Cumulative Stressors and Adolescent Substance Use: A Review of 21st-Century Literature

Stressors and Substance Use

Research has identified myriad factors that affect adolescent substance use, including a family history of use or SUDs, poor family relationships, peer associations, childhood behavioral problems, low self-control, and risk-taking propensities (Chassin et al., 2013; Gray & Squeglia, 2018; Hoffmann, 2020; Leung et al., 2014; Van Ryzin et al., 2012). An antecedent that has been explored for several decades is stressful or adverse experiences—also referred to as stressors—which produce reactions that disrupt one’s homeostasis, resulting in physical and psychological adaptions (Majewska et al., 2002). Whereas stress is an individual’s physiological and emotional response to pressures from the environment (Selye, 1976), stressors are specific types of pressures that are believed to increase stress. Short-term reactions to stressors, such as the fight or flight response, are designed to relieve stress and facilitate a return to homeostasis. According to several studies, one reaction or adaptation to stressors among adolescents is substance use (Chassin et al., 2013; Shiffler et al., 2020).

Much of the research on adverse experiences addresses them singly or in a limited manner. For instance, studies have documented the effects of natural disasters, family disruption, unemployment, victimization, discrimination, parental maltreatment and neglect, or living in poverty or conflict-ridden communities on substance use (e.g., Arkes, 2013; Cambron et al., 2020; Danielson et al., 2017; Wright et al., 2013). But the accumulation of these experiences—cumulative stressors—especially in a limited time frame, can be especially detrimental as one’s ability to adapt and return to a state of homeostasis is hindered, resulting in maladaptive responses such as social withdrawal, physical illness, or mental health disorders (Kort-Butler, 2020; McEwen, 2004; Shiffler et al., 2020). Under these circumstances, substance use furnishes a way to assuage the effects of stressful experiences.

The most common explanation for the link between stressors and substance use is the self-medication hypothesis, which posits that substance use provides a way to diminish the physical and emotional toll of stressful or painful experiences (Grunberg et al., 2011). In particular, substance use alleviates the negative emotions caused by cumulative stressors (Wills et al., 2017). Generally, intoxicating substances furnish short-term relief from stressors. Some intoxicating substances may also boost “extinction learning” and help one forget negative experiences (Grunberg et al., 2011, p. 294).

Cumulative stressors also affect neurological pathways. Exposure to stressors, for example, disturbs cortisol regulation, striatal functioning, DNA methylation, limbic system activation, and other neurological functions, which, in turn, are linked to hostility, impulsivity, and substance use during adolescence and adulthood (Enoch, 2011; Gordon, 2002; Somaini et al., 2011; Tervo-Clemens et al., 2020; Walters & Kosten, 2019). Moreover, if substance use transitions into a SUD, additional stressors may increase emotional distress, resulting in a reciprocal association between stressors and substance use (Brady & Sinha, 2005). Stressors may also increase anhedonia, the inability to feel pleasure, thus leading individuals to seek pleasurable sensations through intoxication (Sussman & Leventhal, 2014). Finally, cumulative stressors are related to cognitive impairments in general, which are associated with a higher risk of substance use and SUDs (Edalati & Krank, 2016).

The neurological impairment model focuses on the indirect effects of cumulative stressors on substance use. Other approaches emphasize conditional effects. The stress process, general strain, or stress-buffering model, for instance, asserts that adverse experiences lead to substance use when conventional coping resources are weak or nonexistent (Agnew & Brezina, 2019; Pearlin & Bierman, 2013). When, say, family relations or social support systems are poor, those who experience cumulative stressors are likely to turn to substances to alleviate their negative psychological states. But, when coping resources are strong, one’s ability to adapt positively to adverse circumstances is enhanced. Similarly, cognitive resources, such as high self-control and emotional maturity, or certain genotypes might attenuate the effects of cumulative stressors on substance use (Blomeyer et al., 2008; Wills et al., 2011).

