Stressful Life Events and Patterns of Polysubstance Use Among U.S. Late Middle-Aged and Older Adults: A Latent Class Analysis

Abstract

The goals of this study were to identify patterns of polysubstance use and their associations with stressful life events among U.S. late middle-aged and older adults and examine whether gender moderates these associations. Adults aged 50 and older (N = 14,738) from the National Epidemiological Survey on Alcohol and Related Conditions-III were included. Latent class analysis was conducted to identify patterns of polysubstance use. Weighted multinomial logistic regression was estimated with a generalized structural equation model. Three different polysubstance use patterns (non-users/low substance users; cannabis and excessive alcohol users; painkiller and sedative/tranquilizer misusers) were identified. Higher levels of stressful life events were associated with patterns of polysubstance use. Gender moderated the association between stressful life events and co-misusing painkillers and sedatives/tranquilizers (p < 0.05). Substance use prevention efforts should consider aging adults’ patterns of polysubstance use and associated stressful life events when designing and implementing gender-specific polysubstance use prevention interventions.

Keywords

polysubstance use latent class analysis late middle-aged adults older adults stressful life events

What this paper adds

• This is the first study that used latent class analysis to identify patterns of polysubstance use among U.S. late middle-aged and older adults on a national scale.

• Stressful life events were associated with an increased likelihood of polysubstance use.

• Gender moderated the association between stressful life events and polysubstance use (i.e., co-misusing painkillers and sedatives/tranquilizers).

Applications of study findings

• Given that the aging population is growing rapidly, substance use prevention efforts should consider aging adults’ patterns of polysubstance use.

• Substance use prevention efforts for preventing polysubstance use, such as early screening and health education, should target aging adults who experience stressful life events as an at-risk population.

• Intervention efforts to reduce polysubstance use should address stressful life events, implement gender-specific interventions, and customize approaches based on the patterns of polysubstance use.

Introduction

Numerous studies conducted over recent decades have shown that one of the greatest public health challenges in the United States (U.S.) is substance use. The increase of substance use among the aging population has been a notable public health concern. For example, one study projected that there will be approximately 72 million older adults misusing substances by 2050 (Youdin, 2019). Substance use has been shown to be associated with drug overdose, attempted suicide, and increased mortality (Schulte & Hser, 2013). Moreover, substance use causes an additional and serious economic burden to families, communities, and governments (Florence et al., 2016). Recently, studies have noted that polysubstance use, referring to using two or more substances concurrently, is an emerging public health concern in the U.S. (Ogbu et al., 2015). For instance, one study found that opioid use commonly overlapped with other substances such as alcohol and sedatives, which has been shown to cause adverse events including overdose deaths (Compton et al., 2021). The additive effects vary by different patterns of polysubstance use. For example, one study reviewed laboratory studies of alcohol and tobacco interactions, suggesting that co-use of alcohol and tobacco was associated with craving and thus increased alcohol and tobacco consumption (Verplaetse & McKee, 2017). In addition, one review study concluded that co-use of opioids and sedatives could result in respiratory depression, which in turn leads to overdose deaths (Hassamal et al., 2016). Given that different patterns of polysubstance use pertain to different additive effects of substances and are associated with various adverse health effects, a better understanding of patterns of polysubstance use is critical for designing tailored prevention efforts.

Polysubstance use has been assessed by summing up multiple substance use indicators such as alcohol and cannabis use statuses (Fendrich et al., 2021). However, such measurement strategies for polysubstance use neither distinguish polysubstance use patterns nor consider the effect of interactions between substances. Moreover, previous polysubstance studies often focused on a single pattern of polysubstance use such as co-misuse of opioids and tranquilizers or co-use of illicit substances and cannabis (Jones et al., 2012; Olthuis et al., 2013). In fact, Timko et al. (2018) conducted a latent class analysis to investigate the patterns of polysubstance use among stimulant users. Yet, the sample characteristic—the mean age of the sample was 32.5 years—limits the study’s generalizability to the aging population. Additionally, the way in which sociodemographics, physical characteristics, and psychological risk factors are associated with each pattern of polysubstance use are not fully explored. A more thorough understanding of polysubstance use patterns and their associations with sociodemographic characteristics and physical and psychological risk factors can better inform interventions to prevent and reduce polysubstance use. It is particularly important to study these relations among older adults given the demographic shifts in our population resulting in increasing numbers of individual in older age groups.

