Social and Environmental Factors Related to Smoking Cessation among Mothers: Findings from the Geographic Research on Welling (Grow) Study

Abstract

Objectives

The current study examined associations between race/ethnicity and psychosocial/environmental factors with current smoking status, and whether psychosocial/environmental factors accounted for racial differences in smoking status in a population-based sample of mothers in California.

Methods

Cross-sectional data from 542 women with a history of smoking were used. Analyses adjusted for age, partner status, and educational attainment.

Results

In models adjusted for sociodemographics, black women had significantly lower odds, and Latina immigrants had significantly higher odds of being a former smoker compared to white women. Persons smoking in the home, having a majority of friends who smoke, having perceptions of their neighborhood as being somewhat or very unsafe, and experiencing food insecurity were associated with decreased odds of being a former smoker. When these variables were entered into a single model, only being a Latina immigrant and having a majority of friends who smoke were significantly associated with smoking status.

Conclusions

Black women demonstrated a notable disparity compared with white women in smoking status, accounted for by psychosocial/environmental factors. Immigrant Latinas demonstrated notable success in ever quitting smoking. Social networks may be important barriers to smoking cessation among women.

Although the smoking prevalence rate among women in the United States (US) is lower than that of men (15.3% vs 20.5%),¹ women often have less success during a quit attempt and demonstrate lower quit rates compared to men.^2–8 Racial/ethnic disparities exist in smoking cessation⁹ but this disparity is rarely examined within each sex.¹⁰ Further, it is well accepted that racial/ethnic disparities are not inherent, but rather, are a function of inequities in social structure and context.¹¹ Experts have called for cessation interventions to take into account the unique needs of women who smoke^12,13 and for research that identifies culturally-relevant or sex-relevant motivators and determinants of cessation.¹⁴ Thus, research is needed to identify factors that account for racial/ethnic disparities in smoking cessation within women.

Research has identified numerous psychosocial and environmental factors associated with smoking behavior,^15–23 but little is known about the role of race/ethnicity in the process of smoking cessation among women. In addition, little is known about whether these psychosocial and environmental factors are uniquely associated with ever having quit among women. Finally, extant research on psychosocial and environmental correlates of smoking behavior that has focused on women has largely utilized samples of women who are pregnant or in the peripartum period.^24–32 Some research suggests that the process of smoking cessation, in particular, may be different in pregnant smokers versus non-pregnant smokers.³⁰ Therefore, the purpose of the current study was to characterize the patterns and correlates of smoking status in a diverse sample of women with a lifetime history of smoking. Such information is useful for identifying populations of women at particular risk for failed cessation and for identifying potential targets of treatment to inform intervention development. This study is unique in that no known published study demonstrates all 3 design characteristics of: (1) an exclusive focus on women in the examination of racial/ethnic disparities in smoking status and correlates of being a former smoker; (2) an examination of factors that account for racial/ethnic disparities among women, and; (3) a sample that is not limited to women in the peripartum period.

Environmental and Psychosocial Factors Associated with Smoking Behavior

Various environmental and psychosocial factors are determinants of smoking behaviors. Low neighborhood level socioeconomic status (SES) is associated with smoking even after adjusting for individual level SES.^33–38 Neighborhood poverty¹⁶ and education and employment³⁴ have been associated with smoking among black women. Neighborhood deprivation has been associated with cessation among white, black and Latino smokers engaged in a quit attempt,³⁹ as has neighborhood level unemployment and level of debt among black smokers participating in a cessation trial.⁴⁰ Neither cessation study further examined associations by sex.

Neighborhood crime and disorder and concerns about neighborhood safety also have been associated with smoking behavior.^26,38,41,42 Burdette et al⁴¹ found that neighborhood safety was associated with smoking among women with young children. Tseng et al³⁸ found that living in a high crime area was associated with current smoking among female ever smokers. Patterson et al²⁶ found neighborhood violence was associated with smoking during pregnancy. These studies did not examine racial disparities. Research has demonstrated a consistent link between discrimination and smoking behavior,^24,43–51 but only one published study found that experiencing a major discriminatory event was negatively associated with cessation among Mexican-American smokers engaged in a quit attempt.⁵² This study controlled for sex rather than examine its effects.