Cumulative Stressors in Childhood and Adolescence

Experiencing cumulative stressors early in the life course is especially detrimental and may increase the likelihood of substance use and SUDs in adolescence and adulthood (Gordon, 2002). Two research literatures support this assertion. First, studies have demonstrated a link between ACEs and substance use in adolescence (Benjet et al., 2013; Shiffler et al., 2020). Second, research that has administered stressful life events inventories to adolescents suggests a positive association with the prevalence and frequency of substance use and SUDs (e.g., Hyman & Sinha, 2009; Roberts et al., 2018). Some studies have also identified a step function, with youth increasingly likely to report substance use as they experience more events or as stressors accumulate (e.g., Cisler et al., 2011; Dube et al., 2006). Many empirical studies and a few summary chapters outlining these associations exist, but a careful review of the literature on cumulative stressors and adolescent substance use has not yet been conducted. Similarly, even though studies have examined factors that moderate or mediate the effects of cumulative stressors on adolescent substance use (e.g., Cooke et al., 2015; Wills et al., 2013), there has been no effort to systematize these factors.

Our objectives in this article are therefore to (1) summarize the literature on cumulative stressors and substance use to determine the consistency of the association across studies; (2) review and organize factors that moderate or mediate the effects of cumulative stressors; and, based on limitations in the literature (3) recommend a path forward for researchers studying the impact of cumulative stressors on adolescent substance use.

Search Procedures

Because we were interested in two often distinct literatures—ACEs and adolescent life events—we used a two-pronged approach to search the research literature for relevant publications. The search terms that guided the review were based on previous assessments of adolescent substance use, ACEs, life events, and cumulative stressors (Arcadepani et al., 2019; Hatch & Dohrenwend, 2007; Heerde & Hemphill, 2016; Schwarzer & Luszczynska, 2013) as well as the research team’s experience investigating these topics (e.g., Hoffmann, 2016; Jones & Pierce, 2020). We employed the search terms and databases listed in Table 1 to explore the two bodies of literature and assessed the reference sections of included studies and narrative reviews to identify additional publications. Google Scholar’s “Cited by” function was also utilized to pinpoint relevant studies that may not have been listed in the searches.

OPEN IN VIEWER

Table 1.

Literature Search Strategy: Keywords and Databases (2000–2020).

Keywords

Adolescence: Life events or Events [adverse, cumulative, family, negative, personal, stressful, traumatic]; Stress or Stressors [adverse, cumulative, family, general, life, negative, personal, traumatic]; Strain [general, life, theory] Childhood: Adverse childhood experiences [cumulative, traumatic]; childhood adversity [cumulative, traumatic]; childhood adverse events [cumulative, traumatic]; ACEs Drug [abuse, dependence, disorder, use]; Substance [abuse, dependence, disorder, use]; Alcohol (abuse, binge, dependence, disorder, heavy, use]; Alcoholism; Alcoholic; Drunk; Drinking [binge, heavy]; Aerosol sniffing; Cigarette; Club drugs; Cocaine; Ecstasy; Hallucinogens; Heroin; Huffing; Inhalant; Nicotine; Marijuana; Meth[amphetamine]; Nonprescription [drug, medicine]; Opiates; Opioids; Pharmaceuticals; Psychotropics; Solvent; Smoking; Stimulants; Tobacco; Vape; Vaping Adolescent; Teen; Teenage; Teenager; Young people; Youth

Databases

Scopus; ProQuest; Google Scholar; Web of Science; EBSCO; MEDLINE

Note. ACE = adverse childhood experience.

We limited the publications using the following criteria that are consistent with the aims of our review and with commonly accepted practices for narrative and systematic reviews (Rodgers et al., 2009; Templier & Paré, 2015):

The study’s results were published between 2000 and 2020.

We did not deliberately omit experimental studies, though the review did not turn up any that were within the outlined parameters. The search was completed in May 2020.

