The Diversity in Tobacco Use Among Women of Reproductive Age (15–49 Years) in Pakistan: A Secondary Analysis of a Multiple Indicator Cluster Survey 2016–2018

Abstract Introduction Tobacco use in women of reproductive age group (15–9 years) may contribute to poor maternal and child health outcomes. We estimated the prevalence of smokeless tobacco (ST), cigarette, and waterpipe use among these women in Pakistan and explored associations with socio-economic differences and birth weights after recent childbirths (past 2 years). Aims and Methods We used the Multiple Indicator Cluster Surveys datasets from three provinces in Pakistan. Prevalence was generated for current use of tobacco among women with recent childbirth (WWRC) and without recent childbirth (WWoRC). We further compared socio-economic status and birth weights between tobacco users and non-users. Results The prevalence of current ST use was 1.4% and 2.5%, and waterpipe smoking was 1.2% and 1.8%, in WWRC and WWoRC, respectively. ST use varied across Pakistan’s provinces by 13 percentage points in WWRC and WWoRC while waterpipe smoking by 10 and 15 percentage points, respectively. The odds of using any form of tobacco were significantly lower with higher levels of education or wealth index. Compared to children born to mothers who did not use tobacco, the birth weight of children born to mothers who smoked waterpipe was on average 0.83 kg (95% CI −1.6 to −0.1) lower whereas it was 0.33 kg (95% CI −0.9 to 0.3) higher for ST users. Conclusions Tobacco use among women of reproductive age in Pakistan varied significantly based on province and tobacco type. Waterpipe smoking was associated with a reduction in birth weight. Raising awareness of the harms of tobacco use among women of reproductive age and targeting interventions in high-burden regions of Pakistan should be prioritized. Implications The prevalence of current smokeless tobacco use and waterpipe smoking varied significantly in Pakistan (~13 and ~15 percentage points, respectively) among women of reproductive age, and there seems to be little change in tobacco use behavior around pregnancy and early motherhood. The study highlights a need to better understand the prevalence of tobacco use in Pakistan, and further contextual research is warranted to understand the reasons for such high tobacco use in certain areas. This information highlights the need for improving current tobacco control interventions and policies, including measures that could help reduce its use and prevent its uptake. Furthermore, compared to children born to mothers who did not use any form of tobacco, we found a statistically significant difference (830 g) in birth weight for babies born to mothers who at the time of the interview smoked waterpipe and a non-statistically significant difference of about 330 g for children born to mothers who at the time of the interview used smokeless tobacco. In addition to our findings, some literature suggests the association of waterpipe smoking with low birth weight. A possible explanation could be the exposure to smoke that contains toxicants from the tobacco product itself in addition to the burning of charcoal, which is required when using waterpipe. Waterpipe smoking also results in high exposures to carbon monoxide and polycyclic aromatic hydrocarbons, which are known to be carcinogenic and teratogenic. In terms of smokeless tobacco use, further research is warranted to understand its impact on birth outcomes through longitudinal studies accounting for diverse products, its constituents and the amount of consumption.


Introduction
2][3][4] This is in addition to several other harms of tobacco, such as cardiovascular diseases and cancers. 5,6Most tobacco risk estimates originate from high-income countries (HIC) and are based on cigarette smoking.Yet in many low-and middle-income countries (LMICs), smokeless tobacco use and waterpipe smoking are equally or more popular, particularly among women. 7,8aterpipe smoking is also gaining worldwide popularity among youth. 9,10 South Asia, tobacco use is particularly diverse, with most women using smokeless tobacco and waterpipe-also known as hookah/sheesha/nargileh. [11][12][13] The age-and sex-specific standardized prevalence of smokeless tobacco among adults aged 15 and older, spanning 204 countries, indicated that approximately 273.9 million people globally used smokeless tobacco in 2019.The majority of these users, about 228 million (83%), were located in the South Asia region, with the highest prevalence observed among females in South Asia (12%). 14hile cigarette smoking rates remain low in South Asia, other tobacco products are highly prevalent among women due to their accessibility and social acceptability. 12,15Social factors such as education, income, perceived health benefits, social acceptability, and peer habits determine its intake. 16,17urthermore, this high prevalence also extends to its use during pregnancy and breastfeeding. 12,18n Pakistan, the overall prevalence of smokeless tobacco use, waterpipe smoking, and cigarette smoking among women is 3.7%, 1.1%, and 1.0%, respectively. 19owever, consumption rates vary; for example, within urban settlements in Karachi, smokeless tobacco prevalence among women can be as high as 42%, 20 while another study reported the prevalence of waterpipe smoking as 41%. 21This indicates that consumption is likely to be determined by socio-ecological factors, which are currently unreported in the literature.A good understanding of these socio-ecological factors is crucial for any future research and policy work to address tobacco use in women of reproductive age group in Pakistan.
Tobacco use in any form is harmful to human health.While, the causal link between cigarette smoking and adverse health effects is well-established, the evidence on the harms of smokeless tobacco and waterpipe smoking is also strong and emerging.7][28][29] Despite the notable harms of smokeless tobacco and waterpipe smoking and its high prevalence in certain populations, there is limited literature on their effect on perinatal outcomes. 30here is a dearth of literature on the prevalence, associations and effects of smokeless tobacco, and waterpipe use among women of reproductive age.Furthermore, recent tobacco use estimates, especially based on the different types of tobacco are not available.These estimates are required to understand the current tobacco-related burden among women of reproductive age.This study aimed to understand the environmental and socio-economic reasons for the variation in prevalence estimates in Pakistan and the association between tobacco use and birth weight.The findings from this study will inform policy and contribute toward intervention development to address tobacco use among women of the reproductive age group.Furthermore, understanding the fundamentals of tobacco use among these women will enable tailored preventive measures to be implemented to avoid the uptake of tobacco in the first place.

