Tobacco consumption and spontaneous quitting at the first trimester of pregnancy

INTRODUCTION
The purpose of this study was to examine the association between pregnant women's socio-demographic characteristics, smoking-related variables and psychological symptoms (anxiety and depression) and both tobacco consumption and spontaneous quitting at the first trimester of pregnancy. In particular, we wished to examine the contribution of depressive symptoms to tobacco consumption and spontaneous quitting, while controlling for anxiety symptoms, socio-demographic and smoking-related variables.


METHODS
The sample was comprised of 901 Spanish pregnant women. Assessment included an ad hoc questionnaire with socio-demographic and tobacco consumption information, the Edinburg Postnatal Depression Scale (EPDS), and The State-Anxiety Inventory (STAI-S). Two multiple logistic regression analyses were performed, respectively to predict tobacco consumption and to predict spontaneous quitting.


RESULTS
Having a partner who smokes (OR=5.578), not having a college education (OR=2.803), higher scores on the EPDS (OR=1.073) and higher scores on the STAI-S (OR=1.027) increase the probability of continuing smoking. Being primiparous (OR=2.463), having a college education (OR=2.141), smoking fewer cigarettes before pregnancy (OR=1.175), and lower scores on the STAI-S (OR=1.045) increase the probability of spontaneously quitting smoking at the first trimester of pregnancy.


CONCLUSIONS
Depressive symptoms were a predictor of tobacco consumption but not of spontaneous quitting; spontaneous quitting was better predicted by anxiety symptoms. These findings support recommendations that women with depressive symptoms are at risk for smoking during pregnancy and highlight that anxious symptoms should be targeted in interventions for smoking cessation during pregnancy.


Introduction
Tobacco is one of the most commonly used drugs, and cigarette smoking during pregnancy has serious health consequences for both the pregnant woman and her offspring. In fact, several medical and psychological problems were reported in the offspring from gestation (Stroud et al., 2009;USDHHS, 2014;Vardavas et al., 2010) to several years after childbirth (Wakschlag et al., 2011), and were associated with the mother tobacco consumption during pregnancy. Despite policies implemented to reduce smoking and efforts to increase awareness of the detrimental effects of smoking during pregnancy, the prevalence of tobacco consumption namely in Europe is still quite high during pregnancy (Smedberg, Lupattelli, Mardby, & Nordeng, 2014;Smedberg, Lupattelli, Mardby, Overland, & Nordeng, 2015).
The body of research on the role of anxiety in tobacco consumption during pregnancy is quite sparse, when compared to the depression. Taking in account the high comorbidity between depression and anxiety symptomatology during pregnancy (Figueiredo & Conde, 2011;Vesga-López et al., 2008), questioning how much of the effect of depression on tobacco consumption can be due to the effect of co-morbid anxiety, when anxiety is not controlled, is relevant.
Quitting smoking during pregnancy was also associated with both the mother´s sociodemographic characteristics and smoking-related variables. Among the factors that have been reported to increase the chances of spontaneous quitting during pregnancy, the most cited are primiparity (Krstev et al., 2012;Torrent et al., 2004), having a planned pregnancy (Fang et al., 2004), a higher educational level Krstev et al., 2012;Vaz et al., 2014), a non-smoking partner (Torrent et al., 2004), and lower pre-pregnancy use and/or a lower level A C C E P T E D M A N U S C R I P T