Aging is associated with social, emotional, and physical changes. Age-related changes as well as associated increasing chronic illness may increase vulnerability to substance use and misuse (Moos et al., 2004). In particular, older adults are at great risk of the adverse effects from substance use because of polypharmacy (i.e., simultaneous use of multiple medications) and pharmacokinetic changes (e.g., decreased renal elimination of drugs, decreased hepatic metabolism) (Kalapatapu & Sullivan, 2010). Further, aging adults may be less likely to perceive substance use as a problematic behavior (Wu & Blazer, 2011). In the U.S., the sizeable aging boomer cohort started turning 65 years old in 2011. By 2050, over 85 million, approximately 22%, Americans will be over age 65 (United States Census Bureau, 2018). Given that the projected increase in the aging population and their substance use and misuse, a better understanding of substance use behaviors and associated risk factors among the aging cohort in the U.S. is imperative. However, previous studies have focused primarily on single substance use among the aging cohort or on polysubstance use among the general adult population. One study suggested that using a lifespan perspective approach to investigate substance use prevention is needed (Schulte & Hser, 2013), implying that it is crucial to break down the study population into different age cohorts to investigate (poly)substance use among middle-aged and older adults. Thus, with a better understanding of polysubstance use among aging cohorts, substance use prevention and intervention programs can be better tailored by addressing multiple substance use behaviors.

A stressful life event is defined as an event that results in psychological and physiological stress responses for an individual (Myers et al., 2014). A growing body of evidence has shown that stressful life events are connected to substance use (Verplaetse et al., 2018) and problem drinking behaviors (Brennan & Moos, 1996). Supported by empirical evidence, general strain theory (Agnew, 2001) posits that individuals who experience strains (e.g., stressful life events) may be upset or produce negative emotions, and substance use serves as a way to manage negative emotions (Akins et al., 2010). Yet, an important limitation of existing studies that explored stressful life events and associated substance use is that they focused on single substance use rather than polysubstance use. In addition, previous studies suggest that gender differences exist in the associations between stressful life events and substance use (Sacco et al., 2014). For instance, compared with females, males are more likely to cope with strains through delinquent behaviors such as substance use (Brezina, 2017). Therefore, building on these previous findings, it is important to further investigate whether gender moderates the associations between stressful life events and polysubstance use. To the best of our knowledge, no studies have investigated the moderating role of gender on the associations between stressful life events and polysubstance use patterns, in particularly among middle-aged and older adults. Understanding gender differences can inform gender-specific intervention design in the context of stressful life events and polysubstance use for the aging population.

Taken together, this study aimed to fill literature gaps by investigating polysubstance use among adults aged 50 and older. The purpose of this study was three-fold: (a) to identify patterns of polysubstance users; and (b) to test the associations between past-year stressful life events and polysubstance use patterns; and (c) to investigate the moderating effects of gender on the relationship between stressful life events and polysubstance use patterns in a nationally representative sample of U.S. adults aged 50 and older. This study hypothesized that: (a) adults aged 50 and older who experienced stressful life events would report more polysubstance use; and (b) male adults aged 50 and older who experienced high levels of stressful life events would have more polysubstance use than their female counterparts.

Methods

Data and Study Sample

Data were extracted from Wave 3 of the National Epidemiological Survey on Alcohol and Related Conditions (NESARC-III), a cross-sectional survey of a nationally representative sample of U.S. non-institutionalized civilians. It was conducted by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) in 2012-2013. NESARC-III used multistage probability sampling to randomly select U.S. adults aged 18 and older (Grant et al., 2015). Of the original sample from NESARC-III, a total of 14,738 adults aged 50 and older were included in the analyses. This study was verified as a non-human subjects study by the Institutional Review Board of the author’s institution due to the deidentified data of NESARC-III (protocol number: 1703680184).

