Social-Environmental Factors Related To Prenatal Smoking

Abstract

Introduction

Cigarette smoking during pregnancy is a significant public health issue that has profound effects on maternal and fetal health. Although many women stop smoking upon pregnancy recognition, a large number continue. Given the higher burden of smoking among low-income women, the focus of this study is to examine the impact of pre-conception social-environmental influences on smoking cessation during the first trimester of pregnancy.

Methods

Pregnant women who presented for prenatal were asked to complete a screening form at their first prenatal appointment. Women who agreed to participate were scheduled for a total of four interviews; a prenatal interview at the end of each trimester and a postnatal interview at 2 months of infant age. The sample for the current report consisted of pregnant women (first trimester) with a partner (N= 316).

Results

After controlling for pre-conception heaviness of smoking, a number of social-environmental factors were associated with smoking during the first trimester. Women were more likely to smoke during the first trimester if their partner was a smoker; however, the presence of other household smokers was not associated with increased risk for smoking. Additionally, women with a greater proportion of friends (but not relatives) who smoked and more frequent exposure to environmental tobacco were more likely to smoke.

Conclusions

This work found differential impacts of the social network on smoking suggesting that understanding relationship type, not simply number of smokers, may be important for smoking cessation efforts. Understanding differences in social network influences on smoking can help to inform interventions.

1.0 Introduction

Maternal cigarette smoking during pregnancy is a significant public health issue that has profound effects on maternal and fetal health. Although many women stop smoking upon pregnancy recognition, a large number continue to smoke. The likelihood of continued smoking during pregnancy is not equally spread across the population with younger, lower income, and less educated women more likely to smoke during pregnancy (Kaneko et al., 2008; Lu, Tong, & Oldenburg, 2001; Ockene et al., 2002). Individuals in lower income groups, compared to those in higher income groups also have higher rates of lung cancer and higher mortality rates (President's Cancer Panel, 2007). In addition to these socio-economic factors related to smoking, heaviness of smoking prior to conception is one of the strongest predictors of smoking during pregnancy (Cnattingius, Lindmark, & Meirik, 1992; Wakefield, Gillies, Graham, Madeley, & Symonds, 1993; Woodby, Windsor, Snyder, Kohler, & Diclemente, 1999).

In addition to individual-level factors such as heaviness of pre-conception smoking and income, a variety of social-environmental influences may be involved in predicting the continuation of smoking during pregnancy. For example, using a general community sample, we have found that smokers married to other smokers were less likely to quit smoking (Dollar, Homish, Leonard, & Kozlowski, 2009; Homish & Leonard, 2005). There is also evidence of partner influence on changes in smoking among pregnant smokers. For example, in a review of smoking during pregnancy, Lu and colleagues (2001) found that having a partner who smoked was associated with increased risk for continued maternal smoking during pregnancy. Although partner smoking appears to be consistent risk factor for continued smoking, it is not clear if the impact is from a partner, per se, or simply the influence of a household smoker. For example, Kahn and colleagues (2002) used a US nationally representative cohort of 8285 pregnant women and found that a greater number of household smokers was associated with a significant decrease in the likelihood of smoking cessation, replicating an earlier study by Wakefield and colleagues (1991). However, in neither of these studies was the nature of the relationship of the pregnant women and the other household members defined. Thus, it is not clear whether the important predictor is simply the presence of any smoker in the household or the particular importance of the behaviors of an intimate partner. The nature of the relationship between the smoker and household member can hold important information to inform interventions.