Data
We analyzed cross-sectional survey data from the Multiple Indicator Cluster Surveys (MICS) VI datasets (most recent) from Sindh (2018), Khyber-Pakhtunkhwa (KP) (2016/17) and Punjab (2017) provinces of Pakistan.The (freely available) MICS data is collected by the United Nations International Children's Emergency Fund (UNICEF).It is a large, representative dataset on maternal and child health collected by standardized methods in each province, by first identifying the rural and urban areas in all districts as main sampling strata, then systematically selecting a specified number of census enumeration areas within each stratum, and finally selecting a systematic sample of 20 households in each enumeration area. 31

Measures
The questions asked during the survey and responses for the variables used in this study are mentioned below:

Tobacco Use
Women were asked if they ever tried or used ST, smoked cigarettes, and smoked other non-cigarette forms of tobacco (referred as waterpipe smoking in this paper), which were reported as a binary variable (yes/no).Women who reported ever use were further asked about their tobacco use within the past month (binary variable: yes/no) referred as current tobacco use in this paper.In this paper, we are focusing on the current use of tobacco.

Birth Weight
All women were asked if they had a live birth in the past 2 years as of the date of the interview (yes/no), and those who responded yes, were further asked additional questions on maternal and child history, including birth outcomes.Birth weight data were collected by asking women whether the child was weighed at birth (yes/no).If the answer was yes, they were further asked if they had the child's weight recorded in the delivery record card.Based on that, birth weight was reported in kilograms specifying whether the value originated from the delivery record card or by recall.

Sociodemographic Covariates
Age (in completed years at the time of the interview) recorded as a continuous value, highest education attained (primary, middle, secondary, and higher), area of residence (rural or urban), ecological data (ie, the divisions and sub-divisions in each province), and quintiles of wealth index (poorest, second, middle, fourth, and richest) were reported.Age and education were individual indicators at the mother's level, while the area of residence, ecological data and wealth index were part of household indicators.The wealth index was constructed using principal components analysis, 32,33 based on various characteristics related to a household's material wealth. 31

Statistical Analysis
The secondary data analysis was performed in STATA (StatsCorp, 2017).Descriptive summary statistics for the characteristics (age, education, and wealth index) of all women interviewed (15-49 years) who reported birth history in the previous 2 years (women with and without live birth in the previous 2 years) were calculated by provinces and their divisions.Prevalence estimates for current smokeless tobacco use, cigarette smoking, and waterpipe smoking, and 95% confidence intervals (CI), were estimated and further stratified by divisions in each province.Pooled estimates for each province and for the country were also generated.Sampling weights were accounted for differential probabilities of selection and participation in the calculation of summary statistics and prevalence estimates.Sampling weights were calculated using the svy command in STATA based on the weight variables provided in the MICS dataset.
Regression analyses were conducted by combining the data from all divisions after accounting for the possible clustering effect by division.The following analyses were performed: Logistic regression was used to explore the effect of maternal covariates (namely, residence, education, socioeconomic status, and age) on current use of tobacco for each of the following tobacco forms: ST, cigarette, and waterpipe smoking.In each case, non-tobacco users formed the reference group.
Multiple linear regression analysis to estimate the effect of current tobacco use on birth weight (outcome variable reported as a continuous variable in kilograms) among women who have had a live birth in the past 2 years, compared to non-tobacco users.The independent variables were current exclusive smokeless tobacco use, exclusive cigarette smoking, and exclusive waterpipe smoking (dual users were not considered due to extremely small observations).For this particular analysis, we created a new variable with exclusive use of any form of tobacco to estimate their effect on birth weight compared to non-tobacco users.
For all regression analyses, we reported 95% CIs for the estimates.A p-value less than .05was considered statistically significant.

