Is irreligion a risk factor for suicidality? Findings from the Nashville stress and health study

Abstract

Suicide is a public health problem and one of the leading causes of death in the United States. Research exploring the linkages between religion and spirituality has received intermittent attention. Data come from the Nashville Stress and Health Study (2011–2014), a cross-sectional probability survey of black and white adults from Davidson County, Tennessee (n = 1,252). Results indicate that those with no perceived belief in divine control had a higher likelihood of suicidality. This study provides a fresh perspective on the links between religious factors and suicidality by (a) considering multiple religious and spiritual domains and (b) focusing on the association between irreligion and suicidality.

Introduction

Suicide is a top ten leading cause of death in the United States. Between 2006 and 2019 alone, the United States witnessed a 33% increase in suicide rates (CDC, 2021). However, global suicide trends have been less consistent during this period. Some areas of the world have seen downward or stable trends, while others have seen increasing rates after the 2008 financial crisis (Alicandro et al., 2019). Given these trends, a growing body of research explores possible risk and protective factors for suicidality, i.e., suicidal ideation and planning, attempted suicides, and completed suicides (e.g., Ivey-Stephenson et al., 2017; Steelesmith et al., 2019; Stone et al., 2018). The domain of religion and spirituality has received intermittent attention from investigators. One strand of work on this topic hews closely to the ecological strategy pioneered by Durkheim (1952 [1897]) in the late 19^th century. Briefly, he associated the suicide rates of cantons—i.e., administrative units in parts of Western Europe—with the religious composition of those units, particularly the proportion of Catholics contained therein. While ecological studies linking religious composition and suicide rates of areal units (cities, counties, nations) have continued to appear in the literature (Pescosolido & Georgianna, 1989; Ellison et al., 1997; Van Tubergen et al., 2005; Stack & Laubepin, 2019), most recent work on religion and suicidality uses individual-level data, including several studies with representative population samples.

Although findings are not unequivocal, the weight of evidence suggests that at least some aspects of individual religious involvement tend to deter suicidal thoughts, plans, and behaviors (Lawrence et al., 2016; Gearing & Alonzo, 2018). Such findings are consistent with a longstanding theoretical and empirical literature on religion, spirituality, and mental health outcomes, ranging from reduced depression, anxiety, distress, and substance (mis)use to higher levels of life satisfaction, happiness, positive self-concept, and other manifestations of flourishing (Koenig et al. 2001, 2012; Smith et al., 2003; Ellison & Henderson, 2011; Schieman et al. 2013, 2017; Krause, 2021; Upenieks & Schieman, 2022).

In recent years, however, some scholars have criticized the religion and health literature for neglecting the study of secular Americans (Hwang et al., 2011; Baker et al., 2018; Speed & Hwang, 2019). This topic has grown more important as the proportion of US adults who eschew religious affiliation, practice, and belief has increased in recent years and among younger cohorts (Schwadel, 2011; Hout & Fischer, 2014; Voas & Chaves, 2016). Taken together with shifts in rates of suicidality in recent years, these developments raise an important question: Are irreligious persons uniquely prone to suicidality relative to those who are at least somewhat religiously involved?

To our knowledge, this is the first study to address this question. Toward this end, we assess data from Vanderbilt University’s Nashville Stress and Health Study (NSAHS), a cross-sectional probability sample of 1,252 residents of Davidson County, Tennessee (i.e., urban Nashville), between 2011 and 2014. We identify persons who report a complete lack of religion according to several indicators, including frequency of attendance and socializing with church members, church-based social support, prayer, religious salience, religious coping, and belief in divine control. Our findings suggest that a complete lack of belief in divine control is a unique risk factor for suicidality, while other forms of irreligion do not predict suicidality. This is also one of the few studies to test non-linear associations between religious measures and outcomes relevant to mental health. After presenting our results, we discuss the possible implications for recent trends in religious involvement and suicide rates in the United States. We also identify several promising directions for future research.

Background

Over the years, a significant body of theoretical and empirical work has examined links between various aspects of religion/spirituality and mental health. This work identifies a number of mechanisms by which religious factors may contribute to greater psychological well-being, including the following: (a) reduced exposure to stressful events and conditions; (b) better health behaviors and lifestyles that are conducive to well-being (e.g., regulation of alcohol, tobacco, and drug use); (c) spiritual coping resources and skills for dealing with problems; (d) congregational social networks and support systems; (e) religious and spiritual meaning systems and interpretive frameworks; (f) character strengths and virtues such as hope and optimism, forgiveness, gratitude, humility, and others (Pargament et al., 2000; Koenig et al. 2001, 2012; George et al., 2002; Krause 2008, 2021; Ellison & Henderson, 2011; Park, 2017).