ACCEPTED MANUSCRIPT
6 of nicotine dependence Morasco, Dornelas, Fischer, Oncken, & Lando, 2006;Torrent et al., 2004;Vaz et al., 2014). Other studies also linked the decision to quit smoking during pregnancy with the awareness of the risks of smoking on the fetus, on health and with the guilt resulting from smoking (Fang et al., 2004;Smerdberg, et al., 2015).
Nevertheless, less attention has been paid to the relationship between quitting smoking during pregnancy and the mother´s psychological symptoms. Prior research indicates that women with high levels of depressive symptoms present a decreased likelihood of quitting smoking during pregnancy (Smerdberg, et al., 2015) and tend to be less confident in their ability to quit smoking (Hauge, Torgensen, & Vollrath, 2011;Prusakowski, Shofer, Rhodes, & Mills, 2011).
Moreover, although psychological symptoms during pregnancy are more prevalent among smokers, and less prevalent among spontaneous quitters (Solomon et al., 2006), no previous studies have differentiated the effects of anxiety and depression symptoms on smoking behavior during pregnancy.
The aim of the present study was, firstly, to analyze the association between Spanish pregnant women's socio-demographic characteristics, smoking-related variables and psychological symptoms (anxiety and depression) with both tobacco consumption and spontaneous quitting in the first trimester of pregnancy. And secondly, to examine how much depressive symptoms contribute to tobacco consumption and spontaneous quitting, after controlling for socio-demographics characteristics, smoking-related variables and anxiety symptoms. The research questions addressed were: Are the characteristics that differentiate women who continue smoking from those who spontaneously quit during pregnancy in Spain similar to the ones found in other countries? Do depressive symptoms contribute to tobacco consumption and spontaneous quitting during pregnancy?
The questions asked in the present study regarding the characteristics that differentiate women who continue smoking from those who spontaneously quit in pregnancy is a relevant one, in order to provide a more adequate intervention to those women at risk of not quitting smoking during pregnancy.

Participants
Pregnant women attending a primary health public care center in two health areas of Galicia (Northern Spain) were consecutively recruited, until the sample size to insure representativeness was reach 1 . Inclusion criteria were: having 18 years or more, less than 20 weeks pregnant and speaking and reading Spanish. A total of 915 women were invited to enroll in the study, 11 pregnant women declined the participation, and three could not be enrolled, as they were not proficient in Spanish. The study sample was comprised of 901 pregnant women at the first trimester of pregnancy, with an average of 10.02 weeks of gestation (SD = 4.27).
The age range of the sample was between 18 to 46 years, with a mean age of 32.73 years (SD = 4.36). More than half of the participants were over 30 years old (72.3%), were employed at the time the survey was carried out (71.0%), were primiparous (66.3%), had university studies (53.2%) and earned a maximum 2000 euros net per month and household (63.5%). Likewise, the majority of the participants were married or cohabiting (83.7%).

Procedure
This study received approval from the General Directorate of Research, Teaching and Research Ethics Committee of Clinical Research of Galicia. It took place between December 2012 and February 2014. The sample was recruited in the first trimester of pregnancy, while the participants were waiting for their first or second midwife consultation at their local health 1 A mathematical formula, which took into account the number of births in Galicia in the year preceding the study according to official data, with a confidence level of 95 % and an accuracy of 3.4%,was used to calculate the sample size to insure representativeness Participants were firstly informed about the objectives of the study and their cooperation was requested.
Once informed consent was signed, two psychologists especially trained for that purpose carried out the interviews individually. The time for individual assessment of each participant was between 15 and 20 minutes on average. No pregnant women withdraw her participation after signing the informed consent and 98.5% of the pregnant women completed all the study measures. Self-reports of smoking abstinence were biochemically verified through cotinine concentration in urine.

Socio-demographic and consumption information
A questionnaire regarding socio-demographic characteristics (e.g., age, marital and employment status, number of children, education, occupation and household income), psychological (e.g., participant´s psychological diagnose) and pregnancy-related variables (e.g., parity, number of gestational weeks and planned pregnancy) was developed for this study. Smoking status was assessed by the following questions: "Have you ever smoked cigarettes regularly?" "Did you smoke before pregnancy?" (if yes, "How many cigarettes did you smoke each day on average?" if no, "Since how many months before pregnancy?") "Have you smoked any cigarettes in the past seven days?" (if yes, "How many cigarettes did you smoke each day on average?" if no, "Since when did you stopped smoking and why?").
Women who reported smoking no cigarettes in the past seven days, but had smoked at least one cigarette per day in the three months prior to learning of their pregnancy, and who reported quit when they suspected /confirmed they were pregnant were classified as "spontaneous quitters". Other variables related to pre-pregnancy smoking included the number of daily cigarettes smoked and the number of years consuming tobacco. We also asked about the perception of how tobacco consumption affects health by the following question: Do you believe that smoking seriously affects health?, and partner´s smoking status.
Self-reports of smoking abstinence were biochemically verified through urinary cotinine by Cotinine Test Medi-marketing (cut-off 200 ng/mL). This biochemical measure was selected instead of carbon monoxide as the cotinine remains for a longer period of time in the human body than the later. The use of cotinine was particularly helpful as the pregnant women were assessed in the morning and previously to a medical appointment, so it was expected that they would not have smoked. Cotinine measurement and self-report were combined to establish smoking status. Pregnant women who reported not smoking performed a cotinine urine test and were classified as smokers when the result was positive.