Smoking is more prevalent among individuals living below the poverty level compared with those living above the poverty level.^1,53 Income has been associated with smoking among women⁵⁴ and among black women, specifically.²³ However, income has not shown a consistent association with smoking among Latinos or Asian Americans.^45,55 Less is known about how income is associated with quitting. Income has been associated with cessation in the general population⁵⁶ but some research suggests it is not a predictor of cessation among Latinos attempting to quit.^57–59 Also, it is unclear whether changes in income impact smoking behavior. Young-Hoon et al⁶⁰ examined the association between one’s change in income over time and status as a former smoker. Increases in income were associated with greater odds of being a former smoker; decreases in income were associated with lower odds of being a former smoker only when the drop in income fell below the poverty level. However, Blakely et al⁶¹ found that income change was not associated with smoking status among adults over age 25. None of these studies examined associations among women separately or by race/ethnicity.

Being unemployed has shown a consistent association with being a smoker,^62–65 although not among Mexican Americans.⁵⁵ However, differences by sex and race/ethnicity are rarely examined. The relationship between employment status and cessation is not clear. Studies have found that unemployment is related to lower odds of quitting among women,⁶⁶ higher odds of quitting among women,³ and no association between income and cessation.⁶⁷ Employment status has not been associated with quitting among Latinos.⁵⁸ Each of these studies controlled for but did not examine sex and/or race/ethnicity.

One’s immediate social circle plays a role in smoking behavior. Having children has shown an inconsistent association with both smoking and cessation,^59,64,65 but this may be accounted for by the moderating effects of age at motherhood, race/ethnicity, and partner status found in other studies.^68–71 However, to our knowledge, no published research has examined whether number of children is associated with smoking or cessation among mothers. Having friends who smoke is associated with smoking^21,72–74 and failure to quit smoking.^{19,20,24–29} Oh et al²¹ found that having friends or family who smoke was significantly associated with ever smoking among European women, and Solomon et al²⁹ found it to be associated with postpartum relapse. People smoking in one’s home is associated with smoking^19,55 and home smoking bans have shown a consistent association with smoking less and greater odds of cessation.^22,75–78 One study has examined the association of a household smoking ban with smoking among women and found greater odds of cessation.²² Social support has been associated with odds of cessation^{2,32,39,79,80} and odds of being a former smoker.⁸¹ However, among the few studies that focus on women, the results are mixed. Creswell et al⁵⁷ found that support was related to cessation among women with weight concerns seeking cessation counseling, as did Ward et al³² among pregnant women. However, Pollak and Mullen²⁷ did not find social support to be associated with relapse among women who quit smoking due to pregnancy. Turner et al⁸² found that, whereas the main effect of social support was not statistically significant, social support buffered the negative impact of depression on cessation among women seeking cessation counseling.

Related to individual and household stressors, food insecurity, or the inability to access adequate food due to limited resources, has been shown to be positively related to smoking behavior,^83–87 including among women specifically.^17,18 However, extant research has not examined its relationship with smoking status. Perceived stress is among the most consistent predictors of smoking^{15,28,31,88–93} and cessation.^{20,25,28,39,93–97} Among women in particular, perceived stress has predicted time to relapse,⁹⁶ smoking prevalence and failure to quit among peripartum women,^25,28,31 and women not in the peripartum period.^20,89 In contrast, few studies have failed to find an association between perceived stress and smoking behavior among women.²⁷

In summary, extant research has demonstrated links between psychosocial and environmental variables and smoking, but in most instances, little or no published research has examined their links with cessation, be it either current smoking status or success in a specific quit attempt. In addition, research with population-based, racially/ethnically diverse samples of women is lacking, but especially research on samples of women not in the peripartum period. Identifying populations of women who are at particular risk for failed cessation and illuminating potential targets of intervention is needed to inform the development of effective interventions. Factors at multiple levels (community, household, individual) are influential in women’s lives and likely serve as barriers or enablers of transitioning from being a current to a former smoker. As such, the current study examined racial/ethnic disparities in smoking status in a population-based sample of women in California with young children, focusing on the known psychosocial and environmental determinants of smoking.