Measurement

Outcome Variable

Polysubstance use pattern was constructed as the outcome variable using latent class analysis (LCA). This study included six, binary (yes/no), indicators of substance use in the past year. Substances that have a euphoriant effect and are commonly used/misused were included, including excessive alcohol use (i.e., binge drinking), cannabis use, cocaine use, and prescriptions drug misuse (painkillers, sedatives/tranquilizers, stimulants) (Dackis & O'Brien, 2001; Mukherjee et al., 2008; Weaver, 2015). Literature shows that excessive alcohol use, illicit drug use (such as cocaine), and prescriptions drug misuse increase the risk for injury and overdose death (Seth et al., 2018). Also, this study included cannabis use as more and more states have implemented medical and recreational cannabis legalization, which has been linked to increased cannabis use among older adults (Han & Palamar, 2020). NESARC-III respondents were asked about their past-year experiences of using these substances. For example, during the interview, they were asked “Have you ever used sedatives or tranquilizers non-medically? For example, barbs, downers, Ambien, Lunesta, phenobarbital, pentobarbital, Halcion, Tuinal, Nembutal, Seconal, Librium, Valium, Xanax, benzodiazepines, tranks, Ativa.” Excessive alcohol use was measured as having 5 or more drinks (for men) or 4 or more drinks (for women) in a period of two hours or less. Notably, this study did not include tobacco use because recent literature suggests that nicotine produces little euphoriant effect (Dar et al., 2007; Wei et al., 2018).

LCA was conducted to identify the patterns of polysubstance use in the study sample. LCA can be used to identify distinct subgroups among study population based on observed categorical variables, classifying individuals with similar behaviors or characteristics into latent classes (Lanza & Rhoades, 2013). A key assumption of LCA is local independence, requiring individuals’ responses to items to be independent. We first estimated the 1-class model, serving as a comparison class with the other classes. We then increased the number of classes until the model no longer converged or had meaningful applications (Lanza et al., 2020). Optimal model selection was based on fit statistics including the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC), as well as model stability and interpretability (Nylund et al., 2007). A lower value (closer to zero) of AIC and BIC indicates a better balance model fit and model parsimony. We finally determined the number of classes and the membership of each class among all participants. This variable was a nominal variable.

Main Predictor

Number of stressful life events was the main predictor, which was assessed based on sixteen questions regarding past-year stressful experiences, including moving or someone new moving in, being laid off, being unemployed and looking for new job, having interpersonal conflict at workplace, starting a new job, getting divorce, experiencing death of a family member or close friend, having financial problems (unable to repay debt, declared bankruptcy), having interpersonal conflict with friends or relatives, being a victim of a crime (property has been stolen or damaged), having a family member or close friend who was the victim of a crime (being physically assaulted or attacked), having trouble with the police or the law, having family member or close friend who had trouble with the police or the law, and being homeless (Verplaetse et al., 2018). Each question was coded as “1” if responded yes. A summed total number of experienced stressful events (range 0 to 16) was used as continuous predictor variable. This scale has been widely used in previous studies and has been shown to be reliable (Lehavot et al., 2018; Vásquez et al., 2019). In this study, Cronbach’s alpha was .82, indicating high internal consistency of this scale.

Moderator

This study tested the moderating effect of gender (male = 0; female = 1) on the associations between stressful life events and polysubstance use patterns.

Covariates

Sociodemographic variables were included as covariates, including age (50–64, 65 and above), gender (male or female), race/ethnicity (non-Hispanic White, non-Hispanic Black, Asian/others, Hispanic), education (less than high school, high school, some college, college, or graduate school), marital status (married or not), unemployment (yes or no), and urbanicity (urban or rural).

Analytical Approach

Descriptive Analysis

This study first computed descriptive statistics for each variable. Unweighted frequencies and weighted percentages were reported for all categorical variables. Weighted means and standard errors were reported for continuous variables.

Generalized Structural Equation Modeling

Generalized structural equation modeling (GSEM) was used to conduct latent class analysis to identify subgroups of participants with distinct polysubstance use behaviors. After identifying the best-fitting model and the number of classes (as described in the measurement section), a multinomial logistic regression was estimated using GSEM to examine the associations between stressful life events and polysubstance use, controlling for all covariates. In the multinomial logistic regression, Class 2 (non-users/low substance users) was used as a reference group. Additionally, an interaction term between gender and stressful life events was included in the analyses to test the moderating effect of gender, which was further validated by an invariance test as a post-hoc test under GSEM. To produce nationally generalizable estimates, all analyses were weighed using the NESARC-III survey weights and sampling scheme. All statistical analyses were conducted using Stata 15.1. GSEM was fitted using the svy: gsem command in Stata.