Edinburg Postnatal Depression Scale (EPDS)
The EPDS (Cox, Holden, & Sagovsky, 1987) is a self-reported questionnaire designed to detect postpartum depressive states. This scale includes 10 items with four response options, each having a single value between 0 and 3, the higher the value the more severe the symptom. As a consequence, the scale scores range from 0 to 30. This is the most internationally used questionnaire for detection of postpartum depressive symptoms, but it is also used for assessments during pregnancy (Figueiredo & Conde, 2011). The cut-off score of the scale is country dependent, ranging between 9 and 13. The EPDS Spanish version of the scale (García-Esteve, Ascaso, Ojuel, & Navarro, 2003) proposed a cut-off of 10 / 11; so scores of 10 or more are indicative of probable depression. In the present study, the EPDS showed good internal consistency (Cronbach's α = .85).

State Anxiety Scale of State-Trait Anxiety Inventory (STAI)
The State Anxiety Scale (S-Anxiety) of STAI (Spielberger, Gorsuch, & Lushene, 1970;Spanish version TEA, 1982) evaluates the current state of anxiety, asking how respondents feel "right now". It consists of 20 items, whose responses are coded using a

ACCEPTED MANUSCRIPT
10 Likert scale of 4 points. The score for subscale ranges from 0 to 60, higher scores indicating higher levels of anxiety. In the Spanish adaptation of the questionnaire a score of 31 has been suggested as a clinical cutoff point for the female population (Sarasua, Zubizarreta, Echeburúa, & Corral, 2007). In this sample, the internal consistency of the STAI-S was excellent (Cronbach's α = .94).

Data Analysis
Data were analyzed with SPSS Statistics version 20. The significance level was at p < .05. To test differences between smokers and all non-smokers (never smokers and quitters) and smokers and quitters, chi-square test for discrete variables and Student's t-tests for continuous variables were respectively performed. Cramers´s V coefficients and Cohen's d were calculated to estimate the effect size 2 .
To predict tobacco consumption and quitting smoking at the first trimester of pregnancy, two forward binary logistic regressions were performed, respectively on two different samples. First, to predict current tobacco consumption, the total sample (N= 901) was used, using tobacco consumption (Yes vs. No) as dependent variable (VD). The subsamples utilized for the analysis were smokers and non-smokers (quitters and never smokers).
Next, to predict spontaneous quitting the sub-sample of 268 smokers before pregnancy was used, and the VD was tobacco abstinence (Yes vs. No). In both models, socio-demographic, health and psychological variables (age, marital and occupational status, parity, educational level, income, planned pregnancy, smoker partner and STAI-S) were used as predictors in the first step and the score of the EPDS was used as predictor in the second step.

Socio-demographic Characteristics and Smoking Status
Tobacco consumption in the first trimester of pregnancy was significantly associated with all the studied socio-demographics variables, with the only exception for age and parity (Table 1). Smoking was associated with not cohabiting (p < .001), being unemployed (p = .009), lower educational level (p < .001) and lower monthly income (p = .006).
Spontaneously quitting smoking in the first trimester of pregnancy was associated with being primiparous (p = .018), higher educational level (p < .001) and higher monthly income (p = .007).