METHODS

Data

The Geographic Research on Wellbeing (GROW) study (2012–2013) is a follow-up survey of women who participated in California’s Maternal and Infant Health Assessment (MIHA) survey during 2003–2007.⁹⁸ MIHA, sponsored by the California Department of Public Health, is an annual statewide, representative survey of mothers delivering live infants in California during February-May, linked with birth certificate data. Women are eligible for MIHA if they are English-speaking or Spanish-speaking California residents, aged 15 years or older, with singleton, twin, or triplet births, and whose addresses are recorded in birth certificates. The sample is selected according to region, education, and race/ethnicity, oversampling black women. During 2003–2007, MIHA surveyed approximately 3500 women who were representative of the approximately 500,000 women who gave birth to a live infant in California that year. Self-administered surveys in English and Spanish are mailed to women starting about 8 weeks after they give birth. Telephone contact is attempted with non-responders. Questionnaires were completed by mail for 69% and by telephone for 31% of respondents; 71% of the surveys were completed in English and 29% in Spanish. Response rates exceeded 70% in each year of MIHA. Additional details about MIHA have been described previously^99–102 and further information is available at http://www.cdph.ca.gov/data/surveys/MIHA/Documents/MIHA%20Technical%20Document%20Web%202011.pdf.

MIHA respondents from 6 largely urbanized counties with the highest number of respondents (Alameda, Los Angeles, Orange, Sacramento, San Diego, and Santa Clara) and who had agreed to be re-contacted for a future study (95% of all MIHA respondents) were eligible for GROW (N = 9256, or 51% of all MIHA respondents). In February 2012, letters were mailed to notify women that they would be receiving a survey in the mail and to ascertain undeliverable addresses before the actual mailing. Several weeks later, the questionnaire packets were mailed in the language (English or Spanish) in which women completed the MIHA survey, and a reminder postcard was mailed several weeks later. Address tracing and re-mailing were performed continuously. Additional contact information for non-respondents was obtained through an approved data linkage request with the Women, Infants, and Children (WIC) Supplemental Nutrition program. Telephone surveys for those who did not return paper surveys were attempted between May 2012 and May, 2013. In March, 2013, a postcard was mailed notifying women of a doubling of the gift card incentive (from $10 to $20) and an additional raffle to win an iPad, iPod touch, or a children’s bicycle. Data collection for GROW occurred during 2012–2013 when the index children from MIHA were 5–10 years old.

The GROW survey was intended to be as similar as possible to MIHA in format, language, and survey administration. The questionnaire included approximately 80 questions regarding demographic, socioeconomic, neighborhood-related, psychosocial, and health-related characteristics, pertaining to themselves and their index child (their infant from the MIHA survey). The survey development process lasted 14 months, beginning with an outline of the measurement domains and a search of existing instruments. Multiple drafts of the instrument were reviewed by the project investigators to refine the survey. Six focus groups were then staggered over time to allow for refinement of the instrument and translation into Spanish. Finally, a convenience sample of 7 women with children was used to pilot test the final mail version for timing and logic and 6 women selected as a convenience sample were interviewed by phone via Computer-Assisted Telephone Interviewing (CATI) to pilot test the phone version. Home addresses were geocoded to the census tract level. Among the eligible respondents who were located (N = 4026), the response rate was 75% (N = 3016); among all eligible respondents (N = 9256), it was 33%. Comparing the weighted GROW sample (N = 3016) to the weighted MIHA sample and the target population of all women in California who gave birth during the relevant time period, GROW is representative in terms of important sociodemographic characteristics. The analytic sample for the current study includes all women in GROW who reported that they ever smoked 100 cigarettes in their lifetime and self-reported their race/ethnicity as non-Hispanic Asian/Pacific Islander, non-Hispanic black, Latina, or non-Hispanic white (N = 542).

Measures

The dependent variable was current smoking status. A woman was defined as a former smoker if in the GROW survey she reported ever smoking 100 cigarettes in her lifetime, but not current cigarette smoking. Sociodemographic control variables included race/ethnicity (defined above), educational attainment (less than high school, high school or GED, some college, college graduate), age (in years), and partner status (single/separated/ divorced/widowed, married/cohabiting).