It is now well-established that religion is a multidimensional construct (e.g., Glock and Stark, 1968; Levin et al., 1995), a central insight to the development of scholarship on religion and mental health (Hill & Pargament, 2003; Idler et al., 2003; Krause, 2021). Over the years, scholars pursuing these issues have focused on several classes of religious predictors, including (a) affiliation and organizational practices, such as attendance at services and other congregational activities (Krause, 2008; Idler et al., 2009); (b) non-organizational practices such as prayer, scripture reading, meditation, and others (Sharp, 2010; Upenieks, 2022); (c) religious orientations and motivations (i.e., intrinsic vs. extrinsic vs. quest) and religious salience and identity (e.g., Hill and Pargament 2003); (d) religious functions that may foster health and well-being, such as spiritual coping and social support (Pargament, 1997; Pargament et al., 2000; Krause, 2008); and (e) specific religious cognitions, especially beliefs about God or a Higher Power and the afterlife (Flannelly, 2017; Park, 2017).

What do we know about the association between individual religiosity and suicidality? Over the years, a voluminous literature has tentatively concluded that religious involvement appears to protect against suicidality. Still, these patterns are complex and vary across religious traditions and sociocultural contexts (Wu et al., 2015; O’Reilly & Rosato, 2015; Lawrence et al., 2016; Gearing & Alonzo, 2018; Jacob et al., 2018; Stack & Laubepin, 2019; Eskin et al., 2020; Poorolajal et al., 2022).

Several notable findings have emerged from recent US-based studies. First, there is tentative evidence that persons who attend worship services regularly are less prone to suicidality. Two high-quality studies support this conclusion. Analyzing data among large prospective cohorts of US nurses and healthcare professionals, these studies found that regular religious attendance predicted lower suicide risk (Chen et al., 2016; VanderWeele et al., 2016). Among US nurses, there was also evidence that social integration, depression, and alcohol consumption partly mediated the protective effects of occasional attendance but not necessarily those of regular attendance (VanderWeele et al., 2016).

Nevertheless, a meta-analysis of 89 studies found that religious attendance generally does not protect against suicidality after accounting for social support, which may confound the effects of religious attendance (Lawrence et al., 2016). However, this same analysis found that attendance did appear to protect against attempted and completed suicides. Likewise, cross-sectional analyses among a representative sample of Canadians found that while social support mediated links between worship attendance and suicidality, attendance still appeared to protect against suicide attempts regardless of social support (Rasic et al. 2009, 2011). Another national longitudinal study of US adolescents found that regular worship attendance correlated inversely with suicidality (Nkansah-Amankra et al., 2012). Kim and colleagues (2021) found that rural US college students who scored higher on religious attendance and private religious practices tended to report lower levels of suicidality, due in part to having fewer depressive symptoms. Overall, in a systematic review of 81 articles published worldwide between 2008-2017, Gearing and Alonzo (2018) concluded that religiosity—typically gauged via self-reported frequency of attendance—is often, but not always, protective against various aspects of suicidality.

In addition to the extensive focus on religious affiliation and attendance, a number of suicide studies have employed other indicators of religiosity. Such investigations have used a wide array of religious measures, ranging from subjective religiosity to religious coping practices, intrinsic religious motivation, and others. They have also based their conclusions on diverse types of samples, including population-based and clinical, among others. Perhaps unsurprisingly, these studies have also reported inconsistent results (Walker & Bishop, 2005; Rasic et al., 2009, 2011; Assari et al., 2012; Rushing et al., 2013; Hirsch et al., 2014; Currier et al., 2017, 2018; Lester & Walker, 2017; Fanegan et al., 2022).

Two noteworthy studies used data from the National Survey of American Life (NSAL), a national probability sample of African Americans and Caribbean-born Blacks. We consider these studies notable because, like our study, the authors included various religious indicators. First, Chatters et al. (2011) focused on the characteristics of church-based social networks. Results were mixed. They found that subjective closeness to church members was inversely associated with suicidality. But frequency of interaction with church members was positively associated with suicide attempts. Moreover, several other predictors were unrelated to lifetime prevalence of suicidality, including emotional support from church members, frequency of attendance, and negative interactions within the church.