Health and Psychological Factors and Smoking Status
Furthermore, to analyze the relationship between smoking status with different health and psychological variables (Table 2), it was found that tobacco consumption in the first trimester of pregnancy was significantly associated with all the studied variables. Smoking was associated with not having planned the pregnancy (p = .036), living with a partner who smokes (p < .001), lower health risk perceptions associated to smoking (p < .001), had been

Predictors of Smoking Status
The results of the logistic regression analyses (Table 3) showed that not having a college education (OR = 2.803), having a smoking partner (OR = 5.578), higher scores on the STAI-S (OR = 1.027) and higher scores on the EPDS (OR = 1.073) determine a higher  Table 3 ______________________________________

Discussion
In this sample of Spanish pregnant women, the prevalence of smoking in the first trimester was high (17.4%) and similar to rates observed in other studies (Maxson et al., 2012;Meghea et al., 2012;Smedberg et al., 2015). The pregnant smoker profile found in the present study corroborate that smokers tend not to cohabit with or have a partner (Dupraz, et al., 2013;Goedhart et al., 2009;Maxson et al., 2012), are more likely to have a lower household income or lower socioeconomic status (Krstev et al., 2012;Maxson et al., 2012) and a lower educational level (De Wilde et al., 2013;Goedhart et al., 2009;Krstev et al., 2012;Linares et al., 2009;Maxson et al., 2012;Smedberg et al., 2015). Additionally, and consistent with prior research (De Wilde et al., 2013;Dupraz et al, 2013), pregnant smokers are more likely to have a partner who smokes, which indicates that household smoking is a significant risk factor for smoking, an assumption that is valid for the general smoking population. However, findings concerning age and parity in the current study do not match all results from the literature since

ACCEPTED MANUSCRIPT
14 no differences were found between smokers and non-smokers in age or in parity.
Our data also showed that a high percentage of women quit smoking on their own initiative once their pregnancy is confirmed. One hundred eleven (41.4%) out of the 268 women who reported smoking before pregnancy spontaneously quit smoking in the first of pregnancy, which coincides with data reported by Schneider et al. (2010)

of 27-47% quitters in European studies
The results in the present study seem to corroborate the fact that pregnancy motivates women to quit tobacco consumption (Hepper et al., 2011) because about half of smokers in the sample spontaneously quit smoking after they found out they were pregnant. However, these optimistic data must be analyzed carefully because the literature also notes that smoking relapse in the postpartum period occurs at high rates (Su & Buttenheim, 2014).
Likewise, differential characteristics were observed in women who spontaneously quit smoking once their pregnancy was confirmed. Those who quit were more likely to be primiparous, have a higher educational level, have higher socioeconomic status and have a non-smoking partner, have a lower number of years of tobacco consumption, smoke a lower number of cigarettes daily before pregnancy, and have lower anxiety and depressive symptoms. These are characteristics consistent with the literature. For instance, Krstev et al. (2012) noted that women who are primiparous and have a higher educational level and socioeconomic status are more likely to spontaneously quit smoking. Likewise Torrent et al. (2004) pointed out that being primiparous and having a non-smoking partner were factors

ACCEPTED MANUSCRIPT
15 related to smoking cessation during pregnancy. Additionally, Smedberg et al., (2015) reported that 32.5% of the women that continued smoking during pregnancy were depressed.
Having a lower educational level, having a partner who smokes and higher scores on anxiety and depression (even when controlling for anxiety symptoms) were predictors of tobacco consumption. Being primiparous, having a college education, consuming fewer cigarettes before pregnancy and having lower scores for anxiety, but not for depression, were factors that predicted smoking abstinence. Depressive symptoms seem to be a major predictor of tobacco consumption but not of quitting smoking in the first trimester of pregnancy, when anxiety symptoms are considered. Nevertheless, anxiety predicted both consumption and spontaneous cessation. One possible explanation for the present findings suggests that smokers use cigarettes to regulate negative affect and other mood states (self-medication hypothesis), and that it is precisely these symptoms that maintain tobacco consumption during pregnancy (Bullock et al., 2001;Ludman et al., 2000;Munafò et al., 2008). The belief that smoking is an effective method of coping with negative affect and potential stressors during pregnancy (e.g., health concerns) provides women who experience emotional difficulties the motivation to continue smoking despite social pressure and awareness of the health risks.
Some studies note that women with higher depressive symptoms continue smoking and present a decreased likelihood of quitting (Hauge et al., 2011;Smedberg et al., 2015).
Anxious symptoms were not considered in these studies and depression symptoms can bias the effect of anxious symptoms due to high comorbidity between depression and anxiety symptoms, especially in women during pregnancy (Figueiredo & Conde, 2011;Vesga-López et al., 2008). Nevertheless, the present study examined how much women's depressive symptoms contribute to tobacco consumption and spontaneous quitting after controlling for anxiety symptoms. As a consequence, we consider it essential to analyze depression alongside with anxiety while studying tobacco consumption and spontaneous quitting during pregnancy.