In addition to the impact of the immediate social network, changes in smoking during pregnancy may be influenced by the larger social-environmental context such as the influence of friends, family, and environmental tobacco exposure. Wakefield and colleagues (1993) assessed the impact of broader social network influences on smoking cessation among pregnant women. Although partner smoking status was significantly related to smoking, the proportion of smokers in the network of friends was unrelated to changes in maternal smoking. Similarly, Woodby and colleagues (1999) examined predictors of smoking cessation during pregnancy and found no influence on changes in smoking on the basis of the proportion of friends who smoked. In contrast, Etcheverry and Agnew (2008) found friends’ smoking status was predictive of smoking. In a qualitative study of factors related to smoking cessation during pregnancy, Koshy and colleagues (2010) found that the influence of friends and family on smoking during pregnancy was related to changes in smoking. Others, however, have found that support from family related to smoking cessation was not predictive of smoking cessation (Woodby, et al., 1999). Cnattingius and colleagues (1992) examined the impact of environmental tobacco exposure in the workplace on smoking risk during pregnancy. There were no differences in quit rates among pregnant smokers who worked in a room with a smoker compared to those who did not have this exposure. Further, the number of fellow workers who smoked also had little influence of persistence of smoking during pregnancy. Taken together, there are consistent findings suggesting that a partner's smoking status is influential in changes in women's smoking during pregnancy; however, the evidence is less clear with respect to the influence of other social network members. The objective of this work is to consider a broader array of social-environmental factors that may relate to changes in smoking from before conception to the first trimester.

In addition to individual-level risk factors for the continuation of smoking during pregnancy, it is clear that a variety of social-environmental factors impact changes in smoking during pregnancy. However, inasmuch as these social-environmental influences are inter-correlated, it is not clear which specific social-environmental influences are the most salient predictors of changes in smoking during pregnancy. The purpose of this study is to examine the impact of social-environmental influences on smoking cessation during pregnancy while considering the heaviness of smoking prior to conception. Given the higher burden of smoking among low-income women, this study focuses on women in a prenatal clinic at a large urban hospital. We consider a variety of factors including partner, household, friend, family and general environmental tobacco exposure conditions to determine the relative influence of each of these factors in predicting changes in smoking from before conception to first trimester.

2.0 Methods

2.1 Sample Selection

Pregnant women who presented for prenatal care at a large city hospital were asked to complete a self-report screening form at their first prenatal appointment. Women who met initial eligibility criteria (n=1671) were invited to participate in an ongoing prospective study of maternal health and child development. Initial eligibility criteria included the following: less than 20 weeks gestation, maternal age of 18 or older, no illicit drug use (other than cannabis), no heavy alcohol (more than 1 drink/day or 4 drinks on one occasion) or cannabis consumption (more than one joint/day or 4 joints on one occasion) after pregnancy recognition, and no multiple births. Of the 1671, 753 were excluded because they were a non-smoker with no match to a smoker, 502 were not recruited because they had a miscarriage, an abortion, failed to show up for the appointment, moved, could not be located, did not speak English or were not interested in participating. Women who smoked less than 5 cigarettes a day and acknowledged quitting at their first prenatal appointment on the health screen were not scheduled for a prenatal interview. Women who agreed to participate were scheduled for a total of four interviews; a prenatal interview at the end of each trimester and a postnatal interview at 2 months of infant age. At the end of each month of recruiting, the closest matching non-smokers (based on age and education) to smokers who completed the first prenatal interview were invited to participate in the study. Smokers were over-sampled so that one non-smoker was recruited for every two smokers (taking the average of age and education) who completed the first interview. Women were eligible to be recruited into the non-smoking group if they did not have a history of smoking in the past year. The current paper uses data from the end of the first trimester (between 12-20 weeks gestation) interview which assesses cigarette use for the 3 months before conception and for the first trimester. The study protocol was approved by the appropriate institutional review board. Participants were informed that data confidentiality was protected by a Federal Certificate of Confidentiality issued by the National Institute on Drug Abuse.

2.2 Participants

At the first assessment, 79% (318/402) of the women reported having a partner. This forms the basis of the sample for the current report. The majority of these relationships (60.7%) were categorized as married or living as if married. Women who were not living with their partner had been in a relationship with their partner for an average of 5.64 years (SD = 5.10 years) and reported seeing their partner for an average of 5.24 days a week (SD = 2.33 days). The average age (standard deviation) of the women was 24.2 years (SD=4.9). Less than half (41.5%) reported currently working at a paid job and the majority of these women worked less than 30 hours per week. About one third of the sample (30.5%) had less than a high school education with another 28.9% having a high school diploma or GED as their highest degree. Only 4% completed a college degree. Twenty-three percent of the women reported that they were primigravidas, with the remainder of the sample reporting an average gravidity of 3.01 (range 1 – 12, mode = 1). In terms of race/ethnicity, 44.0% reported their race/ethnicity as Black or African American, 37.1% reported their race/ethnicity as White or Caucasian and 19.8% reported that they were Hispanic or Latino. There were very small proportions of Asian, Native American, or Native Hawaiian/Pacific Islander (<4%). Participants were able to endorse more than one race/ethnicity.