Results

Table 1 shows descriptive statistics for the study sample (unweighted N = 14,738; weighted N = 102,465,952). Notably, 1.7% of U.S. adults aged 50 and older misused sedatives/tranquilizers, 3.1% misused painkillers, 3.9% used cannabis, 5.7% reported excessive alcohol use, .3% misused stimulants, and .4% used cocaine in past year. The mean count of stressful life events was 1.14 (SE = .02). In addition, 60% were between 50 and 64 years old, 53.2% were female, 75.1% were non-Hispanic White, 63.4% were married, 12.0% were unemployed, and 74.1% lived in urban areas.

Table 1.

Descriptive Statistics of the Study Sample.

	Unweighted n or weighted mean	Weighted % or weighted SE
Past-year substance use/misuse
Sedative/tranquilizer misuse	276	1.7%
Painkiller misuse	529	3.1%
Cannabis use	695	3.9%
Excessive alcohol use	890	5.7%
Stimulant misuse	55	.3%
Cocaine use	81	.4%
Main predictor
Stressful life events	1.14	.02
Covariates
Gender
Male	6392	46.8%
Female	8346	53.2%
Age
50–64	8932	60.0%
65 and above	5806	40.0%
Race/Ethnicity
Non-Hispanic White	9363	75.1%
Non-Hispanic Black	2782	10.0%
Asian/Other	813	5.9%
Hispanic	1780	9.0%
Married
No	7657	36.6%
Yes	7081	63.4%
Education
Less than high school	2536	14.1%
High school	4143	27.1%
Some college	4304	29.3%
College	2273	17.3%
Graduate school	1482	12.1%
Unemployment
No	12,561	88.0%
Yes	2177	12.0%
Urbanicity
Urban	11,537	74.1%
Rural	3201	25.9%

Noted: Weighted N = 102,465,952; Unweighted N = 14,738.

Data source: National Epidemiological Survey on Alcohol and Related Conditions (NESARC-III).

SE = standard error.

Weighted mean and standard error are computed for continuous variables.

Table 2 shows the latent class model fit indices and selection criteria. This study increased the number of classes until the 5-class solution because the 6-class solution did not converge. Models with 1–5 classes were compared to identify the optimal model. As the number of classes increased, the fit indices continued to decrease until the 3-class solution. Therefore, the 3-class solution was a point of diminishing returns in the balance of model fit and parsimony. Based on the model fit and parsimony as well as the considerations of class size and interpretability, the 3-class solution was considered as the optimal model. Results from the 3-class model are summarized in Figure 1 and supports distinct, meaningful latent classes of polysubstance use patterns. Specifically, participants in Class 1 reported co-using cannabis and excessive alcohol in the past year (labeled “cannabis and excessive alcohol users”; 2.4% of the study sample). Among Class 1 participants, cannabis had the highest use-probability (probability = .79), followed by excessive alcohol use (probability = .47). Class 2 was comprised of those who were substance non-users or used a small amount of substances in the past year (“non-users/low substance users”; 96.2%). Participants in Class 3 reported co-misusing painkillers and sedatives/tranquilizers in the past year (“painkiller and sedative or tranquilizer misusers”; 1.4%). Among Class 3 participants, painkillers had the highest use-probability (probability = .90), followed by sedatives/tranquilizers (probability = .76).

Table 2.

Latent Class Model Fit Indices.

Class	Log likelihood	df	AIC	BIC
1	−10,677.25	6	21,366.51	24,112.1
2	−9995.22	13	20,016.45	20,115.22
3	−9885.51	20	19,811.02	19,962.98
4	−9885.51	21	19,813.02	19,972.58
5	−9860.52	30	19,791.04	20,008.99

Noted: AIC = Akaike information criterion; BIC = Bayesian information criterion.

Bolded AIC and BIC indicate the best model fit.

Figure 1.

Latent class analysis of polysubstance use patterns.

