Use of Smokeless Tobacco Before Conception and Its Relationship With Maternal and Fetal Outcomes of Pregnancy in Thatta, Pakistan: Findings From Women First Study

Abstract Introduction Smokeless tobacco (SLT) consumption during pregnancy has adverse consequences for the mother and fetus. We aimed to investigate the effects of maternal pre-pregnancy SLT consumption on maternal and fetal outcomes in the district of Thatta, Pakistan. Aims and Methods We conducted a secondary data analysis of an individual randomized controlled trial of preconception maternal nutrition. Study participants were women of reproductive age (WRA) residing in the district of Thatta, Pakistan. Participants were asked questions regarding the usage of commonly consumed SLT known as gutka (exposure variable). Study outcomes included maternal anemia, miscarriage, preterm births, stillbirths, and low birth weight. We performed a cox-regression analysis by controlling for confounders such as maternal age, education, parity, working status, body mass index, and geographic clusters. Results The study revealed that 71.5% of the women reported using gutka, with a higher proportion residing in rural areas as compared with urban areas in the district of Thatta, Pakistan. In the multivariable analysis, we did not find a statistically significant association between gutka usage and anemia [(relative risk, RR: 1.04, 95% confidence interval, CI (0.92 to 1.16)], miscarriage [(RR: 1.08, 95% CI (0.75 to 1.54)], preterm birth [(RR: 1.37, 95% CI (0.64 to 2.93)], stillbirth [(RR: 1.02, 95% CI (0.39 to 2.61)], and low birth weight [(RR: 0.96, 95% CI (0.72 to 1.28)]. Conclusions The study did not find an association between gutka usage before pregnancy and adverse maternal and fetal outcomes. In the future, robust epidemiological studies are required to detect true differences with a dose–response relationship between gutka usage both before and during pregnancy and adverse fetomaternal outcomes. Implications While most epidemiological studies conducted in Pakistan have focused on smoking and its adverse outcomes among males, none of the studies have measured the burden of SLT among WRA and its associated adverse outcomes. In addition, previously conducted studies have primarily assessed the effect of SLT usage during pregnancy rather than before pregnancy on adverse fetal and maternal outcomes. The current study is unique because it provides an insight into the usage of SLT among WRA before pregnancy and investigates the association between pre-pregnancy SLT usage and its adverse fetomaternal outcomes in rural Pakistan.