In another NSAL study, Taylor et al. (2011) widened the scope of inquiry with additional dimensions of religion. Again, the results were complex. For both African Americans and Caribbean-born Blacks, looking to God for strength, comfort, and guidance appeared to protect against lifetime suicidality. However, relying on prayer for stress-coping predicted increased suicidality among both groups, and attempted suicides for Caribbean-born Blacks. Other patterns added even more complexity. For example, among Caribbean-born Blacks only, subjective religiosity was inversely associated with lifetime suicidality.

Although most work in this area has investigated hypothetical benefits of religion, investigators have recently turned their attention to possible harms. These studies have tested links between “spiritual struggles” and suicidality, especially troubled relationships with God and chronic religious doubting. Several of these studies have focused on unique subpopulations, including US war veterans and persons recovering from substance use disorders (Currier et al., 2017, 2020; Raines et al., 2022). In a study based on a probability sample of US adults during COVID-19, Upenieks (2022) documented a positive association between spiritual struggles and suicidality. However, like earlier work by Currier et al. (2017, 2020), Upenieks (2022) also found that other positive dimensions of religion appeared to mitigate the psychic harms of spiritual struggles.

To be specific, Currier et al. (2020) found that links between spiritual struggles and suicidality depended on how people perceived the nature of God, among other factors. Upenieks (2022) also found that attendance and a sense of divine control moderated the association between spiritual struggles and suicidality. Schieman and colleagues (2006) define the sense of divine control as the general belief that “God exerts a commanding authority over the course and direction of one’s life” (529). The notion of divine control has loomed large in studies of religion and well-being, exhibiting patterns consistent with salutary main effects and stress-buffering effects on depression, distress, and other mental health outcomes (Schieman et al. 2005, 2006, 2010; DeAngelis & Ellison, 2017; DeAngelis, 2018; Upenieks et al., 2022). However, prior to Upenieks (2022), divine control beliefs have been largely absent from studies of suicidality.

In sum, a long tradition of theory and research has linked multiple domains of religion and spirituality with suicidality, including suicidal ideation, planning, attempts, and completion. Although study findings are inconsistent at times—due perhaps to differences in the religious domains considered, and the various populations considered—overall, evidence suggests more religiousness associates with less suicidality. However, recent patterns and trends in religion in the United States raise important questions about this topic that have been neglected to date.

First and foremost, levels of religious affiliation, practice, and (to a lesser extent) belief have been declining among US adults for some time. Much of the scholarly and public attention to this issue has focused on religious affiliation and identification and the reasons for declines in these areas (e.g., Hout & Fischer, 2014). Despite evidence of broad, albeit uneven, religious retraction (Schwadel, 2011; Voas & Chaves, 2016), other religious domains have received less attention, such as non-organizational participation, identity, doctrinal assent, and belief. In particular, the proportion of American adults reporting irreligion appears to be on the rise. This group includes those who report an utter lack of prayer, subjective religious identity, religious coping, access to congregational support systems, or belief in God and the afterlife.

For example, several patterns have become apparent in NORC’s General Social Survey over the last two decades. First, the proportion of US adults who say they “never” pray has increased from less than 1% in 2000 to nearly 20% in 2018 (Davis et al. 2000-2010; Davern et al. 2016-2020). During this same period, the percentage who attend religious services “never” or “less than once per year” has also increased from 28% to 36% (Davis et al., 2000-2010; Davern et al., 2016-2020). Persons who deny an afterlife have also grown from 16% to 23% over this period (Davis et al. 2000-2010; Davern et al. 2016-2020). Finally, while only 6% of US adults espoused atheist or agnostic views regarding the existence of God in 2000, this figure increased to nearly 16% by 2018 (Davis et al. 2000-2010; Davern et al. 2016-2020). Taken together, these trends indicate that well-documented patterns of religious organizational disaffiliation may be followed by declines in religious/spiritual beliefs or behaviors that were once thought to be impervious to changes, such as private prayer and beliefs in God or the afterlife.

These trends raise important and largely neglected questions concerning the associations between irreligion and suicidality. Given the current upward trends in suicide, is it possible that declines in religiosity are at least partially contributing? And given the growing proportion of US adults who eschew religion altogether, how might various dimensions of irreligion associate with suicidality? In what follows, we examine these questions empirically with data from a Nashville-based population sample of Black and White working-age adults. After detailing our methods and results, we discuss the implications of our findings for future research on religion and suicidality, as well as investigations into irreligion and mental health more broadly.

