Relationship of religiosity and spirituality to hazardous drinking, drug use, and depression among sexual minority women

Abstract

Using data from Wave 3 of the Chicago Health and Life Experiences of Women (CHLEW) study (N = 699) we explored whether religiosity and spirituality were associated with risk of hazardous drinking, drug use, and depression among sexual minority women (SMW; lesbian, bisexual) and possible differences by race/ethnicity. Participants were more likely to endorse spirituality than religiosity, and endorsement of each was highest among African-American SMW. We found no protective effect of religiosity or spirituality for hazardous drinking or drug use. An association initially found between identifying as very spiritual and past year depression disappeared when controlling for help-seeking. Among SMW with high religiosity, African-American SMW were more likely than White SMW to report hazardous drinking. Latina SMW with higher spirituality were more likely than White SMW to report drug use. Results suggest that religiosity and spirituality affect subgroups differently, which should be considered in future research on resiliency among SMW.

Introduction

Research on disparities in substance use and psychological distress among sexual minorities suggests that sexual minority women (SMW; lesbian, bisexual) are more likely than heterosexual women to report hazardous drinking, drug use, and depression (Balsam, Beauchaine, Mickey, & Rothblum, 2005; Bostwick, Boyd, Hughes, & McCabe, 2010; Drabble, Midanik, & Trocki, 2005; Hughes, Szalacha, & McNair, 2010; McCabe, Hughes, Bostwick, West, & Boyd, 2009; Trocki, Drabble, & Midanik, 2009; Wilsnack et al., 2008). SMW are also more likely than heterosexual women to seek treatment for psychological distress and substance use problems (Drabble et al., 2005; Grella, Greenwell, Mays, & Cochran, 2009; Jessup & Dibble, 2012). Research to date has focused disproportionately on risk factors, and there is a need to better understand factors that may protect against hazardous drinking, depression and drug use among sexual minorities (Balsam, 2003; Condit, Kataji, Drabble, & Trocki, 2011; Hughes, Matthews, Razzano, & Aranda, 2003; Kwon, 2013).

Studies in the U.S. suggest that religiosity is generally lower among sexual minorities than heterosexuals and that differences are most notable among women (Rostosky, Danner, & Riggle, 2008; Sherkat, 2002). In addition, sexual minorities appear to be more likely to embrace spiritual than religious identities (Eliason, Burke, van Olphen, & Howell, 2011; Halkitis et al., 2009) and endorsement of spirituality may be higher among sexual minorities compared to heterosexuals (Eliason et al., 2011). Spirituality generally refers to individual beliefs and practices and subjective experiences (such as seeking transcendence or feelings connection to a higher power) (Aldwin, Park, Jeong, & Nath, 2014; Allen & Lo, 2010). By contrast, religiosity is typically defined by involvement with doctrinal, organizational, and institutional dimensions of faith traditions (Koenig, King, & Carson, 2012).

In the general population, religiosity has largely been found to protect against hazardous drinking, drug use, and depression in both adults and adolescents (Allen & Lo, 2010; Bonelli & Koenig, 2013; Borders, Curran, Mattox, & Booth, 2010; Edlund et al., 2010; Haber, Grant, Jacob, Koenig, & Heath, 2012; Michalak, Trocki, & Bond, 2007). For example, higher levels of religiosity are associated with abstinence from drinking and lower rates of alcohol dependence (Michalak et al, 2007). Findings from a smaller body of literature that includes measures of spirituality also suggest a protective effect of spirituality in relation to hazardous drinking, drug use, and depression (Bonelli & Koenig, 2013; Chitwood, Weiss, & Leukefeld, 2008; Haber et al., 2012). However, at least one large population-based study in the U.S. found no protective effect of spirituality for past year illicit drug use and increased odds of reporting lifetime cocaine or injection drug use among individuals who reported higher levels of spirituality (Allen & Lo, 2010).

In contrast to research with general population samples, findings from recent studies of sexual minorities suggest that the role of religiosity as a risk or protective factor is mixed in relation to hazardous drinking and psychological distress (Anderton, Pender, & Asner-Self, 2011; Barnes & Meyer, 2012; Eliason et al., 2011; Page, Lindahl, & Malik, 2013; Rostosky, Danner, & Riggle, 2007; Rostosky, Danner, & Riggle, 2010). Using data from the National Longitudinal Study on Adolescent Health, Rostosky and colleagues (2010) found that religiosity was protective against heavy episodic drinking among young adult men (both heterosexual and sexual minority) and heterosexual women, but not among lesbians. Furthermore, this study found that religiosity was associated with increased risk among bisexual women.

Theoretically, the effects of religiosity on health occur primarily through behavioral self-regulation (e.g., encouraging less smoking, alcohol consumption, drug use), whereas spirituality mediates the impact of stress on health by improving emotional self-regulation (e.g., reducing negative arousal and stress through cultivating calmness or refocusing attention) (Aldwin et al., 2014). From this perspective, religious alienation may increase risk of unhealthy behaviors and general distress (Aldwin et al., 2014). Because sexual minorities are not welcome in many religious contexts, and in some cases experience abuse or rejection (Wood & Conley, 2014), religiosity may have mixed effects on the risk of substance use and psychological distress among sexual minorities (Dahl & Galliher, 2010; Gattis, Woodford, & Han, 2014; Page et al., 2013; Ream & Savin-Williams, 2005). For example, some researchers have found that affiliation with non-affirming religious organizations may be associated with greater internalized heterosexism (Barnes & Meyer, 2012; Page et al., 2013; Ream & Savin-Williams, 2005).