Table 3 reports the results of the multinomial logistic regression analysis, indicating that higher levels of stressful life events were associated with co-using cannabis and excessive alcohol (adjusted odds ratio [AOR] = 1.33, 95% CI: 1.10-1.34, p < .001) and co-misusing of painkillers and sedatives/tranquilizers (AOR = 1.65, 95% CI: 1.45-1.87, p < .001). Females were at lower risk for co-using cannabis and excessive alcohol than males (AOR = .20, 95% CI: .17-.37, p < .001). Compared with middle-aged adults (aged between 50 and 64), older adults (aged 65 and older) were less likely to co-use cannabis and excessive alcohol (AOR = .14, 95% CI: .08-.25, p < .001) and co-misuse painkillers and sedatives/tranquilizers (AOR = .47, 95% CI: .26-.82, p < .001). Further, compared with participants who were not married, married participants were less likely to co-use cannabis and excessive alcohol (AOR = .49, 95% CI: .38-.63, p < .001) and co-misuse painkillers and sedatives/tranquilizers (AOR = .41, 95% CI: .27-.63, p < .001). Compared with participants who had less than high school education, those who had graduate school degree were less likely to co-use cannabis and excessive alcohol (AOR = .47, 95% CI: .24-.95, p < .05). Unemployed participants were more likely to co-use cannabis and excessive alcohol (AOR = 1.84, 95% CI: 1.28-2.65, p < .001) and co-misuse painkillers and sedatives/tranquilizers (AOR = 3.51, 95% CI: 2.43-5.07, p < .001) than those who are not unemployed. Participants who lived in rural areas were less likely to co-misuse painkillers and sedatives/tranquilizers than participants who lived in urban areas (AOR = .51, 95% CI: .30-.85, p < .05).

Table 3.

Weighted Multinomial Logistic Regression Estimates of Polysubstance Use by Generalized Structural Equation Modeling.

	Cannabis and excessive alcohol users (vs. Non-users/low substance users)		Painkiller and sedatives/tranquilizers misusers (vs. Non-users/low substance users)
	AOR	95% CI	AOR	95% CI
Stressful life events	1.33^***	1.10–1.34	1.65^***	1.45–1.87
Moderator
Female × stressful life events	1.03	.87–1.11	.84^*	.72–.99
Covariates
Gender
Male (ref.)	–	–	–	–
Female	.20^***	.17–.37	.88	.46–1.69
Age
50-64 (ref.)	—	–	–	–
65 and above	.14^***	.08–.25	.47^**	.26–.82
Race/Ethnicity
Non-Hispanic White (ref.)	–	–	–	–
Non-Hispanic Black	.93	.64–1.36	.87	.49–1.52
Asian/Other	.57^*	.31–1.02	.22^**	.08–.59
Hispanic	.87	.57–1.32	.60	.31–1.19
Married
No (ref.)	—	—	—	—
Yes	.49^***	.38–.63	.41^***	.27–.63
Education
Less than high school (ref.)	—	—	—	—
High school	1.43	.99–2.07	1.32	.79–2.32
Some college	1.10	.74–1.65	1.03	.59–1.82
College	.89	.51–1.57	.79	.39–1.58
Graduate school	.47^*	.24–.95	.67	.22–2.00
Unemployment
No (ref.)	—	—	—	—
Yes	1.84^**	1.28–2.65	3.51^***	2.43–5.07
Urbanicity
Urban (ref.)	—	—	—	—
Rural	1.03	.66–1.63	.51^*	.30–.85

Noted: Weighted N = 102,465,952; Unweighted N = 14,738.

Data source: National Epidemiological Survey on Alcohol and Related Conditions-III (NESARC-III).

AOR: adjusted odds ratio; CI: confidence interval.

*p < .05, **p < .01, ***p < .001.

Regarding the moderating effect of gender on the associations between stressful life events and polysubstance use patterns in this sample of older adults, the results indicate that when experiencing high levels of stressful life events, males have an even higher likelihood to co-misuse painkillers and sedatives/tranquilizers than females (AOR = .84 [female], 95% CI: .72-.99, p < .05). The invariance test further confirmed the significant difference between male and female participants on this association (χ² = 99.31, d.f. = 58, p < .05). However, the moderating effect of gender was not observed on the association between stressful life events and co-use cannabis and excessive alcohol.

