is the probability of an excess zero [36, 37]. Two indices are considered when interpreting the coefficients: Odds Ratio (OR) and Risk Ratio (RR) for two parts of the model.

Multilevel zero-inflated count regression model

For the hierarchical structure of this study, we consider a multilevel zero-inflated count distribution for the response variable (${y}_{ijk}$), where ($i=\text{1,2},\dots ,31;j=\text{1,2},\dots ,{n}_{i};k=\text{1,2},\dots ,{n}_{ij}$). Due to the nature of the data collected in different cities and provinces, three-level (TL) count models were used for data analysis. The TL regression model is as follows

$$\left\{\begin{array}{c}log\left[\frac{{\phi }_{ijk}}{1-{\phi }_{ijk}}\right]={\xi }_{ijk}={a}_{ijk}^{T}\alpha +{\delta }_{i}+{\gamma }_{ij}\\ log\left[{\lambda }_{ijk}\right]={\eta }_{ijk}={b}_{ijk}^{T}\beta +{\tau }_{i}+{v}_{ij}\end{array}\right.$$

(5)

where ${a}_{ijk}^{T}$ and ${b}_{ijk}^{T}$ refer to covariate vectors in two parts of the models. In this context, k^th refers to the households (the first level) in the j^th city (the second level) in the i^th province (the third level). ${\delta }_{i}$ and ${\tau }_{i}$ in the zero and count part of the model refer to the random effect of the province, while ${{\varvec{\gamma}}}_{ij}$ and ${v}_{ij}$ are attributed to the random effects of the city.$\delta$,$\gamma$, $\tau$, and $v$ are assumed to be independent and normally distributed with mean zero and variance${\sigma }_{\delta }^{2}$, ${\sigma }_{\gamma }^{2}$,${\sigma }_{\tau }^{2}$, and ${\sigma }_{\nu }^{2}$ respectively [32].

Model fitting and selection

The average number of cigarettes consumed per day in the household as a function of gender, age and other explanatory variables is modeled using P, NB, GP, zero-inflated Poisson (ZIP), zero-inflated Negative Binomial (ZINB) and Zero-inflated Generalized Poisson (ZIGP) regression models. The same explanatory variables are included in both the zero and count components of the zero-inflated models. In addition, the same variables are used in all models to compare the fitted models. It is important to use a set of fit indices to evaluate the models and carefully determine the best fit. The indices log-likelihood, AIC, BIC, and MSE are used for this purpose [29, 38]. Comparing AIC and BIC values between models can help determine the model that offers the best balance between fit and parsimony. Lower AIC and BIC values therefore indicate a better fitting model. MSE is a measure of the mean squared difference between the predicted and observed values. Lower MSE values indicate a better model fit as the model makes more accurate predictions. Data were analyzed using the library (glmmTMB) package in R, version 4.3.2.

Results

Information from 8571 Iranian households from 31 provinces included in this study. The mean (SD) household size was estimated at 4.40 (2.16). The average monthly income of these households was also $252.54 ($176.03). According to the results, the fewest and the most cities were assigned to the province of Fars with 29 cities and Qom with 1 city, respectively. The number of households in the provinces varies between 128 and 537 households. The mean (SD) cigarette consumption was highest in the three provinces of West Azerbaijan 5.38 (9.69), Chaharmahal & Bakhtiari 4.95 (8.87) and Qazvin 4.16 (8.87). See Table 1 for the number of households and cities in each province. About 86.6% of households were non-smokers, and the mean (SD) number of cigarettes smoked per day and month in all households were 1.91 (5.80) and 57.16 (173.59), respectively. The distribution is heavily skewed to the right, with a significant proportion of households reporting no cigarette consumption. This indicates a significant number of non-smoking households and a smaller subset of households with high cigarette consumption. See Fig. 1 for the frequency distribution of daily smoking in Iranian households. Patterns and differences in the percentage and the average daily cigarette consumption in smoking households in the various provinces are also shown in Fig. 2.