There is a paucity of research on the potential protective effects of religiosity and spirituality among adult populations of SMW. Findings from a recent study examining protective effects of religiosity among women by sexual identity suggested that higher levels of religiosity were protective against hazardous drinking and drug use among both sexual minority and heterosexual women (Drabble, Trocki, & Klinger, 2016). However, this study had a number of limitations including a small sample of SMW, absence of measures of spirituality, and insufficient power to explore possible differences by race/ethnicity. In a study of college students, religiosity, but not spirituality, was protective against substance abuse for non-heterosexual compared to heterosexual students (Eliason et al., 2011). Given these somewhat inconsistent results there is a need for research with more diverse samples of adult sexual minorities.

In the general population, African-American and Latina women generally report lower levels of hazardous drinking, drug use, and depression compared to White women (Hasin & Grant, 2015). By contrast, although some research has found lower rates of hazardous drinking among African American compared to White SMW (Lewis, Mason, Winstead, Gaskins, & Irons, 2016), many studies have found few differences, or even higher rates, of hazardous drinking among African-American and Latina SMW compared to White SMW (Cochran, Mays, Ortega, Alegria, & Takeuchi, 2007; Hughes et al., 2003; Hughes et al., 2006; Matthews et al., 2014; Mereish & Bradford, 2014). Similarly, research findings suggest higher rates of depression among SMW of color compared to heterosexual women of color (Cochran, Mays, Ortega, Alegria, & Takeuchi, 2007; Matthews, Hughes, Johnson, Razzano, & Cassidy, 2002; Mays, Cochran, & Roeder, 2004). Findings regarding depression among African-American and Latina SMW compared to White SMW are inconsistent (Aranda et al., 2015; Balsam et al., 2015; Jeong, Veldhuis, Aranda, & Hughes, 2016). Given that lower rates of hazardous drinking, drug use, and depression typically found among women of color compared to White women in the general population are not seen consistently among SMW, it seems likely that dynamics of risk and protection may not operate in the same way among SMW (Cochran et al., 2007; Hughes, Wilsnack, & Kantor, 2016; Hughes et al., 2006; Matthews et al., 2014; Mereish & Bradford, 2014). For example, cultural, religious, and familial norms against alcohol use, along with other factors that have been found to be protective against heavy/hazardous drinking among women of color (Zapolski, Pedersen, McCarthy, & Smith, 2014), may not provide similar protections for SMW of color.

The current study uses data from a large, diverse sample of lesbian and bisexual women and disaggregates constructs of religiosity and spirituality to explore three questions: 1) How does endorsement of religiosity and spirituality differ by race/ethnicity and other demographic characteristics among lesbian and bisexual women?; 2) Are religiosity and spirituality associated with the study outcomes (past year hazardous drinking, drug use, and depression) among lesbian and bisexual women?; and 3) How do associations between religiosity and spirituality and the study outcomes differ by race/ethnicity?

Methods

Data for the current study are from 699 self-identified lesbian and bisexual women interviewed in Wave 3 (2010–2012) of the Chicago Health and Life Experiences of women (CHLEW) study, a longitudinal, community-based study of risk and protective factors associated with alcohol use and alcohol-related problems among SMW. Participants in the first wave of the CHLEW study (2000–2001) were predominately lesbian and were recruited using a broad range of strategies, including community-based organizations, media outreach, and individual social networks. Recruitment strategies focused on increasing representation of women who had been previously underrepresented in research, including lesbian women of color, older lesbians, and lesbians of lower socioeconomic status. In 2010–2012 (wave 3) a supplemental sample of younger women (ages 18–25), African American and Latina women, and bisexual women (N = 373) were recruited using a modified version of respondent-driven sampling (Martin, Johnson, & Hughes, 2015). Wave 3 data included a retention rate of 79% of the 447 participants in the original Wave 1 study.

Measures

Sexual identity

Sexual identity was assessed using an item that asked participants, “Recognizing that sexual identity is only one part of your identity, how would you define your sexual identity? Would you say that you are: ‘only lesbian/gay,’ ‘mostly lesbian/gay,’ ‘bisexual,’ ‘mostly heterosexual,’ ‘only heterosexual/straight’ or ‘other’” (McCabe, Hughes, Bostwick, Morales, & Boyd, 2012; Skrocki, 1996). Women who identified as only lesbian or mostly lesbian (n = 517), or bisexual (n = 182) were included in the current analyses. Because women who identified as only lesbian and mostly lesbian did not substantially differ on key study variables, they are combined as one group in the current study.