Methods

Data

Data come from the Nashville Stress and Health Study (NSAHS), a cross-sectional probability sample of 1,252 non-Hispanic Black and White women and men aged 22 to 69 who lived in Davidson County, Tennessee, between 2011 and 2014. The NSAHS was designed to assess population stress and health disparities, including mental health and suicidality. The NSAHS sample was collected using multistage stratified sampling techniques. The sampling frame consisted of 2,400 randomly selected households, with 2,065 being contacted to participate in the study. Of those households contacted, roughly 61 percent participated in the study. Trained interviewers conducted the survey and were matched with respondents based on race. Interviews were conducted either at respondents’ homes or at Vanderbilt University. Interviews were computer-assisted and lasted approximately three hours. Respondents were offered $50 to participate in the survey interview (Turner et al., 2017).

Measures

Suicidality.

Suicidality is gauged with three dichotomous measures asking whether respondents ever: (1) seriously thought about suicide, (2) made a plan for suicide, and (3) attempted suicide. Our final measure compares respondents who answer “yes” to either statement with respondents who answer “no” to all statements.

Lack of Religion.

We measure six dimensions of irreligion. First, lack of religious attendance is measured with the question, “Which of the following best describes how often you attend services at a church/temple/synagogue/mosque?” Second, lack of religious/spiritual coping is measured by asking, “How often do you turn to your religion or your spiritual beliefs to help you deal with your daily problems?” Third, lack of religious social support is measured by asking, “How often do people in your church (place of worship) help you out?” Fourth, lack of prayer is measured with the question, “About how often do you pray?” For these first four items, respondents who answer “never” are compared to everyone else.

Fifth, we measure lack of religious identity with the question, “How religious are you?” Respondents who answer “not at all religious” are compared to everyone else. Sixth and finally, we measure the lack of belief in divine control with the following four items (Schieman et al., 2005; Schieman et al., 2006): (1) “I decide what to do without relying on God”; (2) “When good or bad things happen, I see it as part of God’s plan for me”; (3) “God has decided what my life shall be”; and (4) “I depend on God for help and guidance” (α = .83). Respondents who “strongly agree” with item 1 and “strongly disagree” with items 2 – 4 are compared to everyone else.

Covariates.

We also include covariates for age (in years), gender (female = 1, male = 0), race (Black = 1, White = 0), marital status (1 = married, 0 = not married), education (1 = less than high school, 0 = high school or more), employment status (1 = full-time, 0 = not full-time), and household income (ordinal, 0 = $5000 or less to 15 = $135,000 and higher).

Analytic Strategy

We use multivariable logistic regression techniques to predict odds of suicidality. Models 1 through 6 estimate odds of suicidality for those who: never attend religious service (Model 1); never use religion to cope (Model 2); never rely on religious support (Model 3); never pray (Model 4); do not identify as religious (Model 5); and do not believe in divine control (Model 6). Model 7 includes all religious variables simultaneously. Model 8 adds covariates.

Multiple imputation by chained equations (30 iterations) is used to replace missing values (White et al., 2011). The following variables had missing values: religious support (n = 1), prayer (n = 2), perceived divine control (n = 19), and household income (n = 31). All models adjust for probability weighting and cluster sampling by block group to allow generalizability to the Davidson County population of non-Hispanic Black and White adults. All analyses are conducted using STATA 14 (StataCorp, 2009).

Results

Table 1 provides unweighted descriptive statistics of study variables. Regarding suicidality, a little over 20% of the sample reports ever having considered or attempted suicide. Looking at religious attendance, religious coping, prayer, and belief in divine control, less than a fifth of the sample reported not engaging or believing in either religious dimension. However, larger percentages of the sample report they never receive religious support (54%) and do not identify as religious (25%). Additional descriptive statistics are presented in Table 1.

Table 1.

Descriptive Statistics: NSAHS, 2011-2014.