A c c e p t e d M a n u s c r i p t

Implications
While most epidemiological studies conducted in Pakistan have focused on smoking and its adverse outcomes among males, none of the studies have measured the burden of SLT among women of reproductive age (WRA) and its associated adverse outcomes. In addition, previously conducted studies have primarily assessed the effect of SLT usage during pregnancy rather than before pregnancy on adverse fetal and maternal outcomes. The current study is unique because it provides an insight into the usage of SLT among WRA before pregnancy and investigated the association between pre-pregnancy SLT usage and its adverse feto-maternal outcomes in rural Pakistan.
A c c e p t e d M a n u s c r i p t Introduction More than 300 million people consume smokeless tobacco (SLT) worldwide. 1,2 SLT products are relatively cheaper than manufactured cigarettes and some are viewed as safer substitutes for smoking. [3][4][5] In some areas, the use of SLT products is considered a socially acceptable cultural norm resulting in higher usage of SLT. [6][7][8][9] There is a long history of using SLT products globally, including the United States of America, Europe, Asia, and parts of Africa with differences in the epidemiology of SLT usage in these regions. [10][11][12] Consumption of SLT products is increasing, not only among men but also among women in South Asian countries. 1,13 For example, an updated analysis of data from 127 countries reveal that more than 85% of the ST-related burden is found in South and Southeast Asia. 14 Further, India accounts for 70% of the DALYs lost due to SLT followed by Pakistan (7%) and Bangladesh (5%). 14 One of the main reasons for the extensive use of SLT in Pakistan is its ease of access and low prices. 15 In Pakistan 15% of male and 10% of females use SLT in different forms such as gutka, Nass/Naswar, paan, and betel nut. 16,17 Although studies have evaluated the carcinogenic effects of SLT, 18,19 there is a dearth of evidence on assessing the effect of SLT use during pregnancy, in general, and particularly before pregnancy on feto-maternal outcomes. 20,21 The literature reveals SLT use during pregnancy may increase the risk of adverse pregnancy outcomes such as, maternal anemia, miscarriage, and stillbirths, low birth weight (LBW), and decreased birth length. [22][23][24][25][26][27] For example, a study in Sweden demonstrates that different SLT products have modest effects on fetal growth and increased risk of preterm delivery. 28 Likewise, a study conducted in India found that SLT use is associated with an average reduction of 105 g in birth weight and a reduction in gestational age of 6.2 days. 29 A study conducted by Gupta et al found an increased risk of LBW, preterm birth (PTB), and stillbirth among tobacco chewers as compared to non-chewers. 30 Similarly, Pratinidhi et al conducted a cohort study to assess the effect of tobacco use on pregnancy outcomes in Indian women. 21 The authors found a significantly increased risk of LBW and stillbirth among tobacco users when compared with non-users. 21 Likewise, one systematic review and meta-analysis suggested a positive and statistically significant association between SLT use and adverse pregnancy outcomes including LBW, PTB, and stillbirths in Indian women. 31 However, A c c e p t e d M a n u s c r i p t studies included in the meta-analysis had some limitations. First, there were only three studies that were primary conducted in India thus ignoring the effects of SLT in other countries of Asia. Second, some studies did not address potential confounders, such as age, parity, education, body mass index, and working status of the women. Besides, outcomes in these studies were measured using medical records rather than measuring outcomes directly. Moreover, these studies mainly assessed the use of tobacco during a pregnancy instead of assessing the use of tobacco before pregnancy. 31 Although these study findings support the association between SLT usage during pregnancy and adverse fetal and maternal outcomes, there is a lack of evidence regarding the association of SLT use before pregnancy (before conception) and feto-maternal outcomes. Since the exposures to SLT tend to accumulate over time before conception, the preconception period is important to capture these harmful exposures to assess their true effects on adverse outcomes. This is because women might tend to change their SLT use after they become pregnant thus preventing one from capturing actual exposure. Besides, exposures to such harmful exposures require a sufficient latent or induction period to produce the adverse effects, therefore, we aimed to measure the exposure to SLT before pregnancy instead of during pregnancy. Hence, this study aimed to investigate the association between gutka (SLT) use prior to pregnancy and maternal and fetal outcomes in Pakistan.

Study design and sample size
The sample for this secondary analysis comes from the Women First (WF) Preconception study that was conducted in the Democratic Republic of Congo, Guatemala, India, and Pakistan. 32 This was an individually randomized, controlled efficacy trial (NCT01883193) that included 3 arms: women in Arm 1 started a nutritional supplement ≥3 months before conception and continued through delivery; women in Arm 2 started the same nutritional supplement at the end of the first trimester, and women in Arm 3 (control arm) received no nutrition supplements. 32 The main objective of the WF study was "to determine the benefits to the offspring of women of commencing a daily comprehensive maternal nutrition supplement ≥ 3 months before conception versus the benefits of commencing the same supplement at 12-14 week gestation and also to compare offspring outcomes with those of a third Downloaded from https://academic.oup.com/ntr/advance-article/doi/10.1093/ntr/ntaa215/5934513 by Aga Khan University, Health Sciences Library user on 11 December 2020 A c c e p t e d M a n u s c r i p t control arm". 32 Sample size of the primary WF study was based on having 80% power and maintaining a study-wide Type I error of 0.05 across all planned primary hypothesis tests. 32 Assuming an α-level of 0.00625, a 2-sided test, and an SD of 1.0 for the primary outcome (birth length), 192 women per arm in each country were needed to detect an effect size of 0.37 with 80% power for the primary outcome of birth length in the main study. Around 240 women were required per arm enter Phase 2 (pregnancy) within each country after accounting for 20% attrition during pregnancy.
Furthermore, this number was increased to 480 women per arm to be enrolled in each country, assuming that only 50% of women enrolled would become pregnant. 32 Further details of the WF study are discussed in depth elsewhere. 32,33 Since the WF study was not designed to assess the effect of gutka usage on adverse feto-maternal outcomes, we did a power analysis to evaluate whether the sample size for the current study was sufficient to answer the research question. We did power analysis for adverse outcomes reported commonly in the literature such as LBW, PTB, and stillbirth using the effect size from a recently conducted systematic review and meta-analysis. 34 The findings of the power analysis revealed that our study was powered appropriately to detect the desirable effect sizes of 1.88 for LBW, 2.54 for PTB, and 2.85 for stillbirths. 34 More specifically, considering the proportion of various outcomes in our study population (LBW: 15 to 20%; PTB: 15 to 17%; stillbirth: 3.4 to 5.6%; pregnancy loss or miscarriage: 15 to 16%; Anemia: 50 to 75%), [35][36][37] having 80% power and maintaining a level of significance as 5% we need to have a minimum of 65 to 175 women per group. Our study had at least 197 women in each group, thus was sufficiently powered to detect the true difference between gutka users and adverse fetal and maternal outcomes.