Article Selection

Figure 1 depicts the results of the database searches and the article reviews. One member of the research team was responsible for the review of the adolescent life events literature and initially identified 3,605 articles. After screening the titles and abstracts of each, 3,032 were excluded, leaving 573 for a full-text review. Each article was then vetted by the team member using the eligibility criteria, which resulted in 90 publications selected for inclusion in the final review. After forward and backward selection based on the reference lists of these articles, four additional articles were identified, yielding a final list of 94 articles. The second member of the research team utilized the same process to assess the ACEs literature, initially identifying 2,872 articles. After screening the titles and abstracts of each, 2,742 articles were excluded, leaving 130 for full-text review. Once each article was evaluated using the eligibility criteria, 15 publications were selected for inclusion in the final review. Forward and backward selection did not identify any additional publications. The team member who was initially responsible for the life events review then independently assessed the ACEs articles, and the team member who was initially responsible for the ACEs review independently evaluated a random subset (n = 25) of the life events articles. No discrepancies were identified. The analysis of the data was guided by Rodgers et al.’s (2009) recommendations regarding narrative syntheses. A complete list of the final set of articles is available in the Online Appendix.

OPEN IN VIEWER

Figure 1.

Literature search flowchart: Article selection procedures.

The corpus of studies utilized 80 distinct samples mostly from the United States but also from Australia (one), Brazil (one), Canada (three), China (three), western Europe (two), Finland (one), Germany (two), Iceland (one), Israel (one), Italy (one), Mexico (three), the Netherlands (one), New Zealand (one), the Russian Federation (one), South Africa (one), Spain (one), and Turkey (one). Four of the studies were not in English, with three in Spanish and one in Chinese. These were translated into English using transcription software. All of the studies used cross-sectional or prospective longitudinal data, with none relying solely on retrospective data regarding substance use.

Summary information from the final group of studies appears in Table 2, along with a sample of references. Forty-one studies analyzed cross-sectional data and 68 longitudinal data. Seventy-five studies took place in the United States, 32 outside the United States, and two included a sample of the United States and Canadian youth. Fourteen studies used nationally representative data from the United States (12) or Italy (two). The remainder relied on school (48) or community (47) samples. Two of the school samples and five of the community samples were from American Indian communities in the United States or Canada.

OPEN IN VIEWER

Table 2.

Summary Information From the Reviewed Studies.