2.3 Measures

2.3.1 Maternal Substance Use

Participants were interviewed in a private setting by trained interviewers. The Timeline Follow-Back Interview (TLFB; Sobell et al., 1992) was used to assess maternal substance use at each prenatal and the postnatal interview. Participants were provided a calendar and asked to identify events of personal interest (i.e., holidays, birthdays, vacations, etc.) as anchor points to aid recall. This method has been established as a reliable and valid method of obtaining longitudinal data on substance-use patterns, has good test-retest reliability, and is highly correlated with other intensive self-report measures (Brown et al., 1998). At each prenatal appointment, the TLFB was used to gather daily tobacco, alcohol, and cannabis use for the previous 3 months. Women who smoked “blunts” were asked how many joints they could have rolled from the amount of marijuana in the blunt. Thus, self-reported data spanned 3 months prior to conception through delivery. TLFB yielded a number of different measures of substance use. For the purposes of this study, TLFB data was used to ascertain maternal smoking status during pregnancy.

Saliva was collected at each prenatal interview to provide objective evidence of recent exposure. The saliva specimens were analyzed by a commercial laboratory for cotinine, the primary nicotine biomarker, with enzyme-linked immunosorbent assay (ELISA for the initial 46 women recruited into the study) or liquid chromatography-tandem mass spectrometry (LC-MSMS for the remaining women) at 10ng/ml cutoff. Maternal saliva was used to determine maternal smoking status, and was not used for identification of environmental tobacco exposure (ETS).

Maternal smoking status was determined by a combination of maternal report and maternal saliva results. Mothers were included in the smoking group if self-reports were positive, even if saliva results were negative (18% of women in the smoking group). Similarly, mothers who reported that they did not smoke but had positive saliva samples (1% of women in the smoking group) were included in the smoking group. Of the women in the smoking group, 82% had a positive saliva sample, 99% reported smoking, and 80% had both a positive saliva sample and positive self-report.

2.3.2. Social Network Smoking

Potential sources of exposure in the women's social environment consisted of the following: if women had a spouse or partner who lived in the household, if their partner had ever smoked cigarettes, if he was a current smoker, and if the partner smoked inside the home. Women were also asked if their partner smoked other forms of tobacco such as cigars or pipes, with the same follow-up questions. A dummy coded partner smoking status was determined on the basis of these questions. Other questions regarding sources of ETS exposure consisted of the number of smokers in the household excluding the partner, proportion of relatives who smoked (coded none, some, most and all), and proportion of friends who smoked (coded none, some, most and all).

2.3.3 Environmental Tobacco Exposure

Frequency of ETS exposure in the past 7 days was assessed with three questions. Women were asked about the number of days in the past week that they were in the same room, in the same car, or outside with someone who was smoking.

Data Analysis

Descriptive statistics were used to estimate smoking rates prior to conception and at the first trimester. A Generalized Linear Model (GLM) was used to examine the association between key social-environmental influences and the risk for smoking during the first trimester. The GLM model was estimated with a Binomial Family and Log link to generate risk ratios. Although the traditional regression-based approach for a binary outcome is often logistic regression, we used a GLM model because of the high rates of smoking (approximately 60% reporting smoking at the first trimester assessment). When the prevalence of an outcome variable is greater than about 10%, the odds ratio that is generated from a logistic model will overstate the true association and, therefore, models that generate relative risk ratios are preferred (McNutt, Wu, Xue, & Hafner, 2003; Zhang & Yu, 1998). Risk ratios are a measure of the association between a predictor variable and an outcome. Risk Ratios (RR) that are greater than 1 are interpreted as increasing the likelihood of an outcome whereas ratios less than 1 are interpreted as decreasing the likelihood of an outcome. For example, an RR of 1.05 for a predictor indicates a 5% increase in the criterion for each unit increase in the predictor. An RR of .90 indicates a 10% decrease in the likelihood of the outcome for each unit increase in the predictor. Risk Ratios that are equal to 1 are not significantly associated with either increased or decreased likelihood of having the outcome.