Study Participants
Study participants were married WRA residing in the district of Thatta, Pakistan. Women were identified through tracking women in a maternal and newborn health registry that is active in Thatta district, in addition to household surveys, health care facilities, word-of-mouth, and meetings with the traditional birth attendants and community leaders. 32,33 Women were eligible to participate in the A c c e p t e d M a n u s c r i p t primary WF trial if they were 16-35 years old; had 0-5 children, were not using or planned to use family planning methods; and planned to conceive during the following 18 months. 32,33 The flow of study participants during different phases of the study Figure 1 depicts the flow of participants that were included in this secondary data analysis. At the enrollment, we screened 3554 married and non-pregnant women. Of those, 2095 women were eligible for the study. 32 Of the eligible women, 2013 women agreed to participate and were enrolled. Data on gutka consumption were collected from 2013 women before they became pregnant. These 2013 women were followed longitudinally unless they became pregnant within 3 months of enrolment. This number decreased as these women were followed until they became pregnant and delivered their babies (figure 1). As per the protocol of the main WF study, women had to become pregnant after three months of enrollment, otherwise, their participation in the study was discontinued without collecting any further data from them. Of 2013 women, the participation of 1135 women was discontinued, with the most common reason being pregnancy within three months of enrolment (90%). Thus, out of 2013 women, 879 (43.6%) women became pregnant after three months of enrollment, and they were followed until the end of their pregnancy. After pregnancy, maternal and fetal outcomes of 879 pregnant women were monitored and recorded during biweekly home visits.
After becoming pregnant, the participation of 156 additional women (17.7%) was discontinued, mostly due to early pregnancy loss/miscarriage before 20 weeks of gestation (n=152). Thus, 723 (82.2%) women were followed until they delivered their baby. Of these, 124 (17.1%) women delivered before and 599 women (82.8%) delivered their babies after 37 weeks of gestation (full-term deliveries). Of the 723 deliveries, 26 (3.4%) were reported as stillbirths. Birth weight data were available for 702 babies (97%) and 229 (31.6%) babies were found to have a low birth weight (less than 2500gms) and 473 (67.3%) had a normal weight of more than 2500gms. Due to the longitudinal nature of the study and the range of fetomaternal outcomes, the sample size varied for different outcomes in the analysis. For example, the data were collected for maternal anemia at the enrollment and, therefore, data were available for 2013 women to assess the association between gutka and maternal anemia. On the other hand, data for pregnancy loss were only available for 879 women, as A c c e p t e d M a n u s c r i p t these were the women who became pregnant after three months of enrolment and were followed until delivery or pregnancy loss. Due to pregnancy losses, not all women were followed until delivery, therefore, data for the outcome of stillbirth and preterm delivery were available for 723 women.
Lastly, we were able to collect data on birth weight for 700 live births and 2 stillbirths, therefore, the sample size for assessing the relationship between gutka usage and low birth weight was 702.

Study setting and data collection
Although it was a multi-country study, we only analyzed data from Pakistan for the current study. In  38 We included all women who completed the WF primary study and had delivery outcomes recorded. At the time of enrollment, we asked questions from women about their socio-demographic characteristics, food insecurity, indoor air pollution, obstetric history, and maternal health status. We measured their weight, height, body mass index (BMI; in kg/m 2 ), waist circumference, and hip circumference before they became pregnant. Field workers administered a urine pregnancy test to confirm the pregnancy status of the women if a woman had missed their periods, which was followed by conducting an ultrasound to confirm the pregnancy. We assessed SLT usage by asking the question "Do you currently chew tobacco" (Yes or No). Additionally, women may have continued consuming tobacco during pregnancy, or some may have stopped consuming tobacco due to their pregnancy.
Since we did not collect the data of tobacco use during the pregnancy, this study assessed only the association between pre-pregnancy gutka usage and adverse feto-maternal outcomes.