Religiosity and spirituality

Religiosity

Religiosity was assessed using responses to two questions. First, participants were asked, “What is your religious preference now?,” with response options of Protestant, Catholic, Jewish, Other or None. Second, participants who indicated a religious preference (anything other than “none”) were asked a follow-up question: “We would like to know how religious you would say you are. By ‘religious’ we mean how actively you currently follow the teachings of a specific religion and participate in activities of that religion. Would you say that you currently are: very religious, somewhat religious, or not at all religious?” The two religiosity items were used to construct a 3-category religiosity variable: no religious preference/not at all religious, somewhat religious, and very religious. We verified that there were no significant differences between participants reporting no religious preference and those indicating religion was not at all important (after identifying a religious preference) in relation to outcome variables before combining these categories. Spirituality: Spirituality was measured by the following question: “We would also like to know about your spirituality. By ‘spirituality,’ we mean how often you spend time thinking about the ultimate purpose of life or your own relationship to a higher power in life. In this sense, would you say that you currently are: ‘very spiritual,’ ‘somewhat spiritual,’ or ‘not at all spiritual’.” The religiosity measures were adopted from other national surveys (Michalak et al., 2007; Wilsnack, Wilsnack, & Klassen, 1984) and the spirituality measure was created to parallel the religiosity measure. The CHLEW study team used brief measures of spirituality and religiosity in the CHLEW study to minimize participant burden. Although the religiosity and spirituality variables were significantly correlated (.285, p<.001), the size of the correlation was less than .3, which is considered small (Mukaka, 2012). Given sexual identity differences in prior literature we examined the effects of religiosity and spirituality separately for lesbian and bisexual women.

Alcohol, drug, and mental health variables

Hazardous drinking (past 12 months)

A dichotomous measure of hazardous drinking was constructed using four dichotomous variables: consuming six or more drinks in one day on one or more occasion in the past year (heavy episodic drinking), having consumed “enough to feel drunk” once or more in the past year, reporting one or more of five past year dependence symptoms (inability to stop drinking before becoming intoxicated, inability to stop or reduce alcohol consumption over time, memory lapses [blackouts] while drinking, rapid drinking, and morning drinking), and reporting one or more adverse drinking consequences (e.g., driving while intoxicated, concerns/complaints about drinking from family/friends, partner requests to cut down, arguments with partner with partner when drinking, arguments or fights with someone other than a partner or family member when drinking, and negative impact on obtaining work/promotion). Alcohol variables used in this study, such as symptoms of potential alcohol dependence and alcohol consequences, were adapted from other national studies in the U.S. (Midanik & Greenfield, 2000; Wilsnack, Kristjanson, Wilsnack, & Crosby, 2006; Wilsnack et al., 1984). Dichotomized (any/none) responses for each of the four indicators were used to create a hazardous drinking score (0–4). A hazardous drinking indicator was constructed by dichotomizing the 0–4 measure, with a score of 2 or more indicating hazardous drinking. The hazardous drinking index has been validated in the CHLEW sample (Riley et al., 2017) and has been used in research comparing sexual minority and heterosexual women (Drabble, Trocki, Hughes, Korcha, & Lown, 2013; Hughes, Johnson, Steffen, Wilsnack, & Everett, 2014; Hughes, Szalacha, Johnson et al., 2010; Johnson et al., 2013; Matthews et al., 2013). Drug use was constructed as any past year non-medical use of three types of prescription drugs (tranquilizers, sleeping pills, or painkillers) or any use of five types of illicit drugs (uppers/stimulants, cocaine/crack, heroin/illegal methadone, hallucinogens, or club drugs), or use of marijuana. Past year depression. Depression was assessed based on questions from the National Institute of Mental Health Diagnostic Interview Schedule, which approximates a clinical diagnosis of depression (Robins, Helzer, Croughan, & Ratcliff, 1981). Participants were asked whether they ever had two or more weeks when they experienced three or more depressive symptoms, such as appetite changes, trouble sleeping, feeling tired all the time, loss of interest in sex, feeling worthless, and difficulty thinking/concentrating. They were then asked whether these symptoms had persisted for two weeks or more (called an “episode”). Responses to a question about when the last episode of depression occurred was used to construct a past 12-month depression measure (one or more episodes in the past year vs. none).

We used two additional variables in follow-up analyses: internalized heterosexism and seeking help for an alcohol or drug problem. Internalized heterosexism questions were adapted from Herek and colleagues Internalized Homophobia Scale (IHP) (Herek, Cogan, Gillis, & Glunt, 1998; Herek, Gillis, & Cogan, 2009). The 10-item scale in the CHLEW study includes 5 items from a brief version of the IHP that was determined to be more suitable for administration with lesbians and bisexual women by the authors of the scale (Herek et al, 2009), which assess the degree to which participants agree with statements related to feeling uneasy about, and seeking to avoid, same-sex attraction (e.g., “I have tried to stop being attracted to women in general”; “I would like to get professional help in order to change my sexual orientation from lesbian/bisexual to straight”). The scale also includes 5 positive statements (e.g., I am proud that I am lesbian/bisexual), which were reverse-scored. Ten items were averaged for each participant to create a composite internalized heterosexism score (range – 0 to 5). Lower scores on this scale indicate lower levels of internalized heterosexism (α = .83). Past help-seeking was constructed as a dichotomous variable comparing participants who responded in the affirmative with those who did not to either of the following two questions: “Have you ever gone to anyone— a treatment agency, Alcoholics Anonymous, anyone at all—for help with an alcohol problem of your own?,” and “Have you ever gone to anyone—a treatment agency, Narcotics Anonymous, anyone at all—for help with a drug problem of your own?”