There were several independent psychosocial and environmental variables. Number of children in the household was based on the total number of children or grandchildren at any age whom the mother had living with her. Employment status was based on how many hours per week the mother usually worked at one or more paid jobs during the past 2 weeks (did not work at a paid job, part-time [less than 40 hours], full time [at least 40 hours]). Based on our prior work,¹⁰³ neighborhood poverty was based on a latent class growth modeling analysis (LGCM)⁸⁵ of poverty rates for all census tracts in California, using data from the Neighborhood Change Database, 1970–2000 (published by Geolytics, Inc.), and the American Community Survey, 2005–2009. We estimated 3 distinct latent classes, referred to as long-term low poverty, long-term moderate poverty, and long-term high poverty, which are linked to GROW data via geocodes. Neighborhood safety was defined from responses to the question on how safe respondents felt their neighborhood was from crime (somewhat or very unsafe, somewhat safe, very safe). Income was measured by self-report of the previous calendar year’s income from all sources combined with family size and categorized into percent of the federal poverty level (< = 200%, 201%-400%, > 400%). Change in income was based on comparing the respondent’s income at MIHA to her income in GROW (income as a percentage of the poverty level dropped between MIHA and GROW; income did not drop). Friends who smoke was based on the statement: “Most of my friends smoke cigarettes” (yes, no). Smoking in the home, modified from the National Health Interview Survey-Cancer Control Supplement (2005) was based on responses to how often people usually smoked anywhere inside the respondent’s home during the previous year (never, sometimes/most days/daily). Worry about discrimination, modified from the MIHA 2011 survey, was defined in response to the question: “Overall during your life until now, how often have you worried that you might be treated or judged unfairly because of your race or ethnic group?” Response options included “very/somewhat often, not very often/never.” We used a 6-item food insecurity scale developed by the National Center for Health Statistics and Abt Associates Inc. that includes questions referencing the last 12 months and querying mothers on issues such as: “The food I bought just didn’t last, and I didn’t have money to get more,” “I couldn’t afford to eat balanced meals,” and “Did you ever cut the size of your meals or skip meals because there wasn’t enough money for food.”⁸⁶ Women who answered affirmatively to at least 2 of the items were considered food insecure (versus food secure). Perceived stress was defined in response to the question: “How often did you feel that you had more to do than you could comfortably handle” during the past year (very/somewhat often, not very often/ never).¹⁰⁴ Finally, social support measures, modified from MIHA 2011 and the National Health and Nutrition Examination Survey, were divided into 3 types (yes, no), based on the questions: “Do you have someone you turn to if you needed…someone to comfort or listen to you,” (emotional support), “…some extra help financially, like help paying for some bills, the rent or mortgage, or food that you needed” (financial support), or “…practical help, like getting a ride somewhere, help with shopping or cooking a meal, or help watching your children for a short time” (practical support).

Data Analysis

Data regarding smoking status were not comparable between the MIHA and GROW surveys; therefore, we could not examine change in smoking status prospectively. Thus, we examined cross-sectional associations of sociodemographic and environmental variables with smoking status among the 542 women in the GROW sample who reported ever smoking. Logistic regression models were constructed with smoking status as the dependent variable (odds of being a former smoker). First, we examined a sociodemographic model, adjusting for race/ethnicity, age, partner status, and educational attainment. Then, we added each psychosocial and environmental variable to that sociodemographic model to determine whether there was an independent effect of each. Finally, all significant psychosocial and environmental variables were added simultaneously to a final model. All analyses were weighted for non-response and accounted for the complex sample design. Because the data were not highly clustered by census tract (mean of 1.4 respondents per tract; 90% of tracts contained only one or 2 GROW respondents), multilevel modeling procedures were not warranted.

RESULTS

Participant Characteristics

The final sample consisted of all 542 women who reported in the GROW survey ever having smoked at least 100 cigarettes in their lifetime. As summarized in Table 1, 43% identified as non-Hispanic white, 7% as non-Hispanic black, and 15% as Asian/Pacific Islander. Half of the 35% of respondents who were Latina were born in the US, and half were born outside the US. Nearly half of participants were between the ages of 30 and 39, nearly 75% were married to or cohabitating with a partner, and over three-fifths of all women had at least some college education.

Table 1.