Primary exposure and outcome variables
The main exposure variable for this analysis was the use of gutka (the most common type of SLT) before pregnancy. The main feto-maternal outcomes assessed included maternal anemia: Hb levels of less than 12gm/dL; 39 miscarriage: unintended loss of a fetus before the first 20 weeks of gestation as A c c e p t e d M a n u s c r i p t determined by the reported last menstrual period and confirmed by ultrasound; preterm birth: delivery at <37 weeks of gestational age confirmed by ultrasound; 40 stillbirths defined as the birth of a dead fetus at the gestation week of 20 or more confirmed by ultrasound; and LBW: the birth weight of < 2500gm. 41,42 We measured Hb levels (gm/dL) of women by using Hemocue machines. At the time of analysis, we categorized Hb into a binary variable (anemic: < 12gm/dl and non-anemic: ≥ 12 gm/dl). 39

Confounders
Based on the literature review potential confounders included were mother's age (years), parity (number of children ever born after 20 weeks of gestational age), which was categorized into nulliparous (no child born), 1 to 4 children, or at least five children, mother's education (illiterate or literate), current working status (Yes/No), cooking place (separate building to cook, cooking takes place outdoors or in the house), food insecurity (Yes/No), place of residence (urban/rural), BMI was calculated as a continuous variable by taking a ratio of weight (in kilograms) to the square of height (in meters). We categorized BMI as underweight (< 18.5 kg/m2), normal (18.5-24.9 kg/m2), overweight or obese (>25.0kg/m2).

Data Management
Two mechanisms were placed for data management, one in the field in Thatta and the other in the Aga Khan University (AKU) Karachi Pakistan located at the distance of 98 kilometers from Thatta. We hired five experienced field supervisors, who supervised field workers. These field supervisors checked the filled data collection forms randomly and edited the forms every week before transmitting to AKU for second editing. In the field, the forms were edited, and an error list was generated by the field supervisors to be corrected in the field by re-visiting the participant to address the missing information and inconsistent errors. Besides, the principal investigator (PI) made weekly random visits to assess the quality of data being collected in the field. The completed data collection tools were returned to the data management system at AKU Karachi Pakistan. A data quality assurance officer evaluated forms a second time to ensure that all data were collected from the participant before she handed over forms to the data entry operators who entered forms in A c c e p t e d M a n u s c r i p t standardized data entry templates. The data manager and PI reviewed the entered data and assisted in data cleaning before sending the data to the regional triangulate institute (RTI) located in North Carolina-the USA for further quality checks. A monthly monitoring report was generated and shared by RTI with the PI for making necessary corrections in the data before analysis. These monthly RTI reports helped PI to plan for refresher training whenever necessary to improve data quality.

Statistical analyses
Frequencies with proportions were calculated for users and non-users of gutka. A Chi-square test was used to compare gutka users and non-users, for sociodemographic and nutritional characteristics of the study participants. Statistical significance was set at the 5% level. To assess the association between gutka usage before the pregnancy and feto-maternal outcomes, we used Cox-regression analysis to calculate the relative risks (RR) with their 95% CIs.
We constructed five series of multivariable models to assess the effect of gutka on five outcomes including anemia, pregnancy loss, PTB, stillbirth, and LBW separately. Model Series 1 assessed the association between gutka usage and maternal anemia after controlling for mother's age, parity, mother's education, working status, cooking place, BMI, and geographic areas. Model Series 2 and 3 assessed the association between gutka usage and pregnancy loss and preterm birth, respectively.
These two models controlled for all the same confounders as Model Series 1, but additionally adjusted for maternal anemia. Model Series 4 assessed the association between gutka and stillbirth while controlling for all the variables included in Model 1, 2, and 3 and additionally adjusted for gestational age. Lastly, Model series 5 assessed the association between gutka usage and LBW after controlling for mother's age, parity, mother's education, working status, cooking place, BMI, and geographic areas. Statistical analyses were conducted using SPSS Version 20.
A c c e p t e d M a n u s c r i p t