	Mean/ Proportion	SD	Range	N
Focal Variables
Suicidality	.213	–	0–1	1,252
Never attends religious services	.149	–	0–1	1,252
Never uses religious coping	.167	–	0–1	1,252
Never receives religious support	.536	–	0–1	1,251
Never prays	.163	–	0–1	1,250
Does not identify as religious	.248	–	0–1	1,252
No belief in divine control	.121	–	0–1	1,233
Covariates
Age (years)	46.04	11.52	22–69
Female (vs. male)	.530	–	0–1	1,252
Black (vs. White)	.501	–	0–1	1,252
Married (vs. not married)	.447	–	0–1	1,252
Less than high school education (vs. more)	.137	–	0–1	1,252
Employed full time (vs. part-/non-employed)	.597	–	0–1	1,252
Household income (ordinal)	7.98	4.16	0–15	1,252

Notes: Statistics are derived from non-imputed and unweighted data. NSAHS = Nashville Stress and Health Study. SD = Standard deviation.

Table 2 displays the odds ratios (OR) of lifetime suicidality. Only lack of belief in divine control (Model 6) is associated with significantly higher odds of suicidality (OR = 1.71; p ≤ .05). This coefficient indicates that respondents with no belief in divine control have 71% greater odds of reporting lifetime suicidality, relative to their peers who hold at least some belief in divine control. Model 7 shows that this pattern holds even after accounting for other dimensions of irreligion (OR = 2.05; p ≤ .05). Finally, Model 8 confirms that this pattern also holds after accounting for additional covariates (OR = 2.51; p ≤ .01).

Table 2.

Odds Ratios from Binary Logistic Regression Estimates of Suicidality: NSAHS, 2011-2014 (n = 1,252).

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Focal Variables
No religious attendance	1.47 (0.98; 2.22)						1.36 (0.79; 2.31)	1.20 (0.70; 2.07)
No religious coping		0.97 (0.62; 1.52)					0.59 (0.34; 1.01)	0.59 (0.34; 1.02)
No religious support			1.01 (0.75; 1.38)				0.87 (0.51; 1.05)	0.81 (0.56; 1.17)
No prayer				1.27 (0.81; 1.98)			0.89 (0.45; 1.81)	0.95 (0.43; 2.09)
No religious identity					1.30 (0.94; 1.80)		1.21 (0.77; 1.92)	1.21 (0.78; 1.86)
No divine control						1.71 * (1.09; 2.65)	2.05 * (1.09; 3.80)	2.51 ** (1.28; 4.91)
Covariates
Age								0.99
Female								1.36
Black								0.43 ***
Married								0.85
Less than high school								1.90 *
Employed full time								0.79
Household income								0.89 ***

Notes: Odds ratios with 95% confidence intervals for religion variables in parentheses. Estimates average over 30 imputed datasets, and adjust for post-stratification weighting and clustering by block group. NSAHS = Nashville Stress and Health Study.

^*p < 0.05

^**p < 0.01

^***p < 0.001 (two-tailed).

Other predictors are significantly associated with odds of suicidality. Compared to their White peers, Black respondents report 57% lower odds of suicidality (OR = 0.43; p ≤ .001).¹ Higher levels of household income are also associated with lower odds of suicidality (OR = 0.89; p ≤ .001). Lastly, having less than a high school education is associated with a 90% higher likelihood of suicidality (OR = 1.90; p ≤ .05).

Discussion and Conclusion

A long research tradition has investigated linkages between religion and suicide in the United States and elsewhere. Although early work in the social sciences focused on ecological analyses, which correlated the suicide rates of areal units with their religious compositions, much of the contemporary literature has centered on the associations between individual religiosity and suicidality in population-based probability samples (Lawrence et al., 2016; Gearing & Alonzo, 2018; Fanegan et al., 2022). Focusing on suicidality, our study makes an original contribution to the religion-suicide literature in at least two notable ways.

First, whereas most studies have concentrated primarily on organizational aspects of religious involvement, including religious affiliation and frequency of attendance at services, we examined multiple religious domains. These included: (a) organizational participation (i.e., attendance at services); (b) non-organizational participation (i.e., prayer); (c) religious identity (i.e., the extent to which an individual believes they are a religious person); (d) religious functions (i.e., congregational social support and religious coping practices); and (e) religious cognitions (i.e., the belief in divine control). Second, in contrast to studies that estimate religious effects on suicidality as linear or monotonic in form, we focus on a very specific segment of the religious public: the growing portion of individuals who entirely eschew religion. In our sample, these were respondents who never attended religious services, never prayed, did not identify as religious, never received support from church members, never used religion to cope with stress, and fully rejected the notion that God played any authoritative or directive role in their lives.