Discussion

This study contributes to the literature by identifying patterns of polysubstance use and investigating associations between stressful life events and polysubstance use patterns among U.S. adults aged 50 and older, using data from a nationally representative study. Unlike previous studies, this study considered multiple substance use behaviors concurrently to identify patterns of polysubstance users (i.e., cannabis/excessive alcohol, painkillers/sedatives or tranquilizers). Moreover, this study extended previous studies of stressful life events and their associations with substance use. We found that past-year stressful life events were associated with co-using cannabis and excessive alcohol as well as co-misusing painkillers/sedatives or tranquilizers, and that gender moderated the later association.

Our results showed that 3.8% of U.S. adults aged 50 and older reported polysubstance use in the past year, including 2.4% co-using cannabis and excessive alcohol and 1.4% co-misusing painkillers and sedatives/tranquilizers. The most common polysubstance use pattern among adults aged 50 and older was co-using cannabis and excessive alcohol. Such a pattern is consistent with previous finding that middle-aged and older cannabis users were more likely to report excessive alcohol use (Salas-Wright et al., 2017). Previous studies found a higher blood tetrahydrocannabinol (THC) level in individuals co-using cannabis and alcohol compared to those who used only cannabis, indicating an increased risk for risky behaviors such as impaired driving among those co-using cannabis and alcohol (Hartman et al., 2015). In addition, middle-aged and older adults may experience age-related cognitive decline as their age increases (Salthouse, 2009), thus, co-use of cannabis and alcohol might exacerbate the problem resulting in increased incidence of adverse outcomes including dementia, being scammed, and traffic-related injuries and deaths. Substance use prevention programs and other interventions should enhance awareness of the consequences on co-using cannabis and alcohol.

This study identified co-misusing prescription drugs (i.e., painkillers and sedatives/tranquilizers) as the second common polysubstance use pattern among middle-aged and older adults. This is also consistent with prior findings that painkiller misusers are more likely to misuse sedatives/tranquilizers (Jones et al., 2012). Previous findings suggest that the main reason for misusing painkillers is to manage chronic pain (Lipari et al., 2017) as chronic pain is highly prevalent among older adults and often occurs among those who have comorbidities (Reid et al., 2015). In addition, previous studies indicated that two in three adults with chronic pain suffered from insomnia, resulting in higher likelihood of co-misusing painkillers and sedatives/tranquilizers among middle-aged and older adult population (Miller et al., 2018; Rubio et al., 2016). In fact, to mitigate the opioid epidemic and reduce overdose deaths due to concurrent use of opioids and sedatives, the Centers for Disease Control and Prevention (CDC) published guidelines for prescribing opioid analgesics to treat chronic pain, cautioning that concurrent use of opioids and sedatives should be avoided. The CDC also suggested that non-pharmacological therapy combined with non-opioid pharmacological therapy are preferred for adults to manage chronic pain (Dowell et al., 2016). In summary, although only 3.8% of the study sample were past-year polysubstance users, whereas 96.2% were non-users or used only a small amount of substances in past year, one study suggested that 1 in 3 overdose deaths was associated with polysubstance use (Kariisa et al., 2019). Thus, incorporating polysubstance use aspects into substance use intervention and control programs may reduce drug overdose and subsequent mortality. Moreover, because aging is associated with physical changes and increased odds of multi-morbidity, interventions to reduce polysubstance use should be better tailored to the characteristics of middle-aged and older adults.

A key finding from our study was that higher levels of stressful life events were associated with an increased likelihood of co-using cannabis and excessive alcohol as well as co-misusing painkillers and sedatives/tranquilizers among adults aged 50 and older, even after controlling for multiple sociodemographic variables. These findings suggest that substance use prevention efforts for late middle-aged and older adults, such as early screening and health education, should target those who experience stressful life events as an at-risk population. Specifically, around one-third of our study participants (35.24%) experienced the death of family members or a close friend in the past year. A previous study suggested that the death of a family member is associated with a higher likelihood of substance use (Glass et al., 1995). Thus, special attention to substance use prevention efforts should be given to this group. In addition, evidence-based programs such as the Stress Management and Resiliency Training (SMART) Program have been shown to be effective in managing psychological stress of late-life spousal loss and in enhancing resilience (Bui et al., 2018; Chesak et al., 2019). Such programs may have a secondary effect on preventing polysubstance use.