ACCEPTED MANUSCRIPT
16 Limitations in the present study indicate that our results should be interpreted with some caution. This study was cross-sectional, which means it is difficult to establish a causal relationship between the different variables under analysis. Furthermore, no pre-pregnancy psychological assessments were carried out, which does not allow us to know how prepregnancy status affected the process of quitting due to anxious and depressive symptoms.
Depression and anxiety symptoms were self-reported in questionnaires and subject to social desirability bias.
Despite these limitations, this study has several strengths. This includes testing simultaneously and in a large sample all the relevant factors (namely psychological factors) that previous research has found to be related with smoking in pregnancy, and adding a confounding factor, anxiety.The most important strength of this study is that it considers the effects of both depression and anxiety symptomatology on tobacco consumption during pregnancy, testing if previous reported effects of depression on tobacco consumption may result from the effect of comorbid anxiety. Another strength is that previous studies (e.g., Orr, Blazer & Orr, 2012) compared women who smoked during pregnancy with women who did not (thereby combining pregnancy quitters with non-smokers). Others (e.g., Linares-Scott et al., 2009;Massey, & Compton, 2013) compared those who smoked during pregnancy with quitters (only women who quit smoking during pregnancy). This study compared smokers with both non-smokers and quitters, because these are groups with different characteristics, and to our knowledge provides new data on the association between depression and smoking or quitting with adjustment for potential confounders. Moreover, the pregnant women´s selfreports of smoking abstinence were biochemically validated to control for social desirability bias. This is particularly relevant during pregnancy when higher social pressure for women not smoking is present.

17
In summary, smoking during pregnancy is a significant public health issue that endangers both the mother's and fetus's current and future health. It is therefore vital to achieve abstinence from tobacco in this period of life, as soon as the first trimester of pregnancy. In fact, major adverse effects of tobacco consumption in the fetus occur as early as from the first trimester of gestation (USDHHS, 2014;Vardavas et al., 2010). Most pregnant women seem to be aware of these health risks, but awareness alone is not sufficient to prompt all of them to quit smoking. On the other hand, most women who stop smoking spontaneously do so in the first trimester of their pregnancy (Fingerhut, Kleimman, & Kendrick, 1990). In our sample, 87.2% of the participants self-reported tobacco abstinence around their first month of pregnancy. This indicates that early pregnancy is the appropriate time to develop and improve interventions for smoking cessation because of the emergence of new motivational factors that have the potential to support behavior change.
Depressive symptoms were a predictor of tobacco consumption but not of spontaneous quitting; spontaneous quitting was better predicted by anxiety symptoms. These findings support recommendations that women with depressive symptoms are at risk for smoking during pregnancy and highlight that anxious symptoms should be targeted in interventions for smoking cessation during pregnancy. The main clinical implications of our results concern the relevance of certain factors, mainly women´s anxiety levels and the presence of a smoking partner, in the screening of women at risk of continued smoking during pregnancy. Also, reinforce the idea that prenatal smoking cessation interventions must take into account their emotional factors, and get smoking partners involved in the treatment. Further research is needed to understand the complex relationship between pregnancy and smoking across different cultures and countries.