3.0 Results

Prior to pregnancy conception, 68.6% (N=218) of the women reported smoking. Given the goals of the main study to examine the impact of prenatal tobacco exposure on development, it is important to note that the study oversampled smokers. The average number (standard deviation) of cigarettes per day was 8.4 (8.7) with a range of 1 to 41 cigarettes per day. The prevalence of smoking at the first trimester did not differ significantly (first trimester 67.8% were smokers). This similarity also meant that there was very little change in smoking status between the two assessments with only 1% of women starting to smoke at the first trimester and 1.8% of women stopping smoking. This low rate of cessation may be due to the fact that we were not targeting individuals who expressed an interest in quitting. The average number of cigarettes smoked per day, however, was significantly reduced (mean[standard deviation] preconception= 8.4 [8.7]; first trimester =5.7 [6.1]; paired sample t-test, p < .001). During the first trimester, 55.0% of partners were also smokers (Table 1). The concordance of smoking between women and their partners was also very high (72% reported similar behaviors). In terms of other social-environmental influences (Table 1), 70% of the women reported that in the past week they were in a room with someone who was smoking with almost a third reporting daily exposure. The majority of women (58.8%) reported being outside with someone who was smoking in the past week and 41.5% reported being a car while someone was smoking. The prevalence of smoking among friends and family was also high; only 16% reported that none of their close relatives were smokers and only 25% reported that none of their close relatives smoke. In addition to a partner smoking, 19% of women reported another household member smoking.

Table 1.

Descriptive Statistics of Social Network Factors

Partner is Smoker

No: 55.0% (175)

Yes: 45.0% (143)

Number of Other Household Smokers

0: 81.0% (259)

1: 14.4% (44)

2+: 4.6% (15)

In past week, number of days in a room with someone who was smoking

0 days: 30.8% (98)

1-2 days: 23.0% (73)

3-4 days: 10.7% (34)

5-6 days: 4.7% (15)

7 days: 30.8% (98)

In past week, number of days in a car or truck with someone who was smoking

0 days: 58.5% (186)

1-2 days: 18.2% (58)

3-4 days: 9.8% (31)

5-6 days: 2.2% (7)

7 days: 11.3% (36)

In past week, number of days outside with someone who was smoking

0 days: 41.2% (131)

1-2 days: 16.7% (53)

3-4 days: 10.4% (33)

5-6 days: 4.1% (13)

7 days: 27.7% (88)

Not including relatives who live with you, how many close relatives smoke?

None: 16.0% (51)

Some: 41.2% (131)

Most: 27.7% (88)

All: 15.1% (68)

How many of your friends smoke?

None: 25.2% (80)

Some: 32.8% (104)

Most: 20.5% (65)

All: 21.5% (68)

A GLM model was used to examine the association between social-environmental influences and first trimester smoking (binary outcome). Several key factors were associated with an increased risk for smoking during the first trimester. After controlling for the number of cigarettes smoked per day prior to conception (Table 2), women whose partners were smokers were significantly more likely to report smoking during the first trimester (Risk Ratio [RR]=1.37, 95% Confidence Interval [CI]: 1.16-1.63; p < .001). Thus, there was a 37% increase in the risk for women's smoking in the first trimester among women who had smoking partners. Past week exposure to individuals smoking was also related to increased risk for first trimester smoking. However, this was true only for exposure to smoking in a room (RR= 1.10, 95% CI: 1.02-1.11, p < .01) or outside (RR= 1.06, 95% CI: 1.02-1.10, p < .01). Women with a greater proportion of friends who are smokers were also more likely to smoke during the first trimester (RR= 1.17, 95% CI: 1.10-1.25, p < .001); however, the number of close relatives or smokers in the household (other than the partner) was not related to increased risk for smoking at the first trimester. Each of the above factors was in the model simultaneously, thus, the impact of the other variables was accounted for in the model.