Table 1 Number of households, cities, and the mean and standard deviation of cigarette consumption in Iranian households with at least one child under 5 years by province, Iran 2021

Fig. 1

Distribution of daily cigarette consumption in Iranian households with at least one child under the age of 5 years, based on the Income and Expenditure survey conducted in 2021

Fig. 2

Variations in the percentage of smoking (left) and the average daily cigarette consumption (right) in Iranian households with at least one child under 5 years old, categorized by province in 2021

Table 2 shows the results on the characteristics of the households and their heads. Based on the results, about 62% of the household heads were 35–55 years old and the majority of household heads were male (97%). About 32% of the heads were illiterate and almost 13% had had an academic education. Only 9.5% of them were unemployed. The percentage of urban and rural households was almost equal (51% versus 49%). Household size was 4 or more in 77% of households. Most households (70%) had 2 to 3 educated members and 81% had only one working person. In 36% of households, monthly health expenditure was less than $1. About 65% of households were homeowner and most of them (60%) had low to middle income. In addition, 85% had internet access.

Table 2 Descriptive statistics and comparisons of response component (zero value or count value) by variables

The comparison of model fit indices between the different models shows that the ZIGP model generally provides the best fit according to LL, AIC and BIC values, while the ZIP model has the lowest MSE, indicating better predictive performance. The Poisson model consistently performs the worst on all indices. All indices for comparison of different statistical models are summarized in Table 3. The results show that the ZINB and ZIGP models perform better than other models due to overdispersion and zero inflation in the data. Table 4 presents the coefficients, standard errors, and confidence intervals (CI 95%) for the odds ratios (OR) and risk ratio (RR) associated with the fitted three-level Zero-Inflated Generalized Poisson (TL-ZIGP) regression model with random effects. Age significantly influences the likelihood of smoking in households. Compared to household heads under 35 years, those aged 35–54 have a decreased likelihood of not smoking, with an OR of 0.74 (95% CI: 0.62–0.88). Similarly, household heads aged 55 years and older have an OR of 0.60 (95% CI: 0.41–0.88), indicating a reduced likelihood of not smoking. Additionally, the daily smoking risk in households with heads aged 35–54 is about 14% higher than those under 35, with an RR of 1.14 (95% CI: 1.03–1.25). Gender also plays a critical role in smoking behavior. With an OR of 0.16 (95 CI: 0.07–0.37), female heads are more likely to lead a smoke-free homes than male household heads, suggesting a higher prevalence of smoking among men. The level of education of the household head is inversely related to smoking prevalence. Higher education levels correspond to higher odds of not smoking. For instance, heads with academic education have an OR of 4.30 (95% CI: 3.01–6.15), showing they are significantly less likely to smoke compared to illiterate heads. Furthermore, households with academically educated heads smoke 0.701 times less than those with illiterate heads (RR = 0.70, 95% CI: 0.56–0.88).

Table 3 Comparison of three-level model fit indices, n = 8571

Table 4 Estimates, standard errors and confidence intervals (CI 95%) for the odds ratio and relative risk associated with the fitted Three-level ZIGP regression model with random effects

Results also shows, higher health expenditures are associated with a higher likelihood of smoking. Households spending $5–30 on health monthly have an OR of 0.82 (95% CI: 0.69–0.98), and those spending $30 or more have an OR of 0.79 (95% CI: 0.62–1.00). Housing ownership status affects smoking behavior. Owner households are more likely to be non-smoking compared to those renting or under mortgage, with an OR of 0.83 (95% CI: 0.69–1.00). Additionally, renter households have a lower relative risk of cigarette consumption (RR = 0.90, 95% CI: 0.81–1.00). Households in urban areas are less likely to smoke compared to those in rural areas, with an RR of 0.92 (95% CI: 0.85–0.99). Larger living spaces are associated with lower smoking prevalence. Households with a living area of 100–150 (m²) and those with 150 (m²) or more have lower smoking prevalence compared to those with less than 50 (m²), with ORs of 1.55 (95% CI: 1.12–2.15) and 1.53 (95% CI: 1.00–2.33), respectively. High-income households have a significantly higher risk of smoking, with an RR of 1.30 (95% CI: 1.03–1.64). Ownership of a tablet or personal computer is associated with a decreased likelihood of smoking in households (RR = 0.86, 95% CI: 0.75–0.98). However, internet access does not show a significant effect on smoking behavior.