Characteristics and Prevalence of Quitting among Ever Smokers, Geographic Research on Wellbeing (GROW) Study, 2012–2013 (N = 542)

	Weighted % (N)	Cessation Prevalence Weighted % (95% CI)
Race/Ethnicity
Asian/Pacific Islander	14.6 (42)	72.1 (56.6–87.6)
Black	7.2 (65)	40.5 (26.8–54.1)
Latina, immigrant	17.4 (73)	82.0 (72.6–91.4)
Latina, US-born	17.4 (87)	69.6 (59.1–80.1)
White	43.4 (275)	75.3 (69.5–81.2)
Age
20–29 years	17.9 (79)	58.4 (46.6–70.3)
30–39 years	48.9 (250)	73.7 (67.3–80.1)
40+ years	33.2 (213)	78.4 (72.1–84.7)
Partner Status
Married/cohabitating	73.3 (397)	79.2 (74.6–83.8)
Previously/never married	26.7 (139)	55.2 (45.6–64.8)
Educational Attainment
Did not complete high school	14.7 (65)	63.2 (49.6–76.8)
High school graduate/GED	23.5 (108)	70.0 (60.1–79.9)
Some college	33.1 (178)	67.4 (59.6–75.3)
College graduate	28.7 (188)	84.8 (79.1–90.6)
Number of Children in Household
0–1	13.1 (66)	68.4 (54.2–82.5)
2–4	78.8 (423)	73.3 (68.8–78.4)
5 or more	8.1 (43)	68.7 (53.4–84.0)
Employment Status
Did not work	44.5 (231)	75.8 (69.8–81.8)
Part-time	26.1 (138)	72.2 (62.8–81.6)
Full-time	29.4 (167)	68.6 (60.4–76.7)
Long-term Neighborhood Povertya
High	9.5 (40)	60.3 (44.6–76.0)
Moderate	22.6 (107)	68.1 (58.1–78.1)
Low	67.9 (361)	76.6 (71.6–81.6)
Perceived Neighborhood Safety
Very safe	36.5 (207)	78.6 (71.9–85.3)
Somewhat safe	46.7 (241)	71.3 (64.7–77.8)
Somewhat/very unsafe	16.8 (90)	61.5 (50.0–73.0)
Income
Missing	8.5 (39)	66.7 (48.1–85.4)
≤200% federal poverty level	39.5 (195)	62.1 (54.3–69.8)
201%–400% federal poverty level	25.4 (126)	76.1 (67.9–84.3)
>400% federal poverty level	26.6 (182)	86.5 (81.2–91.8)
Change in Income Since Baseline
Dropped	20.5 (90)	64.4 (53.0–75.8)
Did not drop	79.5 (392)	75.3 (70.4–80.3)
Most Friends Smoke Cigarettes
Yes	14.7 (74)	36.6 (24.0–49.2)
No	85.3 (460)	79.5 (75.4–83.7)
Smoking in the Home
Never	97.0 (522)	74.1 (69.8–78.4)
Sometimes or daily	3.0 (19)	24.6 (2.6–46.7)
Worry about Discrimination
Not very often/never	86.2 (471)	73.9 (69.3–78.6)
Very/somewhat often	13.8 (71)	63.6 (50.4–76.9)
Food Insecurity
Yes	20.8 (113)	59.3 (49.1–69.5)
No	79.2 (427)	75.8 (71.0–80.6)
Perceived Stress
Not very often/never	49.8 (257)	75.0 (69.0–80.9)
Very/somewhat often	50.2 (285)	70.1 (63.8–76.4)
Emotional Support
Yes	95.1 (516)	73.5 (69.1–78.0)
No	4.5 (24)	53.7 (31.7–75.7)
Financial Support
Yes	79.4 (432)	75.3 (70.6–79.9)
No	20.6 (105)	65.6 (55.6–75.6)
Practical Support
Yes	91.0 (495)	74.1 (69.7–78.6)
No	9.0 (46)	57.5 (41.3–73.7)

Note.

^aExcludes inaccurate geocodes at the census tract level (N = 16) and women not residing in California at the time of GROW (N = 34).

The majority of women had 2–4 children at home, and over half were employed in the previous 2 weeks. About one-third of participants lived in a neighborhood that had experienced long-term high or moderate poverty levels, and nearly 17% perceived their neighborhood to be somewhat or very unsafe. Nearly half of participants earned an income less than 200% of the federal poverty level, and over one-fifth experienced a drop in income since their baseline MIHA survey. Almost 15% reported that most of their friends smoked cigarettes, but only 3% reported that people smoked inside their home. Fourteen percent reported worry about discrimination very or somewhat often; one-fifth of respondents reported food insecurity. Finally, although half of the participants perceived relatively high stress levels (more than they could comfortably handle very or somewhat often), they also reported high levels of social support, depending on the type of support (79%-95%).