Characteristics of Study	Examples of Studies
Cumulative stressors measure
ACEs (15)	Benjet et al. (2013), Brown & Shillington (2017), Carliner et al. (2016), and Dube et al. (2006)
Adolescent life events inventory (94)	Aseltine et al. (2000), Bares et al. (2020), Blomeyer et al. (2008), Brook et al. (2006), Cheadle & Whitbeck (2011), Cisler et al. (2011), Drapela (2006), Innamorati & Manglio (2015), Lee et al. (2010), Li et al. (2003), Liu (2003), Kühn et al. (2020), Pampel et al. (2015), Rafnsson et al. (2006), Spicer et al. (2003), Wills et al. (2001, 2003, 2011, 2013), Windle (2004), and S. Wu et al. (2020)
Data type
Cross sectional (41)	Baldwin et al. (2011), Brook et al. (2006), Calvete & Estévez (2009), Carliner et al. (2016), Castillo et al. (2018), Dawson et al. (2007), Eitle (2002), Li et al. (2003), Forster et al. (2017), Liu (2003), Unger et al. (2001), Villegas-Pantoja et al. (2014), and S. Wu et al. (2020)
Longitudinal (68)	Aseltine et al. (2000), Bares et al. (2020), Brown & Shillington (2017), Byrne & Mazanov (2003), Charles et al. (2017), Cheadle & Whitbeck (2011), Cheney et al. (2018), Cisler et al. (2011), Doom et al. (2016), Drapela (2006), Elsayed et al. (2018), Hoffmann et al. (2000), Koval et al. (2000), Kühn et al. (2020), Mazanov & Byrne (2008), Pampel et al. (2015), Spicer et al. (2003), Wills et al. (2001, 2003, 2013), and Windle & Wiesner (2004)
Country/region
United States (76)	Aseltine et al. (2000), Baldwin et al. (2011), Bares et al. (2020), Blomeyer et al. (2008), Carliner et al. (2016), Charles et al. (2015, 2017), Cheney et al. (2018), Cisler et al. (2011), Drapela (2006), Eitle (2002), Makini et al. (2001), Pampel et al. (2015), Roberts et al. (2008), Wills et al. (2001, 2003, 2008, 2011, 2013), Windle (2004), Windle & Wiesner (2004), and S. Wu et al. (2020)
Europe (16)	Blomeyer et al. (2008), Calvete & Estévez (2009), Manglio & Innamorati (2014), Kühn et al. (2020), Pokhrel et al. (2018), Rafnsson et al. (2006), Seo et al. (2019), Shafer et al. (2015), Tay et al. (2019), and Yilmaz et al. (2015)
Mexico or Canada (7)	Barnes et al. (2005), Benjet et al. (2013), Castillo et al. (2018), Cheadle & Hartshorn (2012), Cheadle & Whitbeck (2011), Koval et al. (2000), and Villegas-Pantoja et al. (2014)
Australia or New Zealand (4)	Byrne & Mazanov (2001, 2003), Fergusson & Horwood (2000), and Mazanov & Byrne (2008)
Brazil (1)	Gonçalves et al. (2016)
China (3)	Li et al. (2003), Liu (2003), and Unger et al. (2001)
Israel (1)	Schiff & Fang (2014)
South Africa (1)	Brook et al. (2006)
Substance use outcome
Alcohol or heavy/binge use only (18)	Aseltine et al. (2000), Brown & Shillington (2017), Cheney et al. (2014), Dawson et al. (2007), Dube et al. (2006), King et al. (2009), Makini et al. (2001), Rafnsson et al. (2006), Ramos-Olazagasti et al. (2017), Schmid et al. (2010), and Windle (2004)
Tobacco/cigarette use only (15)	Byrne & Mazanov (2001, 2003), Cheney et al. (2018), Cisler et al. (2011), Li et al. (2003), Liu (2003), and Yilmaz et al. (2015)
Marijuana use only (6)	Barnes et al. (2005), Fite et al. (2015), Hoffmann (2018), Maniglio & Innamorati (2014), Rebellon & Van Gundy (2006), and Shafer et al. (2015)
Polysubstance use (55)	Baldwin et al. (2011), Brockie et al. (2015), Brook et al. (2006), Calvete & Estévez (2009), Carliner et al. (2016), Castillo et al. (2018), Charles et al. (2015), Forster et al. (2017), Pokhrel et al. (2018), Ramírez García et al. (2010), Tay et al. (2019), Walker et al. (2007), and Wills et al. (2011, 2013)
Latent classes (8)	Cheadle & Hartshorn (2012), Cheadle & Whitbeck (2011), Hoffmann (2016), Schiff & Fang (2014), Slocum (2010), Thompson et al. (2020), and S. Wu et al. (2020)
SUD, abuse, or dependence (7)	Benjet et al. (2013), Cheadle & Hartshorn (2012), Cheadle & Whitbeck (2011), Eitle (2002), Kühn et al. (2020)
Moderators
Genetic/neurological (8)	Bares et al. (2020), Blomeyer et al. (2008), Cooke et al. (2015), Elsayed et al. (2018), Goyal et al. (2016), Pampel et al. (2015), Schmid et al. (2010), and Tay et al. (2019)
Intrapersonal (21)	Aseltine et al. (2000), Calvete & Estévez (2009), Cheney et al. (2014), Dawson et al. (2007), Drapela (2006), Leban & Gibson (2020), Lee et al. (2010), Seo et al. (2019), and Wills et al. (2001, 2003, 2008, 2009, 2011)
Interpersonal (15)	Brown & Shillington (2017), Charles et al. (2015), Doom et al. (2016), Hoffmann & Cerbone (2002), and Wills et al. (2008, 2013)
Mediators
Peer substance use (9)	Barnes et al. (2005), Cheadle & Hartshorn (2012), Cheadle & Whitbeck (2011), Walker et al. (2007), Wills et al. (2001, 2004a, 2008, 2009, 2013)
Other interpersonal (2)	Hodson et al. (2006; emotional attachment to others) and Ramírez García et al. (2010; family cohesion)
Intrapersonal (3)	Fishbein et al. (2011; depressive symptoms), Fite et al. (2015; aggression), and Unger et al. (2001; depressive symptoms)

Note. See the Online Appendix for the complete list of studies. The numbers in parentheses in the first column are the number of studies in each category. ACEs = adverse childhood experiences.