The random effects below Table 4 provide information on the variability of smoking behavior at different geographical levels, especially in cities and provinces. While there is some evidence of city-level differences in both smoking counts and the likelihood of non-smoking households, these effects are relatively small (σ² = 0.04 (0.20) for Count part; σ² = 0.09 (0.30) for Zero part). However, there is a substantial difference in the likelihood of non-smoking households between provinces (σ² = 0.90 (0.95)), suggesting that provincial factors strongly influence smoking habits.

Discussion

The study investigated smoking habits and the factors influencing cigarette consumption in Iranian households with at least one child under five years of age using zero-inflated multilevel count models. The average number of cigarettes smoked per month in Iranian households in 2017 was 85.25 [21]. However, we calculated 57.16 cigarettes per month in 2021. As studies have shown, the decrease could be due to the global decline in smoking as a result of anti-smoking policies [15] or low cigarette consumption in households with at least one child under 5 years of age, as parents are more aware of the potential health risks to their children from secondhand smoke [39]. Nevertheless, Kangavari et al. (2021) underline the considerable tobacco consumption in Iran. Based on the 2021 STEP data, they found an overall prevalence of 14.01 across all age groups [22]. In our study, 13.6% of households also reported smoking. In numerous studies, zero-inflated count models such as ZINB and ZIP were also used instead of P and NB when the data had an excessive number of zeros [40, 41]. A high number of zeros, almost 86%, i.e. non-smoking households, justified the use of zero-inflated count models in our study. In addition to the zeros, the more than 15-fold difference between the variance and the mean of smoking in the households clearly indicates an overdispersion of the data in the present study [33]. Studies have shown that ZINB regression often outperforms ZIP regression when handling over-dispersed count data with excess zeros [42]. Fatih Tuzen's 2017 research on Turkish adolescents' daily cigarette consumption indicated that ZINB and Negative Binomial Hurdle models outperformed others [38], while a 2020 study identified the NB model as the best fit for daily cigarette consumption of college students [43]. Lastly, Andriyana et al. (2023) highlighted the effectiveness of the minimax concave penalty in analyzing cigarette consumption factors in poor households using penalized ZINB regression [23]. While many studies have applied known count distributions and their zero-inflated models to data sets such as smoking, fewer have examined these models in a multilevel framework. Moghimbeigi et al. (2009) used a TL-ZIP model for Iranian adolescent smoking habits [44], while Almasi et al. found that the multilevel ZIGP model outperformed multilevel ZIP and multilevel ZINB models for DMFT data [45]. This study considers geographic patterns (Fig. 2) indicating a higher prevalence of smoking in cold, mountainous regions of Iran [17], and uses multilevel models to account for differences in smoking behavior between provinces and regions [46]. Although the TL-ZINB and TL-ZIGP models yielded similar assessment indices, the TL-ZIGP model fitted the data best, while the TL-ZIP model fitted the data worst.

The results of the TL-ZIGP model show that the socio-demographic characteristics of household heads and various household factors significantly influence cigarette consumption. The most important influencing factors include income, cigarette prices, school performance, work-family dynamics and urban or rural residence [21, 47]. Despite some progress, tobacco use remains a major public health problem in Iran, especially among men [48, 49]. It is noteworthy that in households headed by women, the likelihood of smoking and average cigarette consumption are lower than in households headed by men, which is consistent with the findings of other studies [17, 23].