Demographic Characteristics

Demographic measures included age (<30 years, 30–39 years, 30–49 years, 50 and older), education (high school/GED or less, some college or 2-year degree, bachelor’s degree, and graduate or professional degree), employment status (employed vs. not employed), and relationship status (partnered vs. not partnered). Race/ethnicity was constructed as a four-category variable (White, African American, Latina, or Asian/Pacific Islander [API]/Other); the small number of participants who identified as Asian/Pacific Islander, Native American or a biracial or multiracial were combined. Dichotomous variables were created for having any children living in the household (children in the home vs. no children in the home).

Data Analysis

We used Chi square analyses in comparisons of categorical variables and analysis of variance (ANOVA) in comparisons using continuous outcome measures. Follow-up tests were conducted on variables found to be significant in bivariate analyses (logistic regression for categorical variables and Bonferroni post hoc tests for ANOVA).

We used logistic regression analyses to test three models predicting past year hazardous drinking, drug use, and depression. First, we examined main effects of religiosity and spirituality and demographics in relation to the three outcomes. Second, we added interaction terms to our regression models to examine potential interaction effects of race/ethnicity and religiosity and spirituality (separately) in predicting past year hazardous drinking, past year drug use, and past year depression (controlling for age, education, employment, partnership status, and children living in the home). For ease of interpretation, religiosity and spirituality were entered into the models as continuous variables, with higher numbers indicating higher levels. Third, we tested the interaction effects of race/ethnicity and religiosity and spirituality, when including internalized heterosexism and seeking help for an alcohol or drug problem respectively.

Findings

Demographics and differences in religiosity and spirituality

Table 1 summarizes the distributions of demographic and key study variables. In response to the question about current religious preference, 42.3% of participants indicated no religious preference and 57.7% endorsed some form of religious preference (34.7% Protestant, 12.2% Catholic, 3.7% Jewish, and 7.0% other, such as Buddhist or Pagan). Participants reporting religious affiliation, but describing themselves as “not at all religious” were combined with participants who reported “no religion.” Using this constructed variable, 51.7% of participants were classified as having no religious preference/not at all religious, whereas 34.8% indicated that they were somewhat religious, and 13.5% said they were very religious. In contrast, the distribution of responses to the question about how spiritual participants perceived themselves to be was skewed toward higher spirituality: not at all spiritual, 9.6%; somewhat spiritual, 42.9%; and very spiritual, 47.5%.

Table 1.

Sample Demographics, Religiosity, Spirituality, and Outcome Variables (N = 699)

	Percent (n)
Religiosity
Very religious	13.5 (94)
Somewhat religious	34.8 (242)
No preference/not at all religious	51.7 (360)
Spirituality
Very spiritual	47.5 (332)
Somewhat spiritual	42.9 (300)
Not at all spiritual	9.6 (67)
Sexual identity
Lesbian	74.0 (517)
Bisexual	26.0 (182)
Age
18–30	32.2 (225)
31–40	21.3 (149)
41–50	20.2 (141)
50 and over	26.3 (184)
Race/Ethnicity
White	37.3 (261)
African American	36.1 (252)
Latina	23.2 (162)
Other	3.4 (24)
Relationship status
Partnered	60.9 (424)
Not partnered	39.1 (272)
Education
High school or less	20.5 (143)
Some college	31.0 (217)
College graduate	21.0 (147)
Post graduate	27.5 (192)
Employment status
Employed	67.6 (472)
Not employed	32.4 (226)
Any children living with participant	20.2 (141)
Hazardous drinking index (2+)	43.9 (297)
Past year drug use	42.2 (295)
Past year depressive episode	30.3 (212)
Internalized heterosexism (Scale 1–5), Mean (sd)	1.47 (0.54)

Table 2 summarizes levels of religiosity and spirituality by participant demographic characteristics. Religiosity and spirituality varied significantly by race/ethnicity (p<.001). African-American/Black SMW were more likely to identify as somewhat or very religious than Latina and White SMW. Participants in the oldest (51+) age group were significantly more likely to identify as somewhat or very religious than those in 18–30 or 31–40 year age categories and were more likely to identify as very spiritual compared to 31–40 year olds. SMW with less education were significantly more likely to say they were somewhat religious compared to those with higher education (BA or post graduate degrees). SMW who had some college education were significantly more likely to say they were very spiritual compared to SMW with post graduate degrees. SMW with children living in the home were more likely to say they were somewhat or very religious compared to those without children living in the home.

Table 2.