Our analyses of data from Vanderbilt University’s Nashville Stress and Health Study revealed two critical patterns. First, with one exception, persons who were irreligious according to the criteria outlined above did not report any greater incidence of suicidality in their lifetimes than their peers who were at least somewhat religiously involved. Thus, although suicide rates are rising among key segments of the US population, there appears to be little correlation with most dimensions of irreligion, at least according to the indicators in our study. Second, however, there was one glaring exception to our mostly null results: Persons who rejected any notion that God influenced the course of their lives were substantially more likely than others to have experienced suicidality. This result was consistent after controlling for other dimensions of irreligion and sociodemographic characteristics.

Taken together, these findings raise two possibilities. First, declines in organizational religious involvement, religious identity, and other forms of practice may have little bearing on suicidality. Second, the erosion of belief in a personal deity may have more concerning implications for suicidality. This second point could be critical, as much of the current discussion regarding secularization in the United States has emphasized declining rates of religious affiliation, attendance, and other public religious indicators.

It is useful to reflect on our findings from the standpoint of prominent theoretical works on religion and suicidality, especially those derived from Durkheim ([1897] 1952). One important strand of work in this area is based on Berger’s (1967) notion of religious “plausibility structures,” or cohesive systems of meaning and interpretation. According to Berger (1967: 40), plausibility structures involve the complex interplay of personal beliefs, collective religious experiences, and networks of coreligionists, which make beliefs seem reasonable. Moreover, Berger argued that the legitimating power of religion may be especially significant when confronting “marginal situations in which the reality of everyday life is put in question” (1967: 43-44). That is, religious meaning systems become most salient in contexts of major traumatic events.

A number of scholars over the years have applied Berger’s insights to empirical studies of religion and suicidality, anomie, and related outcomes (e.g., Bjarnason, 1998; Brashears, 2010). In the context of our findings concerning irreligion and suicidality, these theoretical touchstones raise several significant questions. For example, given that the lack of collective religious participation (i.e., attendance, support) is unrelated to suicidality, have the irreligious developed robust secular rituals and networks of like-minded persons from which to cultivate meaning and purpose? How does this happen? Do links between disbelief in divine control and suicidality vary between persons who are loosely attached to their faiths, versus those who are committed atheists or agnostics? Distinctions like these have been suggested by several recent commentators (e.g., Hwang et al., 2011; Speed & Hwang, 2019). Although we cannot address these and other relevant questions here, they warrant careful attention from future investigators given the apparent steady rise in non-belief.

The association between disbelief in divine control and suicidality adds to a growing literature on the role of religious cognitions in mental health, much of which is fairly recent. For some years, the religion-health field largely neglected consideration of religious beliefs or meaning systems, focusing instead on social practices and functions that appeared relevant for various religious groups and traditions. In retrospect, it appears the neglect of religious cognitions was an oversight. A burgeoning literature shows that perceptions of God and the afterlife are associated with mental health and psychological well-being (Flannelly et al., 2006; 2008; Silton et al., 2014; Flannelly, 2017; Park, 2017; Upenieks et al., 2022). Such beliefs appear to have direct mental health benefits, and also moderate associations between religious practices, psychosocial stressors, and mental health outcomes (Ellison et al., 2014; DeAngelis & Ellison, 2017; DeAngelis, 2018; Bradshaw & Kent, 2018; Upenieks, 2022).

In recent work, the sense of divine control has loomed especially large. Divine control beliefs have been linked with an array of outcomes like self-concept (Schieman et al., 2005; Schieman et al., 2010; DeAngelis, 2018), positive reappraisal coping (DeAngelis & Ellison, 2017), psychological distress (Schieman et al., 2006; Upenieks et al., 2022), and stress-related biomarkers (DeAngelis et al., 2023). Our study extends this list of outcomes to suicidality. We also highlight the unique implications of completely lacking belief in divine control.

Several factors might explain the link between disbelief in divine control and suicidality. For example, disbelievers may have suffered traumatic stress in childhood that both undermined their mental health and belief in God (i.e., confounding). This group may also engage in certain maladaptive coping behaviors as adults that stem from disbelief in God, and ultimately contribute to suicidality (i.e., mediation). Our goal in this study was to establish a baseline association between irreligion and suicidality. Future studies can investigate potential sources of confounding, mediation, and moderation in the associations we uncovered here.