Prevalence of Ever Quitting

The overall quit ratio (proportion of women that had ever smoked who were no longer smoking at the time of the survey) for the GROW sample was 73%. When examined by race/ethnicity (Table 1) the quit ratio was lowest for black women (41%) and highest for immigrant Latinas (82%). As expected, the quit ratio increased with age, and was higher among married/cohabitating versus previously/never married women. The quit ratio was highest for women with the highest level of educational attainment.

Quit ratios did not vary much by the number of children in the household (68%-73%) or employment status (67%-76%). Quit ratios were lowest for women living in neighborhoods that have experienced long-term high poverty (60%) and neighborhoods that are perceived as being unsafe (62%) compared with their counterparts. Quit ratios were also lowest for women with low incomes (62%) or whose income dropped since baseline (64%) compared with higher income women or those whose income did not drop. Quit ratios were very low for women reporting that most of their friends smoke cigarettes (37%) or women reporting that people smoke inside her home (25%). However, quit ratios were 70% for women not reporting discrimination (not very often/never), those who were food secure, those not experiencing stress very/somewhat often, and those with social support.

Sociodemographic Associations with Smoking Status

Compared with white women, black women had significantly lower odds of being a former smoker (odds ratio [OR] = 0.22, 95% confidence interval [CI]: 0.12–0.43) in unadjusted analyses (Table 2). No other racial/ethnic group differed significantly from white women in their odds of being a former smoker in unadjusted analyses. This disparity remained after adjusting for age, partner status, and educational attainment (adjusted odds ratio [AOR] = 0.47, 95% CI=0.23–0.98; Table 2). In addition, immigrant Latinas had significantly higher odds of being a former smoker compared to white women in adjusted analyses (AOR = 2.82, 95% CI: 1.17–6.82). Previously/never married women had lower odds of being a former smoker compared with married or cohabitating women (AOR = 0.44, 95% CI: 0.26–0.75), and women with lower educational attainment had lower odds of being a former smoker compared with college graduates.

Table 2.

Associations between Sociodemographic Characteristics and Quitting, Geographic Research on Wellbeing (GROW) Study, 2012–2013 (N = 542)

	Unadjusted Odds Ratio	95% Confidence Interval
Race/Ethnicity
Asian/Pacific Islander	0.85	0.36–1.94
Black	0.22**	0.12–0.43
Latina, immigrant	1.49	0.73–3.04
Latina, US-born	0.75	0.42–1.35
White	1.00
	Adjusted Odds Ratio	95% Confidence Interval
Race/Ethnicity
Asian/Pacific Islander	1.01	0.46–2.23
Black	0.47*	0.23–0.98
Latina, immigrant	2.82*	1.17–6.82
Latina, US-born	1.20	0.61–2.34
White	1.00
Age	1.03	0.99–1.07
Partner Status
Married/cohabitating	1.00
Previously/never married	0.44**	0.26–0.75
Educational Attainment
Did not graduate high school	0.33*	0.14–0.78
High school graduate/GED	0.47*	0.23–1.00
Some college	0.52*	0.28–0.96
College graduate	1.00

^*p < .05;

^**p < .01

Psychosocial and Environmental Associations with Smoking Status

Each variable was tested for its association with smoking status in a separate logistic regression, controlling for race/ethnicity, age, partner status, and educational attainment. These results are summarized in Table 3. Estimates for race/ ethnicity, age, partner status, and educational attainment were generally stable across models and similar to those presented in Table 2.

Table 3.

Associations between Psychosocial and Environmental Characteristics and Quitting, Geographic Research on Wellbeing (GROW) Study, 2012–2013 (N = 542)