General Findings

The 109 studies reported 263 specific results that examined the association between cumulative stressors and adolescent substance use. Most models were estimated with linear regression (95) or binary logistic or probit regression (111). The remainder were assessed using analysis of variance, multinomial logistic, count, growth curve, survival, or regression tree models. The most frequent outcome was a measure of alcohol use, with 55 specific results, followed by marijuana use (29), cigarette use (23), and binge or heavy alcohol use (12). The others were measures that aggregated substances, such as alcohol and marijuana. About 84% of the results met the authors’ criteria for statistical significance (typically p < 0.05), whereas 16% did not.

Standardized regression coefficients were reported in about 15 studies, with a range of approximately .10 and .20 for additive frequency outcomes but between .20 and .35 for outcomes estimated as latent variables. For example, Baldwin et al. (2011) estimated the standardized slope of life events on a latent measure of alcohol, marijuana, and problem use to be .32, after adjusting for a host of covariates (see also Schiff & Fang, 2014; Slocum, 2010).

The models that examined binary outcomes, such as any report of marijuana or other substances in the past month or year, revealed odds ratios that ranged from about 1.02 to 1.86. There was no clear pattern regarding the type of substance examined, though.

The effects of cumulative stressors on substance use differed in cross-sectional and longitudinal models. The median linear regression coefficient in the cross-sectional models was .23, whereas it was .13 in the longitudinal models, which is not surprising given the more distal nature of the association in the latter models. The median sizes of the odds ratios were 1.26 in cross-sectional models and 1.20 in longitudinal models.

Eight studies estimated latent classes of substance use, usually of a longitudinal nature, and examined whether cumulative stressors predicted membership among adolescents. Of the 33 latent classes identified across the studies, 24 were predicted by cumulative stressors and nine were not. S. Wu et al. (2020), for example, determined that the odds of membership in a serious or a moderate substance-using group was 45% or 37% higher with each unit increase in cumulative stressors. On the contrary, Cheadle and Whitbeck (2011) found that cumulative stressors failed to predict membership in an early or mid-adolescent alcohol trajectory group once feelings of anger or friends’ delinquency was included in the model.

Multiple studies, especially in the ACEs literature, provided evidence of a dose–response relationship (e.g., Benjet et al., 2013; Duke, 2018). Duke (2018) determined, for example, that, compared to youth who reported zero ACEs, those reporting two ACEs were about twice as likely, but those reporting four more ACEs were about four times as likely to initiate alcohol or marijuana use.

Substance Use Problems

Seven studies assessed the risk of a SUD, abuse, or dependence. Of these, six reported statistically significant associations between cumulative stressors and one of these problem measures. For instance, Hoffmann and Cerbone (2002) utilized the Composite International Diagnostic Interview to identify youth from a community sample with a drug abuse disorder. They determined that each one-unit increase in stressful life events (range 0–13) was associated with a 4% increase in the risk of drug abuse. The one study that reported a null effect used a conditional inference tree and found that a measure of life events did not discriminate youth with and without an alcohol use disorder. Instead, gender, early sexual experiences, and disorderliness (e.g., low self-control, dishonesty) were the most accurate predictors of a disorder (Kühn et al., 2020).

Null Findings

As mentioned earlier, about 16% of the results failed to show a statistically significant association between cumulative stressors and substance use. Null findings were rarer in models that examined polysubstance use (6%) than in those that analyzed alcohol, tobacco, or marijuana separately (about 20% each). But nonsignificant results were more likely in studies that used data from outside the United States (28%) than from the United States (11%). As discussed in the next section, the associations reported in several studies were diminished with the introduction of variables, such as friends’ substance use, that operated as mediators.