Results
The study included data of 85 412 women of reproductive age: 31 210 women with and 54 202 women without a live birth in the past 2 years.Furthermore, birth weight data were only available for 15.4% (n = 4790) of women with recent childbirth.The mean ages of women with and without recent childbirth were 28.3 and 36 years, respectively.About 70% (women with recent childbirth) and 63% (women without recent childbirth) lived in rural areas; more than half (52% and 57% among women with and without a recent childbirth, respectively) had no formal education.Wealth index distribution was similar among women with and without a recent childbirth.Detailed characteristics of these women are reported in Tables 1 and 2.
Higher education was inversely associated with all forms of tobacco; odds were significantly lower for smokeless tobacco use (adjusted odd ratio (aOR) = 0.1, 95% CI: 0.05-0.3),cigarette smoking (aOR = 0.1, 95% CI: 0.00.6) and waterpipe smoking (aOR = 0.2, 95% CI: 0.1-0.4) as compared to non-users of that type of tobacco.The association between education and smokeless tobacco use was statistically significant across all levels; that is, with every incline in educational level, the odds of smokeless tobacco use declined significantly.Socio-economic status was also inversely associated with all three forms of tobacco use across each wealth quintile compared to the poorest quintile; odds for tobacco use decreased significantly with every increase in wealth quintile.This disparity gradient for education and socio-economic status is shown as forest plots in Figure 2. Further details of the regression analyses are listed in Table S3.
Birth weights were recorded for 15.4% (n = 4790/31 210) of the women with recent childbirth: 14% (n = 732) of these had the weight logged in the delivery record card, 70% (n = 3545) reported birth weight based on recall, and 11% (n = 513) did not provide any birth weight information.As most women did not have birth weight logged in the delivery record card, to make use of all available data, we used the birth weight reported both via delivery record and by recall.The mean birth weight was 3.55 kg (SD = 2.46).We found a statistically significant difference in birth weight of about 0.83 kg (95% CI: −1.6 to 0.1) in children born to mothers who, at the time of the interview, smoked waterpipe compared to non-tobacco users.While for smokeless tobacco use (0.33 kg, 95% CI: −0.9 to 0.3) and cigarette smoking (0.35 kg, 95% CI: −1.5 to 2.2) the differences were statistically not significant (Table 3).