The results based on the logistic regression model on cigarette smoking for the 2017 income-expenditure data show a higher probability of smoking among married and unemployed older people. [19]. In our study, marital and employment status of household heads showed no significant correlation with smoking probability, though single, divorced, or widowed heads were slightly more likely to smoke than married ones. A 2023 study found that the age of the household head and socioeconomic status significantly influence smoking behavior [46]. Our findings suggest that both the probability of smoking and average household cigarette consumption increase with the age of the household head, corroborated by Sohrabi et al. (2023), which noted rising tobacco use prevalence among men as they age [17].

Numerous studies have established a link between smoking and education and have shown that smoking is more prevalent in societies with lower levels of education [18, 23, 49]. Our result show that higher levels of education among household heads correlate with smoke-free households and lower smoking rates. Interestingly, we also found that high-income households tend to have higher smoking rates than low-income households. However, a 2012 study suggests that improvements in income above the poverty line may increase smoking cessation rates [50]. Handra (2016) found that in poorer households, higher income is associated with higher spending on smoking, implying that lower cigarette consumption in low-income households may be due to financial constraints on spending on tobacco products [51].

Furthermore, the results indicate that while tenant households are less likely to be non-smokers compared to owner households, they still exhibit lower smoking rates. Keshavarz et al. (2019) found that homeowners are generally more likely to be non-smokers in both rural and urban areas [47].

The results of our study indicate that despite a lower smoking rate in these households, the prevalence of smoking remains relatively high in Iran, especially in the western regions compared to the south and east. Therefore, Iranian households need to reconsider their smoking habits, which underlines the importance of strengthening tobacco control policies and programs. Policies can reduce smoking and protect vulnerable populations from SHS by raising awareness of the harms of smoking, supporting anti-smoking campaigns, levying high taxes, banning smoking in the media and promoting smoke-free environments, especially in households with young children and adolescents.

Limitations

Since this study was a secondary study of the Iranian Household Income-Expenditure survey, a better understanding of the patterns of cigarette consumption in Iranian households requires more information on the number of smokers, the role of smokers in the household members such as parents, grandparents or siblings, information on the diseases of household members, the social status of individual members, and so on, that were not available in the survey. Another limitation of this study also is the possibility of measurement error resulting from the use of self-report data. Self-reported smoking behavior may be subject to recall error, social desirability bias, or underreporting, which may lead to an underestimation of smoking prevalence and associated factors. Furthermore, due to the cross-sectional design of the study, it is not possible to establish causal relationships between the variables. Although associations were found, longitudinal studies are needed to draw causal conclusions.

Future directions

Future studies could lead to a more accurate assessment of smoking behavior in Iranian households by including comprehensive data on households and the individual, social, psychosocial and health characteristics of their members. New qualitative studies are also needed to explore the effective factors that influence smoking behavior in Iranian households and provide insights for culturally appropriate smoking cessation programs. In addition, future studies should evaluate the effectiveness of smoking cessation interventions implemented in Iran to refine these interventions and optimize resource allocation.

Conclusions

To summarize, choosing the most appropriate model for the data in this structure is crucial, as there is no single model that fits best. The data suggest that women are more likely to abstain from smoking than men and that higher levels of education are associated with a greater likelihood of not smoking. Higher levels of education are also often associated with better health literacy and greater awareness of the risks of smoking. Broadcasting educational programs or short clips at different levels of society, whether in schools or universities as well as in the public media, can therefore effectively inform all family members about the harmful effects of smoking. This study underlines the role of education and information dissemination in public health campaigns. The results also show that the prevalence of smoking correlates with increased health care costs in households, which could exacerbate the consequences of smoking and highlights the economic burden of smoking in these households. Therefore, it could be helpful to implement effective smoking cessation programs and policies, such as increasing cigarette prices and restricting access to cigarettes. Therefore, cooperation between the government and cultural leaders in implementing effective laws and promoting a supportive culture can significantly reduce smoking and its prevalence.