Moderators and Mediators

The neurological impairment and stress process models posit that the effects of cumulative stressors on adolescent substance use are moderated or mediated by genetic/neurological, intrapersonal, and interpersonal factors (Gordon, 2002; Pearlin & Bierman, 2013). Consistent with these perspectives, several of the reviewed articles examined moderators or mediators that fit into each area.

Study Limitations

Directions for Future Research

The limitations discussed in the last section provide guidance for future research (see Table 4). First, studies of cumulative stressors should attend more carefully to the measurement of adverse experiences. Some studies attempted to gauge how distressing experiences actually are by (1) asking respondents their perceptions of events or (b) requesting that “experts” rate the severity of events. But this may not solve the problem of many inventories that combine volitional and nonvolitional events. Additional research is needed to create theoretically appropriate event inventories to administer to adolescents. The ACEs inquiries tend to furnish a clearer focus on events outside of the control of the target children and thus may provide guidance for research with adolescents.

OPEN IN VIEWER

Table 4.

Implications for Research and Practice.

Research Limitations	Future Research Needs
Measurement of cumulative stressors	Examine the psychometric properties of cumulative stressors measures in more detail. Study memory and scoping effects among adolescents. How valid and reliable are parents’ reports of ACEs? Respondent reports?
	Consider measurement strategies that focus on uncontrollable or unexpected experiences; avoid or be circumspect with adverse events that have a volitional component (e.g., parent–child conflict).
Theoretical models—limited construction and testing	Consider theoretical models that integrate genetic/neurological, intrapersonal, and interpersonal factors. Models should include moderators and mediators, such as peer substance use or related factors that channel youth toward substance use.
Cumulative stressors tend to address childhood or adolescent adversity	Build and test models that examine the continuity of cumulative stressors from childhood to adolescence (integrate ACEs and adolescent life events).
Diversity—research has been limited to comparisons between American Indian or African American and White youth or between females and males	Build and test models that compare different racial and ethnic groups. Consider differences in exposure to cumulative stressors across diverse groups based on race/ethnicity, gender identity, family socioeconomic status, and neighborhood of residence.
Implications for practice
Practitioners need to be mindful of cumulative stressors, rather than focusing on discrete events, to better identify youth at risk of substance use.
Develop research-informed methods for targeting at-risk youth by considering factors that lead to substance use when stressors accumulate, such as gender, parental SUDs, lack of family support, peer relations, low self-control, and high-stress sensitivity.

Note. SUD = substance use disorder.

Second, research should be guided by theories of the stress process and integrate genetic, psychological, and social factors as well as examine cumulative stressors from childhood through adolescence. Most studies that address genetic or neurological factors, though they often include demographic control variables, have not considered in a conceptual manner interpersonal and environmental conditions. Research has also not attended to the possible reciprocal effects of substance use and physiological reactions to the stressful experiences (Brady & Sinha, 2005). We have a rich knowledge base of the myriad factors that play a role in the cumulative stressors—adolescent substance use relationship, but there have been too few efforts to construct a conceptual model out of these building blocks. Future efforts to understand adolescent substance use should develop and test models that integrate genetic, intrapersonal, and interpersonal factors and consider the continuity from childhood to adolescence in the cumulative stressors process.

Third, besides a few comparisons between American Indian and White youth, there has been little comparative research on cumulative stressors among other racial and ethnic groups. Many studies have compared substance use among White, African American, Latinx, and Asian youth in general (e.g., Terry-McElrath & Patrick, 2020; L. T. Wu et al., 2011), so it is important that researchers investigate whether the effects of cumulative stressors differ across these groups.

In conclusion, the studies examined in this review demonstrated a consistent, though modest, association between cumulative stressors and adolescent substance use. But the association appeared stronger when certain genetic, intrapersonal, and interpersonal conditions were present. Peer substance use also mediated the effects of cumulative stressors, which implies a socialization mechanism that warrants additional exploration by stress and substance use researchers. By understanding the pathways that lead to cumulative stressors and that link them to substance use and other risk behaviors, responsible adults will have better information to guide prevention and intervention programs designed to help youth lead healthy lifestyles.

Implications for Policy and Practice