	Adjusted Odds Ratio	95% Confidence Interval
Number of Children in Household
0–1	1.00
2–4	1.11	0.58–2.15
5 or more	1.26	0.47–3.40
Employment Status
Did not work	1.00
Part-time	0.66	0.36–1.21
Full-time	0.58	0.33–1.03
Long-term Neighborhood Povertya
High	0.60	0.24–1.51
Moderate	0.67	0.36–1.25
Low	1.00
Perceived Neighborhood Safety
Very safe	1.00
Somewhat safe	0.65	0.38–1.11
Somewhat/very unsafe	0.46*	0.24–0.88
Income
Missing	0.48	0.19–1.22
≤200% federal poverty level	0.36**	0.17–0.76
201%–400% federal poverty level	0.57	0.29–1.13
>400% federal poverty level	1.00
Change in Income Since Baseline
Dropped	0.66	0.36–1.21
Did not drop	1.00
Most Friends Smoke Cigarettes
Yes	0.20***	0.10–0.37
No	1.00
Smoking in the Home
Never	1.00
Sometimes or daily	0.18**	0.05–0.66
Worry about Discrimination
Not very often/never	1.00
Very/somewhat often	0.65	0.33–1.28
Food Insecurity
Yes	0.55*	0.32–0.95
No	1.00
Perceived Stress
Not very often/never	1.00
Very/somewhat often	0.74	0.46–1.19
Emotional Support
Yes	2.15	0.92–5.03
No	1.00
Financial Support
Yes	1.36	0.74–2.50
No	1.00
Practical Support
Yes	1.82	0.82–4.06
No	1.00

^*p < .05;

^**p < .01;

^***p < .001

Note.

Each covariate was tested in a separate logistic regression model controlling for race/ethnicity, age, partner status, and educational attainment.

^aExcludes inaccurate geocodes at the census tract level (N = 16) and women not residing in California at the time of GROW (N = 34).

Perceptions of one’s neighborhood as somewhat or very unsafe, having low income, having a majority of friends who smoke, having persons who smoke in the home, and experiencing food insecurity each was associated with decreased odds of being a former smoker. When these 5 variables were entered into a single model with covariates to examine their unique associations with smoking status, only having a majority of friends who smoke retained statistical significance (Table 4). In addition, the association between identifying as black and smoking status was no longer significant after adjusting for the 5 significant psychosocial and environmental variables; however, the association between self-identification as immigrant Latina and smoking status remained statistically significant.

Table 4.

Final Model of Associations between Psychosocial and Environmental Characteristics and Quitting, Geographic Research on Wellbeing (GROW) Study, 2012–2013 (N = 542)

	Adjusted Odds Ratio	95% Confidence Interval
Race/Ethnicity
Asian/Pacific Islander	1.34	0.59–3.05
Black	0.72	0.32–1.65
Latina, immigrant	3.64*	1.28–10.32
Latina, US-born	1.27	0.62–2.61
White	1.00
Age	1.00	0.95–1.04
Partner Status
Married/cohabitating	1.00
Previously/never married	0.59	0.33–1.06
Educational Attainment
Did not complete high school	0.47	0.18–1.25
High school graduate/GED	0.62	0.28–1.39
Some college	0.62	0.31–1.26
College graduate	1.00
Income
Missing	0.49	0.17–1.38
<200% federal poverty level	0.45	0.20–1.05
201%–400% federal poverty level	0.63	0.30–1.36
>400% federal poverty level	1.00
Perceived Neighborhood Safety
Very safe	1.00
Somewhat safe	0.80	0.46–1.41
Somewhat/very unsafe	0.72	0.34–1.51
Most Friends Smoke Cigarettes
Yes	0.23**	0.12–0.44
No	1.00
Smoking in the Home
Never	1.00
Sometimes to daily	0.26	0.07–1.01
Food Insecurity
Yes	0.83	0.45–1.55
No	1.00

^*p < .05;

^**p < .001

DISCUSSION

Our study examined racial/ethnic differences in sociodemographic, psychosocial, and environmental variables relevant to smoking status in a diverse sample of women with a history of smoking. As such, it is among the first to characterize areas of disadvantage in a sample of women who have ever smoked, and adds to the limited extant research on correlates of cessation among women.

Black women had half the odds of being a former smoker compared with white women after adjusting for sociodemographics. However, this disparity was attenuated after additionally adjusting for important psychosocial and environmental variables. This is a notable finding, as the social and environmental factors identified here are amenable to change, thereby allowing for more opportunities for intervention. For example, environmental-level interventions which may promote smoking cessation include implementing a smoking ban in the home or in housing units, community-level programs to reduce crime, and food assistance programs. In addition, these interventions may have beneficial health effects beyond cessation. Increasing interpersonal skills for navigating and avoiding social environments where smoking is common also may be important. More generally, these results underscore the importance of social context as a determinant of racial/ethnic health disparities and the need to consider a more holistic approach to supporting smoking cessation among women that addresses both psychosocial and environmental determinants.