Discussion
We utilized data from a large representative sample of women and estimated the prevalence for three types of tobacco use (smokeless tobacco, cigarettes, and waterpipe) according to the history of recent childbirth.The data suggests that there is variation in prevalence between provinces and by type of tobacco, with cigarette smoking as least prevalent across all three provinces, smokeless tobacco use most prevalent in Sindh (~6%) and waterpipe smoking most prevalent in Punjab (~3%).Furthermore, the prevalence of tobacco use varied even more significantly across the administrative divisions; 13 percentage points for smokeless tobacco use in Sindh and 15 percentage points for waterpipe smoking in Punjab.Among a subset of women (31 210) with recent childbirth, similar prevalence patterns were observed.The findings are worrying because it indicates that there is little change in tobacco use behavior around pregnancy and early motherhood.Furthermore, the high prevalence of tobacco use in certain geographical areas (administrative divisions) may go unnoticed due to relatively lower national estimates and the opportunity to intervene and benefit public health may not be available.Potential factors influencing these variations include cultural association and social acceptance of specific tobacco use among women in certain regions and limited awareness of tobacco-related harms. 13,34High smokeless tobacco use in certain areas may also be driven by easy access to cheap products manufactured locally.Additionally, there may be variations in regulating tobacco products and support for women to reduce or quit tobacco. 13,20This highlights the need for tailored public health measures to increase awareness and offer support to women in reducing or quitting tobacco, across all types of tobacco products.
Higher education was significantly associated with a decrease in tobacco use.This inverse association was most distinct and significant across all educational levels with smokeless tobacco use; aOR dropped from 0.5 to 0.1 from primary to higher level education compared to those with no formal education.Socioeconomic status was also significantly associated with a reduction in all forms of tobacco use.These  education and socioeconomic disparities are consistent with existing literature. 11,35However, the significantly lower odds of tobacco use found among women with high levels of education in Pakistan is an important finding for a country where female literacy remains less than 50%. 36This is crucial because tobacco use and social inequalities are recognized as factors that exacerbate health disparities. 37,38These disparities occur due to easy access to cheap unregulated and highly toxic tobacco products in poor neighborhoods, limited awareness about tobacco-related harms, poor health literacy, and access to public health initiatives among those with poor educational attainment, which can in turn lead to lower rates of tobacco cessation, thus further widening of the health gap. 13,37To assist women in reducing or quitting tobacco use, public health measures should be designed to incorporate socio-cultural contexts and consider implementing pictorial warnings not only on cigarette packaging but also across all forms of tobacco products.We found a statistically significant difference (830 g) in birth weight for babies born to mothers who at the time of the interview smoked waterpipe compared to children born to mothers who did not use any form of tobacco.There is some literature on this suggesting the association of waterpipe smoking with low birth weight.A study conducted among pregnant women in Egypt reported lower mean birth weight (p-value <.001) compared to non-smokers, however women who smoked included cigarette and waterpipe smoking, though waterpipe was more prevalent. 39Another study based on a prospective cohort in Iran, reported the adjusted risk of low birth weight to be twice among waterpipe smokers. 40he possible explanation could be the exposure to smoke that contains toxicants from the tobacco product itself in addition to the burning of charcoal which is required when using a waterpipe. 41Waterpipe smoking also results in high exposures to carbon monoxide and polycyclic aromatic hydrocarbons, which are known carcinogens and teratogens. 41,42In terms of smokeless tobacco use and birth weight, our findings suggest that the birth weight was about 330 g less for children born to mothers who at the time of the interview used smokeless tobacco compared to children born to mothers who did not use any form of tobacco, however this was not statistically significant.One other study conducted in Pakistan also reported no statistically significant association between smokeless tobacco use before conception and birth weight (RR = 0.96, 95% CI 0.7-1.3). 43However, other studies have reported a weak to moderate evidence on smokeless tobacco use during pregnancy and low birth weight. 44,45  of 105 g (p value = .006)in birth weight, which was consistent after adjustment for gestational age. 45The strong evidence on the effect of combustible tobacco on birth weight, may not be applicable for smokeless tobacco, and needs further longitudinal studies, accounting for the diverse smokeless tobacco products and amount of consumption.Furthermore, our findings of birth weight and cigarette smoking were not statistically significant, possibly due to very small number of smokers in our cohort, however this association is already well-established in the literature. 2,46e highlight a significant missed opportunity; most newborns were either not weighed at birth or did not have a record of it.This is consistent with previous MICS datasets and warrants the urgent need for better recording of birth events to help improve maternal and child health. 47Especially with diverse tobacco use in South Asia and the dearth of literature, such large, nationally representative data hold significant value in understanding the effect of these forms of tobacco on peri-natal outcomes.

Strengths
The prevalence estimates of tobacco use (smokeless tobacco, cigarette smoking and waterpipe smoking) from a nationally representative sample along with estimates based on administrative divisions is novel.Certain areas with high levels of tobacco use (geographical location and type of tobacco) need further exploration to understand the contextual factors.The association of smokeless tobacco use and waterpipe smoking with birth weight from a representative sample is also a novel finding.

Limitations
First, among women with recent childbirth, <15% had birth weight of their child recorded which indicates the challenges in obtaining this data.However, we compared the sociodemographic characteristics of women who reported birth weight based on delivery record and recall (Table S4) and found that women who provided birth weight data based on recall were residing more in rural areas, were less educated and belonged to low socio-economic status.Second, this is a cross-sectional survey and hence causal links cannot be assumed.In addition, for the analysis looking at the association between birth weight and tobacco use, there is a major assumption that those who are currently using tobacco might have also used the same form of tobacco during their pregnancy.Furthermore, we did not adjust for several factors that affect birth weight (eg, gestational age, maternal weight) due to the unavailability of such data.Further research such as a large cohort is required to understand the effect of different forms of tobacco on birth weight.Only a few women reported cigarette smoking, which meant that we did not have sufficient power to provide precise estimates of this sub-group.

Figure 1 .
Figure 1.Prevalence estimates (weighted) of current tobacco use (smokeless tobacco use, cigarette smoking, and smoking waterpipe) among women with and without recent childbirth.

Figure 2 .
Figure 2. Forest plots showing associations between education and wealth index and current tobacco use by type.(cigarette smoking, smoking waterpipe and smokeless tobacco use).

Table 1 .
Socio-Demographic Distribution (Weighted) of Women With a Live Birth in the Past 2 Years (15-49 Years)

Table 2 .
Socio-Demographic Distribution (Weighted) of Women Without a Live Birth in the Past 2 Years (15-49 Years)

Table 3 .
Estimates of Multiple Linear Regression Analysis for BirthWeight and Current Smokeless Tobacco Use, Cigarette Smoking, and Waterpipe Smoking

Multiple linear regression analysis Coef. (95% CI) p-Value Current tobacco use (reference = non-tobacco users)
Note: Total observations = 4507; the absolute numbers of women who exclusively used smokeless tobacco were 63, smoked cigarettes were 7, and smoked waterpipe were 37.