Ethical Approval
The ethics review committees of Aga Khan University and the institutional review board of the University of Colorado Denver approved the study. All women provided written informed consent. Table 1 shows more than 50% of the women were < 25 years old and more than 80% of the women were illiterate. Regarding working status, 14% of the women and more than 90% of their husbands reported working for earning, respectively. More than a third of the women (38%) were nulliparous and 56% of the women had one to 4 children. Nearly half of the women (48%) reported cooking in a separate building and 44% reported cooking outdoor. For nutritional characteristics, nearly threefourths of the women (73%) were found to be anemic, about one-third of the women (35%) had a BMI of less than 18.5kg/m2, and around half of the women (45.4%) reported food insecurity in their houses.

Sociodemographic and nutritional characteristics of study participants
The data suggests that 1440 women (71.5%) reported using gutka in district Thatta, while 573 (28.5%) women were non-users. Gutka users were resided more in the rural areas, ranging from 68% to 80% with an average of 74% when compared with the urban areas, where the proportion of gutka users varied from 63% to 84% with an average of 68% (figure 2). We did not find any differences between users/non-users of gutka for socio-demographic variables among WRA (Table 1).

Fetao-maternal outcomes associated with gutka usage and other important factors
Around three fourth (74.2%) of the gutka users and 72.3 % of non-users were anemic (P-value: > 0.05). There was no difference between gutka users and non-users for any maternal and fetal outcomes as shown in Table 2.
The univariable analysis showed that gutka usage was not significantly associated with any of the adverse maternal and fetal outcomes (Table 2). Likewise, in the multivariable analysis, after adjusting for maternal age, education, parity, working status, BMI, geographic area, and treatment arm, we did A c c e p t e d M a n u s c r i p t not find any significant association between gutka usage with maternal and fetal outcomes such as  15-1.28)].

Discussion
The current study examined the association of gutka usage during the pre-pregnancy period with adverse maternal and fetal outcomes in district Thatta, Pakistan. The present study depicts that around three-fourths of women consume gutka in district Thatta, Pakistan and most of them live in rural areas. Gutka users were slightly more anemic and had more stillbirths and preterm deliveries, however, results were not statistically significant. After adjusting for potential confounders, we did not find a statistically significant association between gutka usage during the pre-pregnancy period and various adverse fetal and maternal outcomes.
Our findings are consistent with few studies where maternal SLT use appears to have modest to no effect on birth weight as compared to non-users. 43,44 However, our study findings contradict studies, where significant birth weight deficits are found among SLT users during pregnancy. 28,29,45 For instance, a population-based prospective cohort study conducted in Mumbai India on 1217 women found an average reduction of 105 grams in birth weight among users of SLT as compared to nonusers. 29 Likewise, in Sweden, a study conducted on 12,284 women found that adjusted mean birth weight was reduced significantly in snuff users by 39 grams as compared to non-users . 28 Furthermore, these studies found a positive and significant association between SLT and preterm and stillbirth, which differs from our study findings. 29,45 The study conducted in Sweden found that the risk of preterm delivery was high among snuff users as compared to non-users. 28 Similarly, another study conducted in India on 918 women with hemoglobin data showed a positive association of SLT with anemia among pregnant women, further contradicting our study findings. 46 Likewise, findings Downloaded from https://academic.oup.com/ntr/advance-article/doi/10.1093/ntr/ntaa215/5934513 by Aga Khan University, Health Sciences Library user on 11 December 2020 A c c e p t e d M a n u s c r i p t from a meta-analysis using three studies found a positive and significant association between SLT during pregnancy and adverse outcomes such as stillbirth, PTB, and LBW. 31 These differences across various studies could be due to several reasons that make it challenging to study the effects of SLT on pregnancy outcomes. For example, the differences could be due to variation in the products for nicotine content, bioavailability, nicotine delivery, presence of additives, toxin levels, and/or portion size of gutka. Furthermore, product type and timing of exposure can vary between women, making it challenging to identify cohorts of women with identical type, intensity, and duration of SLT exposure before and during pregnancy. Lastly, unlike our study, most of the studies described earlier have assessed the effect of SLT usage during pregnancy on the outcomes instead of assessing the effect of SLT use before pregnancy.
We also did not find any association of other important factors with the adverse fetal and maternal outcomes except for food insecurity associated with maternal anemia. This association implies that women who faced the problem of food insecurity were 1.23 likely to develop anemia when compared with their counterparts. This finding is comparable with other studies conducted in developing countries of Asia and Africa. [47][48][49] Collectively, these findings can be explained by the fact that women with food insecurity might consume inadequate diets, with a limited variety of food groups, with poor iron content and coping mechanisms during a food shortage. Besides, it is a cultural norm in Pakistan that women usually feed their husbands, children, and other family members first and consume leftover foods themselves thus resulting in anemia. Moreover, it might be possible that these women with food insecurity have low purchasing power with greater family size thus not able to purchase the nutritious food required to prevent them from becoming anemic. 50 A c c e p t e d M a n u s c r i p t