Religiosity and Spirituality by Demographics and Outcome Variables (N = 699)

	Very Religious	Somewhat Religious	Not Religious/No Preference		Very Spiritual	Somewhat Spiritual	Not at all Spiritual
DEMOGRAPHICS	% (n)	% (n)	% (n)	p value	% (n)	% (n)	% (n)	p value
Sexual Identity
Lesbian	13.6 (70)	36.1 (186)	50.3 (259)	.401	49.3 (255)	42.4 (219)	8.3 (43)	.088
Bisexual	13.3 (24)	30.9 (56)	55.8 (101)		42.3 (78)	44.5 (81)	13.2 (24)
Age
18–30	8.9 (20)	28.9 (65)	62.2 (140)	<.001	39.6 (90)	49.8 (112)	10.7(24)	.006
31–40	11.4 (17)	31.5 (47)	57.0 (85)		43.6 (65)	42.3 (63)	14.1 (21)
41–50	17.3 (24)	41.0 (57)	41.7 (58)		56.7 (80)	38.3 (54)	5.0 (7)
51+	18.0 (33)	39.9 (73)	42.1 (77)		53.5 (98)	38.6 (71)	8.2 (15)
Ethnicity
White	10.0 (26)	26.4 (69)	63.6 (166)	<.001	38.3 (100)	47.5 (124)	14.2 (37)	<.001
African-American	18.1 (45)	46.6 (116)	35.3 (88)		59.5 (150)	36.1 (91)	4.4 (11)
Latina	10.5 (17)	33.3 (54)	56.2 (91)		43.2 (70)	45.7 (74)	11.1 (18)
Other	25.0 (6)	12.5 (3)	62.5 (15)		50.0 (12)	45.8 (11)	4.2 (1)
Relationship Status
Partnered	14.7 (62)	35.7 (151)	49.6 (211)	.289	46.0 (196)	44.3 (188)	9.7 (41)	.641
Not partnered	11.5 (31)	33.3 (90)	55.2 (149)		49.6 (135)	41.2 (112)	9.2 (25)
Employment
Employed	13.2 (62)	32.7 (154)	54.1 (255)	.151	45.6 (215)	44.1 (208)	10.4 (49)	.300
Not employed	14.3 (32)	39.3 (88)	46.4 (104)		51.3 (116)	40.7 (92)	8.0 (18)
Education
HS or Less	15.5 (22)	51.4 (73)	33.1 (47)	<.001	46.9 (67)	44.8 (64)	8.4 (12)	.264
Some college	10.6 (23)	37.5 (81)	51.9 (112)		53.9(117)	39.2 (85)	6.9 (15)
Bachelor’s degree	14.4 (21)	20.5 (30)	65.1 (95)		42.9 (63)	45.6 (67)	11.6 (17)
Graduate school	14.6 (28)	30.2 (58)	55.2 (106)		44.3 (85)	43.8 (84)	12.0 (23)
Children at home
Children at home	18.6 (26)	43.6 (61)	37.9 (53)	.001	52.5 (74)	42.6 (60)	5.0 (7)	.086
No children at home	12.2 (68)	32.6 (181)	55.2 (307)		46.1 (257)	43.1 (240)	10.8 (60)
OUTCOMES	% (n)	% (n)	% (n)	p value	% (n)	% (n)	% (n)	p value
Hazardous Drinking (Past yr)
2+ Indicators	11.8 (35)	30.7 (91)	57.4 (170)	.048	40.4 (120)	47.8 (142)	11.8 (35)	.014
0–1 Indicators	14.6 (55)	37.6 (142)	47.9 (181)		51.5 (195)	40.1 (152)	8.4 (32)
Drug Use (Past yr)
Any drug use	11.2 (33)	33.2 (98)	55.6 (166)	.144	45.4 (134)	44.4 (131)	10.2 (30)	.635
No drug use	64.9 (61)	35.9 (144)	48.9 (196)		49.0 (198)	41.8 (169)	9.2 (37)
Depression (Past yr)
One or more	12.8 (27)	37.4 (79)	49.8 (105)	.662	49.1 (104)	43.9 (93)	7.1 (15)	.287
No episode	14.4 (67)	34.0 (161)	51.9 (246)		46.4 (221)	42.6 (203)	10.9 (52)

Table 2 also summarizes the relationship between religiosity and spirituality and key outcome variables. Past year hazardous drinking was higher among participants who were not at all religious and among those who were not at all spiritual compared with participants who were very religious or very spiritual, respectively. Past year drug use and depression did not differ by religiosity or spirituality. Separate analyses for three types of drug use (marijuana use, non-medical use of prescription drugs, and use of illicit drugs) were conducted; none were significant and the dichotomous indicator of any drug use (vs. none) was used in subsequent analyses.

Relationship between Religiosity and Outcomes

In bivariate analyses, there was no effect of religiosity on drug use or depression and only one significant main effect of religiosity on hazardous drinking: women who reported being somewhat religious had significantly lower odds of hazardous drinking than women who were not religious, OR = .68 (95%CI [0.49, 0.96]), p < .05. However, when controlling for demographic variables (Table 3), religiosity was not statistically associated with any of the outcomes.

Table 3.