Immigrant Latinas had almost 3 times higher odds of being a former smoker compared with white women in analyses adjusted for sociodemographics. That this association was not found in unadjusted analyses suggests a strong suppression effect¹⁰⁵ of sociodemographic variables on the relationship between immigrant Latina status and smoking status. That is, immigrant status and sociodemographics, particularly educational attainment, share variance that contributes to the association between immigrant Latina status and smoking status. Thus, future research with immigrant Latinas should consider age, partner status, and educational attainment to gain a clear understanding of the process of smoking cessation among this population.

The association between being an immigrant Latina and smoking status was not attenuated after additionally adjusting for psychosocial and environmental variables. Thus, additional research is needed to identify factors that account for this association to identify additional potential treatment targets. It is not known to what extent, if at all, the immigrant status variable used here is associated with acculturation, which is a cultural construct thought to consist of attitudes, values, and practices rooted in one’s cultures.¹⁰⁶ However, immigrant status has consistently been associated with acculturation.^106–108 Thus, future research may build upon current findings by examining the extent to which acculturation is associated with smoking status or cessation. In particular, studies using theoretically grounded multidimensional measures may help to clarify which specific aspects of acculturation are important to cessation. Alternatively, immigrant status may be associated with other, non-culturally specific determinants of cessation which were not assessed here (eg, family support, partner support). Thus, future research may also benefit from examining whether immigrant status and/or acculturation is associated with known determinants of cessation among Latinas.

In individual analyses, 5 psychosocial and environmental variables were significantly associated with smoking status above and beyond sociodemographic variables. These were neighborhood safety, low income, having a majority of friends who smoke, allowing persons to smoke in the home, and food insecurity. When these 5 variables were entered simultaneously into a single model, only having a majority of friends who smoke remained a significantly associated with smoking status. This relationship speaks to the relative importance of smokers in one’s social network to the process of cessation among women. However, given the cross-sectional design of the study, it cannot be determined whether quitting encourages women to seek social networks with fewer smokers or vice versa.

Future research may benefit from prospective studies that test causal hypotheses about the relationship between these variables (neighborhood safety, having a majority of friends who smoke, allowing persons to smoke in the home, and food insecurity) and smoking cessation. In addition, it will be important to investigate the mechanisms through which these variables exert effects on cessation. Consistent with social-cognitive theory of drug use and relapse,¹⁰⁹ these psychosocial and environmental variables may be relatively distal contextual determinants that indirectly affect cessation through more proximal intrapersonal determinants. For example, having smokers in the home may increase craving for cigarettes or decrease one’s self-efficacy to abstain from smoking, which may in turn undermine a quit attempt. As another example, food insecurity or concerns about neighborhood safety may increase the experience of negative emotions or stress, which may in turn undermine quitting. Research into the mechanisms that account for these associations not only serves to inform both theoretical models of the process of smoking cessation, but also has implications for intervention development by identifying additional intrapersonal treatment targets.

Our study has some limitations. First, with cross-sectional data, we cannot speak to the causal nature of the observed associations among the variables. Similarly, because smoking status data were not comparable from MIHA to GROW and we cannot determine precisely when women quit, we could not examine change in smoking status prospectively. Prospective research among women actively engaged in a quit attempt is needed to gain insight on causal relationships and mechanisms. Second, the sample size (N = 542) is considered small by epidemiological standards, as are the sample sizes of the racial/ethnic minority groups examined here. Generalizability of our findings, particularly for the black and Latina immigrant groups, should be considered with this limitation in mind. Third is the use of categorical variables to measure constructs that may be more accurately represented on a continuum (eg, perceived support, perceived stress). This can result in a loss of variability and reduce the ability to detect some relationships. Lastly, the women in the sample were all mothers of young children, and this sample characteristic may influence the observed relationships between the variables examined here. Additional research with other samples of women is needed to support generalizability of the current findings.

In summary, we examined racial/ethnic disparities in smoking status and cross-sectional associations between psychosocial and environmental variables with status in a diverse sample of women who have ever smoked cigarettes. The current research sheds light on potential determinants of cessation among women outside of the peripartum period, where the majority of research on women smokers has focused. Results underscore the importance of considering psychosocial and environmental variables in the process of smoking cessation among women, but prospective research is needed.