Strengths and limitations
This study has several strengths. It is the first to examine maternal and fetal outcomes among rural Pakistani women using SLT. It was conducted in an area with a high prevalence of SLT use and addresses local concerns about exposure during pregnancy to SLT. This was a population-based study with a large sample size, which measured different maternal and fetal outcomes objectively by using ultrasound (gestational age) and Hemocue machines (anemia).
This study has few limitations. First, we did not measure tobacco use during pregnancy by asking specific questions related to gutka consumption. Women might have stopped consuming gutka once they became pregnant, which is less likely because, during biweekly visits, we found women continued using gutka during their pregnancy. Since we did not collect data by asking questions about tobacco use during pregnancy, therefore, our analysis was limited to pre-pregnancy gutka usage.
Second, we also did not measure any biomarkers, such as the amount of nicotine in the urine of gutka users. However, our method of assessing exposure is consistent with other longitudinal studies that have measured exposure subjectively by relying on self-reported data. Third, we did not collect information about frequency, and duration as well as the composition of gutka, such as nicotine quantity in gutka. Fourth, stigmatization around tobacco use may discourage from disclosing their tobacco use due to social desirability bias, resulting in exposure misclassification. However, this was less likely in our study due to visible teeth staining due to gutka. Fifth, data on exposure was from women at baseline (n > 2000) but our data on feto-maternal outcomes were only on < 50% of those women. Moreover, almost three-fourths of the women in our study reported using gutka; therefore, it might be difficult to delineate the effects of gutka if the majority of the women consumed gutka. Fifth, the rates of adverse maternal and fetal outcomes are high (anemia: 75%, LBW: 17.9%, Preterm birth: 15-17%, stillbirth: 56.5 /1000 births)) in Thatta. [35][36][37] Considering the multifactorial nature of the outcome we adjusted the results of main exposure for important factors but there might be some other unknown factors to explain the likelihood of the different outcomes. To the extent that gutka tends to suppress appetite, it may also have an indirect effect of limiting dietary intake that could have impacted the outcomes with high rates of maternal undernutrition. Pakistani participants had among A c c e p t e d M a n u s c r i p t the lowest mean intakes of calories, protein and key micronutrients. 51 This is further supported by the positive impact of the nutritional intervention on birth anthropometry in this population. 33 Despite the null findings, it is important to consider that all tobacco products contain nicotine 52 . Data on the adverse effects of nicotine on the developing fetus are sufficient to classify nicotine as a developmental toxin. 53 Use of any products containing nicotine likely will have adverse effects on fetal neurological development. 54 Therefore, our study findings should be interpreted with caution.

Conclusion
Our study did not document an association between pre-pregnancy gutka usage and adverse fetal and maternal outcomes in Thatta, Pakistan. Since we did not measure the frequency and duration of gutka usage and did not assess the intake of gutka during pregnancy, findings must be interpreted cautiously. Longitudinal and robust epidemiological studies are required to detect true differences with the dose-response relationship between gutka usage before pregnancy and adverse feto-maternal outcomes. Continuing evaluation of potential associations between SLT exposure and reproductive health outcomes is necessary to develop accurate estimates of the burden of disease and public health recommendations related to gutka use. M a n u s c r i p t A c c e p t e d M a n u s c r i p t

Birth weight
A c c e p t e d M a n u s c r i p t