Multivariable logistic regression predictors for religion and past year hazardous drinking, drug use, and depression symptoms

	Hazardous Drinking	Drug Use	Depression
	Adj OR (95% CI)	Adj OR (95% CI)	Adj OR (95% CI)
Religiosity
No preference/not religious 1
Somewhat religious	0.75 (0.50, 1.12)	0.93 (0.64, 1.35)	1.25 (0.85, 1.85)
Very religious	1.04 (0.60, 1.80)	0.83 (0.49, 1.40)	1.17 (0.69, 2.00)
Race/Ethnicity
White 1
African-American	0.92 (0.57, 1.46)	1.08 (0.69, 1.96)	0.49 (0.30, 0.77) **
Latina	1.00 (0.63, 1.60)	0.81 (0.52, 1.27)	0.74 (0.46, 1.18)
Other	0.37 (0.12, 1.15)	1.55 (0.62, 3.89)	1.64 (0.66, 4.07)
Age
51+ 1
41–50	1.94 (1.03, 3.19) *	1.16 (0.65, 1.82)	1.12 (0.66, 1.92)
31–40	4.09 (2.39, 7.00) ***	2.29 (1.41, 3.73) **	1.18 (0.70, 2.00)
18–30	10.37 (6.14, 17.52) ***	4.13 (2.60, 6.56) ***	1.53 (0.95, 2.47)
Employment
Employed 1
Not employed	0.58 (0.38, 0.89) *	0.98 (0.67, 1.43)	1.11 (0.75, 1.64)
Education
High school/less 1
Some college	0.54 (0.32, 0.90) *	0.84 (0.53, 1.34)	1.02 (0.63, 1.64)
Bachelor’s degree	0.35 (0.19, 0.66) **	0.78 (0.45, 1.36)	0.55 (0.30, .99) *
Graduate school	0.39 (0.21, 0.72) **	0.44 (0.25, 0.78) *	0.44 (0.24, 0.80) **

¹Reference group

^***p<0.001,

^**p<0.01,

^*p<0.05

Note: Demographic variables not significant across multivariable logistic regression analyses are excluded from table: Relationship status, children living in the home

Differences by race/ethnicity

We examined the interactions between race/ethnicity and religiosity (as a continuous variable) predicting past year hazardous drinking, past year drug use, and past year depression controlling for demographic characteristics and internalized heterosexism. Only one significant interaction was found: African-American SMW who had higher levels of religiosity had elevated odds of hazardous drinking compared to white SMW AOR = 1.87 (95%CI [1.00, 3.49]), p =.05.

Follow-up analyses: Internalized heterosexism

Because internalized heterosexism or internalized stigma has been associated with religiosity in prior studies (Barnes & Meyer, 2012; Page et al., 2013), we re-ran our primary analyses and interaction analyses to examine whether internalized heterosexism would attenuate the relationship between religiosity and hazardous drinking among African-American SMW. Internalized heterosexism was significantly associated with religiosity (but not spirituality), in bivariate analysis: internalized heterosexism was significantly lower among participants reporting that religion was not at all important/no religious preference compared to participants who indicated that religion was somewhat important (p<.01) or very important (p<.05). However, internalized heterosexism was not significant in the multivariable logistic regression analyses.

Relationship between Spirituality and Outcomes

In bivariate analyses, there was a significant main effect of spirituality on hazardous drinking. Specifically, very spiritual SMW had significantly lower odds of hazardous drinking compared to SMW who reported that they were not at all spiritual, OR = .56 (95%CI [0.33, 0.97]), p < .05; however, this relationship was no longer significant in the model when controlling for demographic variables (Table 4). There was no significant relationship between spirituality and drug use in bivariate or multivariable regression models. There was a significant relationship between spirituality and depression; participants who were very spiritual had greater odds of reporting depression compared to those who were not at all spiritual (p = .05).

Table 4.

Multivariable logistic regression predictors for spirituality and past year hazardous drinking, drug use, and depression symptoms

	Hazardous Drinking	Drug Use	Depression
	Adj OR (95% CI)	Adj OR (95% CI)	Adj OR (95% CI)
Spirituality
Not spiritual 1
Somewhat spiritual	0.86 (0.47, 1.58)	0.94 (0.53, 1.67)	1.77 (0.92, 3.41)
Very spiritual	0.65 (0.35, 1.21)	0.94 (0.53, 1.68)	1.94 (1.00, 3.77) *
Race/Ethnicity
White 1
African-American	0.96 (0.60, 1.452)	1.05 (0.68, 1.62)	0.47 (0.29, 0.74) **
Latina	1.02 (0.64, 1.63)	0.81 (0.52, 1.26)	0.73 (0.46, 1.16)
Other	0.39(0.13, 1.22)	1.53 (0.61, 3.84)	1.54 (0.63, 3.81)
Age
51+ 1
41–50	2.00 (1.13, 3.54) *	1.15 (0.68, 1.93)	1.07 (0.63, 1.84)
31–40	4.02 (2.35, 6.88) ***	2.34 (1.43, 3.79) **	1.19 (0.71, 2.00)
18–30	10.32 (6.13, 17.35) ***	4.24 (2.68, 6.70) ***	1.51 (0.94, 2.41)
Employment
Employed 1
Not employed	0.59 (0.39, 0.90) **	0.96 (0.65, 1.40)	1.11 (0.75, 1.64)
Education
High school/less 1
Some college	0.59 (0.35, 0.99) *	0.85 (0.53, 1.35)	0.96 (0.60, 1.54)
Bachelor’s degree	0.40 (0.21, 0.73) **	0.79 (0.45, 1.36)	0.51 (0.28, 0.92) *
Graduate school	0.43 (0.23, 0.78) **	0.46 (0.26, 0.80) **	0.42 (0.23, 0.76) **

¹Reference group

^***p<0.001,

^**p<0.01,

^*p<0.05

Note: Demographic variables not significant across multivariable logistic regression analyses are excluded from table: Relationship status, children living in the home

Differences by race/ethnicity

There was only one significant interaction between race/ethnicity and spirituality (entered as a continuous variable in interaction models) in predicting the study outcomes when controlling for demographics. Latinas who reported higher levels of spirituality had higher odds of reporting drug use compared to white SMW, AOR = 1.97 (95%CI [1.08, 3.74]), p < .05.