Regarding moderation, for example, some readers may be curious about ethnoracial variation in associations between irreligion and suicidality. Indeed, African Americans are more religious than their White peers by almost all indicators and especially disinclined toward irreligion (e.g., Chatters et al., 2009; DeAngelis et al., 2023), or the complete lack of practice or belief (Brown et al., 2013). Moreover, the role of religion in the health of African Americans has received considerable research coverage, (e.g., Ellison et al., 2010; Nguyen, 2020; DeAngelis et al., 2023). African Americans also tend to show lower rates of suicidality than Whites, although signs suggest this trend is slowly changing (Bommersbach et al., 2022; Curtin, 2022; Garnett et al., 2022). In ancillary analyses of NSAHS data (not shown, but available upon request), we found that African Americans indeed reported lower levels of most forms of irreligion, in addition to suicidality. In these data, however, associations between irreligion and suicidality did not vary by race (not shown, but available upon request). These patterns raise the intriguing possibility that (ir)religion may underlie differences in suicidality between African Americans and Whites, a topic worthy of careful investigation in future studies.

Although our study makes a novel contribution to the religion-health literature, we acknowledge several key limitations. First, the NSAHS data are cross-sectional and preclude temporal ordering among variables. This is especially important because we use a lifetime measure of suicidality, owing to extremely small cell sizes for 12-month measures. To be sure, other high-quality studies (e.g., Chatters et al., 2011; Taylor et al., 2011) have also relied upon lifetime measures of suicidality; nevertheless, this undercuts our ability to make causal arguments regarding the temporal association between (ir)religion and suicidality.

Second, although the NSAHS has a reasonable variety of religious indicators, there is no measure of religious affiliation. Given the aforementioned increases in religious non-affiliation, this may have been useful to know. We could have also benefited from knowing whether respondents identified as atheists, agnostics, or unaffiliated theists (Hwang et al., 2011; Speed & Hwang, 2019). In the few studies that have explored the relevance of this distinction for the religion-health connection, self-described atheists and agnostics tended to fare better than unaffiliated believers (Baker et al., 2018). This is an area that is certainly ripe for future investigation.

Third, the NSAHS sampled working-age non-Hispanic Black and White adults of Davidson County, TN (i.e., urban Nashville). Generalizing our findings beyond this population would thus be inappropriate. It is also important to consider the distinctive character of the NSAHS sample. Like much of the South, the Nashville area has long been dominated by Baptists and other evangelical Protestants, including conservative non-denominational evangelicals, and levels of religious practice and belief have traditionally been higher than the national average. Previous theoretical and empirical works have described the role of religion in the South as a “semi-involuntary institution,” particularly for Black residents, but perhaps extending to the wider regional population as well (Ellison & Sherkat, 1995; Hunt & Hunt, 1999; Hunt & Hunt, 2001; Sherkat & Cunningham, 1998). This insight has rarely been brought into the religion-health literature, but a handful of studies have suggested that such a distinctive religious culture may make it more difficult to detect links between religious factors and health outcomes in the South (Ellison, 1995; DeAngelis et al., 2023). If anything, this underscores the need for replicating and extending our findings using other diverse samples.

Limitations notwithstanding, our study provides a fresh perspective on the links between religious factors and suicidality. We considered multiple religious and spiritual domains, and focused on irreligion as a risk factor for suicidality. We believe that studying the links between multiple dimensions of irreligion—including diverse identities of non-believers like atheists, agnostics, and unaffiliated theists—and suicidality, or other aspects of mental health, is an area that is overdue for further exploration. Additional research along the lines sketched above will hopefully shed fresh light on this neglected topic.

Biographies

Colton L. Daniels is an assistant professor in the Department of Criminal Justice and Criminology at St. Mary’s University. His current research focuses on the impact of personality traits on the prevalence of intimate partner violence, and how religious beliefs and practices can dampen the effect of risk-taking behaviors.

Christopher G. Ellison is Dean's Distinguished Professor of Sociology at the University of Texas at San Antonio. His longstanding research interests include: (a) the role of religion in shaping mental and physical health; (b) religious variations in family life; and (c) the influence of religion on politics and public opinion. His research also focuses on gun culture and gun policy issues in the United States.

Reed DeAngelis is a postdoctoral researcher at the Center for the Study of Aging and Human Development in the Duke University School of Medicine. He researches health inequities, stress-coping, and aging.

Katherine Klee recently completed her M.S. in Sociology at the University of Texas at San Antonio. Her core research interests include suicidality, with particular attention to the problem of adolescent suicide.