Table 2.

Preconception Social-Environmental Factors Predicting Smoking during the First Trimester of Pregnancy

Factor	Risk Ratio	95% Confidence Interval
Maternal Cigarettes Per Day Preconception	1.03***	1.02, 1.04
Partner Smoking (Yes: 1)	1.37***	1.16, 1.63
Number Of Other Household Smokers	1.07	0.97, 1.19
Number Of Relatives Who Smoke	0.98	0.92, 1.05
Number Of Friends Who Smoke	1.17***	1.10, 1.25
Past Week Number Of Days In A Room With A Smoker	1.06**	1.02, 1.11
Past Week Number Of Days Outside With A Smoker	1.06**	1.02, 1.10
Past Week Number Of Days In A Car With A Smoker	0.96	0.92, 1.01

Sample size: 316

^***p < .001

^**p< .01

Each of the above factors was in the model simultaneously, thus, the impact of the other variables was accounted for in the model.

4.0 Discussion

The goal of this paper was to investigate the impact of a variety of social-environmental factors on smoking cessation in a community sample of low income pregnant women. We found high stability in smoking from preconception assessment through the first trimester. This could, in part, be due to our sampling criteria. Light smoking women (below 5 cigarettes a day) who acknowledged quitting at their first prenatal appointment on the health screen were not scheduled for a prenatal interview. This was because the focus of the larger study was on prenatal exposure effects on child development. Most other studies that did not have this recruitment criterion have reported cessation rates of 23%-43%. However, there were significant decreases in the number of cigarettes smoked which may lead to greater rates of smoking cessation in this sample of low income pregnant women. Given the ongoing nature of this research, in future work we will be able to examine the trajectory of smoking during pregnancy on the basis of social-environmental influence.

Concordance of smoking between the pregnant women and their social network was high. For example, among those women who were smokers, 47% had a partner who also smoked. This high level of concordance between adult intimate partners is common for smoking (Graham & Braun, 1999; McBride et al., 1998) as well as alcohol (Graham & Braun, 1999; Homish & Leonard, 2007; McLeod, 1993) and other substance use (Homish, Leonard, & Cornelius, 2008). After controlling for the number of cigarettes smoke prior to conception, a number of social-environmental factors were identified as risk factors for smoking during pregnancy. Women whose partners were smokers were more likely to continue to smoke during pregnancy compared to women with nonsmoking partners. This finding is consistent with research on smokers in the general population (Dollar, et al., 2009; Etcheverry & Agnew, 2008; Homish & Leonard, 2005) as well as studies examining smoking cessation among perinatal women (Schneider, Huy, Schutz, & Diehl, 2010; Wakefield, et al., 1993). Others, however, have found that the influence of a partner on smoking cessation was more limited. For example, McBride and colleagues (1998) found cross-sectional evidence to suggest that having a partner who smoked was negatively associated with cessation in early pregnancy, but partner smoking status was unrelated to smoking cessation in later pregnancy. However, this sample consisted of women who were in a randomized trial to evaluate smoking cessation strategies for pregnant women. Therefore, it is possible that the intervention impacted the influence of partner smoking status.

Importantly, we were able to simultaneously examine the impact of partner smoking and other household members on smoking changes during pregnancy. When partner and other household members smoking status was modeled as one variable, there was a significant negative association between smokers in the household and cessation; however, when partner smoking was separated from other household smokers, the influence of partner remained significant while other household members smoking status was no longer related to women's smoking during pregnancy. Others have also found that partner smoking exerted a unique influence on an individual compared to the smoking of other social network members (Etcheverry & Agnew, 2008).