Follow-up analyses: Help-seeking

We conducted follow-up analyses to determine whether having sought help for alcohol or drug problems in the past attenuated the relationship between identifying as very spiritual and increased odds of reporting depression (in multivariable regression analyses) or increased odds of reporting drug use (among Latinas in the interaction analysis). Our rationale for this was two-fold. First, as noted in the introduction, SMW are more likely than heterosexual women to seek help for substance use problems and mental health concerns. Second, endorsement of spirituality has been identified as important among individuals seeking help for alcohol and drug problems through formal treatment or self-help groups in both the general population and among SMW (Bliss, 2011; Hall, 1996; Walton-Moss, Ray, & Woodruff, 2013). Past help-seeking was reported by 9.9 % (n = 133) of participants and a majority of those who reported past help-seeking also reported being very spiritual (64.4%), which was significantly higher (44.0%, p <.05) than among participants who had never sought help. (There was no significant relationship between help-seeking and religiosity.) Not surprisingly, the odds for help-seeking were significantly associated with each of the outcome variables (hazardous drinking, AOR = 2.02, 95%CI [1.26, 3.24]), p < .01, drug use AOR = 1.69 (95%CI [1.10, 2.59]), p < .05), and depression (AOR = 2.30 (95%CI [1.50, 3.55]), p < .001). The increased odds of reporting depression among participants who identified as very spiritual (reported in Table 4) was attenuated when help-seeking was added into the model (p = .078). However, the relationship between drug use and spirituality among Latinas remained significant, AOR = 1.96 (95%CI [1.3, 3.72]), p < .05.

Discussion

We found that SMW in the CHLEW study were more likely to endorse spirituality (rather than religiosity) as very or somewhat important. It is noteworthy that although a majority of participants (51.7%) identified as not at all religious, only a very small percentage (less than 9.6%) said they were not at all spiritual. This finding is consistent with other studies and suggests that spirituality is more salient than religiosity among SMW (Halkitis et al., 2009; Sherry, Adelman, Whilde, & Quick, 2010). We also found significant differences in endorsement of religiosity and spirituality by race and ethnicity; specifically, African-American SMW endorsed higher levels of both spirituality and religiosity than did White SMW.

In contrast to findings from a recent study of the protective effect of religiosity among SMW in the National Alcohol Survey (Drabble et al., 2016), we found no evidence of a protective effect for religiosity in relation to hazardous drinking, drug use, or depression in the full sample. The absence of a protective effect for religiosity among SMW has also been found in studies of young adults (Rostosky et al., 2007; Rostosky et al., 2010), and underscores the need for additional research about the role of religion and spirituality in health behaviors and mental health among diverse populations of SMW. Initially, we found that the odds of reporting depression among women who identified as “very spiritual” were nearly twice that of participants who identified as “not at all spiritual.” However, this relationship lost significance in the follow-up analysis that added help-seeking for alcohol or other drug problems to the model. Given that spirituality is often emphasized in self-help and recovery programs, those who seek treatment may become more spiritual that those who do not. Because the co-occurrence of depression and substance-use disorders is high among SMW (Bostwick, Hughes, & Johnson, 2005; Jeong et al., 2016), it is possible that participants with depression and substance use disorders may embrace spirituality in the process of help-seeking and recovery. It is also possible that individuals who identify as very spiritual may be more likely to seek help for problems. There is therefore a need for longitudinal research to identify the causal pathways in the associations among depression, spirituality and help-seeking.

This study found few differences by race/ethnicity in relation to hazardous drinking, drug use, or depression. Only the odds of reporting past year depression were significantly lower among African-American SMW compared to White SMW; there were no significant differences by race/ethnicity in hazardous drinking or drug use. Likewise, in previous longitudinal analyses of a large probability sample of young adults, heterosexual-identified women compared to men, and people of color compared to Whites, were less likely to report binge drinking in the previous two weeks. Religiosity was also protective for the heterosexual-identified young adults. However, in the sexual minority young adult group, being a person of color was protective against binge drinking, but being female or reporting higher religiosity was not (Rostosky, Danner, & Riggle, 2007; 2008). In a closer examination of within-group differences after accounting for the protective effect of being a person of color, religiosity was not protective against heavy episodic drinking (HED) in lesbian women and was associated with greater risk of HED in bisexual-identified women (Rostosky, Danner, & Riggle, 2010).

In the current study, analyses that focused on the potential protective role that religiosity and spirituality may play in hazardous drinking, drug use, and depression among women of color revealed two significant interactions. First, the likelihood of hazardous drinking was higher among African-American women than White women with higher levels of religiosity (significant in models without internalized heterosexism and marginally significant when controlling for internalized heterosexism). We also found a significantly higher likelihood of drug use among Latina compared to White women with higher levels of spirituality. Together, these findings suggest that religiosity and spirituality may not provide the same protective effects for SMW of color that they do for women of color in the general population, and in fact may increase risk.