In terms of other social network influences, evidence from the current work suggested that friends, but not relatives, who were smokers increased the likelihood that the women would continue to smoke during pregnancy. The influence of the peer group may exert it is influence in terms of subjective norms (e.g., Ajzen, 1991) such that pregnant women with peer networks that are supportive of smoking resulting in a lower likelihood of successful cessation. It is not entirely clear why the influence of friends compared to family members was different. This differential influence may be related to the amount of time the women spend with friends compared to relatives. It is also possible that the nature of the relationship is important to consider. In a community sample of newly married couples, the influence of intimate partners and social network on heavy drinking and drinking and problems was examined (Leonard & Homish, 2008). After controlling for individual- and partner-level factors, the social network was related to both heavy drinking and drinking problems; however, the nature of the relationship was important to consider because the number of heavy drinkers in the network and the number of “drinking buddies” (individuals with whom activities centered on drinking) provided unique contributions to the prediction of the alcohol outcomes. Taken together, these findings suggest that the nature of relationship may be important to consider when examining risk or protection that the social network imparts.

Exposure to environmental tobacco was also longitudinally predictive of increased risk for smoking during pregnancy. A greater frequency of exposure to individuals who were smoking was associated with an increased risk for smoking. This was true, however, only for being exposed to smoking in a room with a smoker or outside with a smoker. Being in a car with a smoker was unrelated to increased risk. Given the low income nature of the sample, the prevalence of riding in a car with a smoker was lower compared to the other exposures. Thus, it is possible that individuals in the sample had less exposure to smokers in a car due to a reliance on public transportation. In addition to mainstream smoke, pregnant women exposed to environmental tobacco inhale side-stream smoke which contains higher concentrations of toxins than the mainstream smoke that is directly inhaled by the smoker (World Health Organization, 1985). This side-stream smoke readily enters the bloodstream (Gillies, Kristmundsdottir, Wilcox, & Pearson, 1986) and, consequently, may have significant physiological and neurological influences on the developing fetus (Best, et al., 2009). These findings, potentially, have important implications for policy and public health outreach reinforcing the need for pregnant women to avoid activities or locations where environmental tobacco exposure is likely. Additionally, as noted in the current findings, environmental tobacco exposure may reduce the likelihood of smoking cessation during pregnancy. Although these environmental tobacco exposure variables contributed to the risk for smoking, it is difficult to differentiate the impact of these exposure variables compared to the influence of the number of smokers within one's social network.

Future work will need to examine the mechanism of action behind each of these social-environmental factors to determine if a passive influence process (e.g., modeling) is the primary source of influence or if the influence is a more direct process which involves prompting or persuasion to impact behavior. There is clearly evidence the social network factors can be positive or negative with respect to smoking cessation (Koshy, et al., 2010), thus understanding mechanisms of action can help to aid in the developing of interventions. Because there are few studies that consider both the context of exposure and social relationships, this study adds to our knowledge about how social-environmental influences relate to smoking cessation.

There are several limitations to this study that must be considered when examining the findings. First, this is a sample of lower socio-economic pregnant women; therefore, these findings cannot be generalized to pregnant women in general. However, given the higher burden of smoking on this group, it is important to consider factors that influence changes in their smoking. Second, information about social network factors was based only on the woman's report of her social network's smoking. It is possible that women endorsed network members as smokers incorrectly. It is important to note that we did not ask women to quantify the amount of smoking by network members, but rather we simply asked the women to report on the status of the network member as a smoker vs. nonsmoker. We have focused on a narrow window of time during pregnancy (preconception to the first trimester), it is possible that social network factors become more (or less) important as women progress through the pregnancy. Finally, women who participated in the study may have been more likely to change their health behaviors to improve the health of their children. However, given the high rates of continued smoking, it appears as if this effect would have been very minimal. Despite these limitations, this work highlighted several key social-environmental factors that were associated with an increased risk for smoking during the first trimester. Importantly, we were able to more fully describe the impact of partner smoking compared to the presence of any household member and found that a partner's smoking was independently related to increased risk for smoking among pregnant women. Future work will need to examine if there are changes in smoking throughout the pregnancy on the basis of social network factors. Additionally, future work should attempt to describe the mechanisms of how the social network impacts changes in substance use can help to inform intervention efforts.

Highlights.

• This study examined pre-conception social-environmental influences on smoking cessation during pregnancy

• Pregnant women (n=316) presenting for prenatal care completed a screening form

• Women were more likely to smoke during the first trimester if their partner was a smoker

• However, the presence of other household smokers was not associated with increased risk for smoking

• This suggests that understanding relationship type, not simply number of smokers, may be important