Possible associations between religiosity and risk of hazardous drinking or other health concerns among women of color are complex. Among African-American women, spirituality, but not religiosity, has been associated with higher levels of happiness (Battle & DeFreece, 2014). Exposure to homophobic messages within some religious institutions may decrease SMW’s comfort with their sexual identity and increase internalized heterosexism and overall minority stress (Miller & Stack, 2013; Page et al., 2013; Walker & Longmire-Avital, 2013). At the same time, participation in church or other religious institutions may serve a strong community and resilience function as being a member of both religious and sexual minority communities may provide separate sources of support and meaning (Miller & Stack, 2013). Furthermore, dynamics of support or intolerance in religious contexts may also be complex; for example, at least one recent study found that individual moral views, but not Black Protestant church affiliation, predicted intolerant attitudes about civil rights of sexual minorities (Ledet, 2016).

Although there is very little research on religiosity/spirituality among Latina SMW, qualitative studies have stressed the role of religious messages (e.g., expectations related to heterosexuality, gender conformity, and motherhood) as a source of stress and conflict among Latina lesbians (Reyes, 1998; Tuthill, 2016). Latina SMW who wish to retain connection to their family/community but who find the religious teachings of their church to be unsupportive of their sexual identity may define themselves as more spiritual than religious (Tuthill, 2016). One well-designed general population study found that spirituality was not protective for past year illicit drug use, and was positively associated with lifetime use of cocaine and injection drug use (Allen & Lo, 2010). The authors hypothesized that characteristics associated with spirituality (in contrast to religion) such as emphasis on self-actualization, individualism, and personal freedom may be predictive of drug use. Additional research with diverse samples of SMW is needed to better understand how spirituality, including dimensions of individualism and self-actualization, vary by ethnicity or other demographic characteristics and if these dimensions of spirituality relate to health behavior and mental health outcomes. Although one study found that disclosure of identity to nonfamily members (e.g., people at work or in community) was associated with less depression (Aranda et al., 2015), additional research is needed to explore possible relationships between religiosity/spirituality, disclosure, and drug use among Latina lesbians compared to other SMW.

Limitations

Several limitations should be considered when evaluating the results of this study. First, the sample was recruited using non-random methods, which limit generalizability. Second, because we used a cross-sectional design, causality and temporality could not be examined. Third, all measures relied on self-report, and thus responses may have been vulnerable to social desirability bias. Fourth, we had insufficient power to disaggregate participants who identified as Asian/Pacific Islander, Native American or a biracial or multiracial; these groups were combined in the analysis and therefore we were unable to explore possible differences among these groups. Fifth, the study used a single-item measure of religiosity precluding a more comprehensive assessment of the potential protective role of religiosity. For example, we had no measure of actual involvement or attendance at religious institutions, how positively participants viewed religiosity/spirituality in their lives, or how participants perceived the influence of religiosity/spirituality in their lives and well-being. Future research should examine these factors in more detail to better understand the potential mechanisms linking religiosity and spirituality with health behavior. Finally, we did not assess the level of acceptance or rejection SMW participants may have experienced in the context of their religious/spiritual affiliations. Previous research has suggested that a religion’s level of sexual minority affirmation may moderate the association between religious affiliation and psychological outcomes (Gattis et al., 2014), but additional research is needed to more fully understand mechanisms that may account for this finding.

Implications for Research and Practice

Our findings suggest that religiosity and spirituality are distinct constructs that may fulfill very different psychosocial needs for SMW. Traditionally, religiosity has been associated with institutional religion and relevant measures tend to focus on attendance and participation. Spirituality, on the other hand, is conceptualized as a more proximal, personal, and intrinsic source of meaning and purpose. In the current study, only a small percentage of women indicated that they were not at all spiritual. This finding is consistent with recent qualitative findings that many sexual minority individuals find positive ways to integrate their spiritual and sexual identities (Rosenkrantz, Rostosky, Riggle, & Cook, 2016). Future research on SMW’s spirituality and health-related behaviors will benefit from the use of multi-dimensional measures of spirituality that can be used to test theoretically-driven hypotheses about its possible function a source of support and resiliency.

Understanding the role of religiosity and spirituality in the lives of SMW of color is particularly important in light of the findings from this study. Variations in risk for hazardous drinking and drug use in analyses of interactions between race/ethnicity and religion and spirituality highlight the necessity of attending carefully to the complex intersections of identities when addressing problems related to substance use and abuse in research and practice. Practitioners, for instance, should inquire about the specifics of SMWs’ religiosity and spirituality and the extent to which they serve as sources of positive purpose, meaning, and support, as well as whether they serve as sources of rejection, recrimination, guilt, and shame. In the current study, we were not able to determine the specific mechanisms by which aspects of religiosity and spirituality may be associated with increases or decreases in health risk behaviors such as substance use. Nevertheless, given the high endorsement of spirituality among SMW, practitioners might refer clients to positive sources of spiritual support in the community and online. These types of resources might be particularly welcome to SMW who have difficulty finding ways to cope effectively with minority stress resulting from internalized negative beliefs that are often reinforced by religiously-motivated rejection from family and religious community members (Rostosky, Riggle, Brodnicki, & Olson, 2008).