Our study results also showed that gender moderated the association between stressful life events and co-misusing painkillers and sedatives/tranquilizers. In the current study sample, males who experienced high levels of stressful life events were more likely to co-misuse painkillers and sedatives/tranquilizers than females experiencing the same level of stressful life events. This result is in line with general strain theory and previous findings that males are more likely to cope with strains by using a substance (Rubio et al., 2016). Therefore, prevention and intervention efforts that pay special attention to males who experience stressful life events to enhance their coping skills should be considered. Moreover, general strain theory suggests that coping skills mediate the associations between strains and substance use. Thus, future studies should incorporate a moderated mediation approach to examine coping as a pathway through which stressful life events and polysubstance use are connected, and how gender may moderate this pathway.

The current study also has secondary findings. Adults aged 65 and older were less likely to engage in polysubstance use compared with adults aged between 50 and 64. In particular, the majority of our study participants aged 50 to 64 were baby boomers who had greater exposure to substances in their youth stage than earlier generations (Specht et al., 2021). The result implies that targeting baby boomers in substance use prevention and reduction programs should be emphasized. This study found that married participants were at a lower risk for polysubstance use, which is consistent with previous studies (e.g., Salvatore et al., 2020). This study also, aligned with prior studies (e.g., Moody et al., 2016), found that unemployment is a risk factor of polysubstance use. However, the causal relationship between unemployment and polysubstance use is still unclear. Future study is needed before drawing conclusions of causality such as whether some people consider polysubstance use as a strategy to cope with unemployment.

There are limitations of the present study. First, due to the cross-sectional nature of NESARC-III, this study cannot determine the temporal sequence between stressful life events and polysubstance use. Future studies should employ a longitudinal study design to examine causal relationships between stressful life events over the life course and polysubstance use. Second, the information about stressful life events and past-year substance use/misuse was self-reported by respondents, thus, the information is subject to recall and response bias. Further, as described earlier, the general strain theory suggests that coping may serve as a critical mediator of the associations between stressful life events and substance use, which implies that coping skills may be potential pathways through which stressful life events are connected with polysubstance use. However, coping was not assessed as part of NESARC-III and the data are cross-sectional in nature, preventing consideration of mediation.

Despite these limitations, to our knowledge this is the first study that used LCA to identify the patterns of polysubstance use among adults aged 50 and older in the U.S. and that examined the associations between past-year stressful life events and polysubstance use. The LCA reveals the three most common polysubstance use patterns among aging adults on a national scale, making a unique contribution to present knowledge of polysubstance use among middle-aged and older adults. Given that the aging population is growing rapidly, our study findings could inform substance use intervention programs to incorporate aging adults’ polysubstance use behaviors. Substance intervention efforts to reduce polysubstance use should address stressful life events, implement gender-specific interventions, and customize approaches based on the patterns of polysubstance use. Through well-designed, tailored programs targeting at-risk subgroups, polysubstance use among middle-aged and older adults in the U.S. could be reduced.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical Approval

This study was verified as a non-human subjects study by the Institutional Review Board of Indiana University due to the deidentified data of NESARC-III (Protocol Number: 1703680184).

ORCID iDs

Tzung-Shiang Ou

Hsien-Chang Lin

References

Agnew

(2001). Building on the foundation of general strain theory: Specifying the types of strain most likely to lead to crime and delinquency. Journal of Research in Crime and Delinquency, 38(4), 319–361. https://doi.org/10.1177/0022427801038004001

Akins

Smith

C. L.

Mosher

(2010). Pathways to adult alcohol abuse across racial/ethnic groups: An application of general strain and social learning theories. Journal of Drug Issues, 40(2), 321–351. https://doi.org/10.1177/002204261004000204

Brennan

P. L.

Moos

R. H.

(1996). Late-life problem drinking: Personal and environmental risk factors for 4-year functioning outcomes and treatment seeking. Journal of Substance Abuse, 8(2), 167–180. https://doi.org/10.1016/s0899-3289(96)90227-8

Brezina

(2017). General strain theory. In Oxford research encyclopedia of criminology and criminal justice. Oxford University Press.

Bui

Chad-Friedman

Wieman

Grasfield

R. H.

Rolfe

Dong

Park

E. R.

Denninger

J. W.

(2018). Patient and provider perspectives on a mind–body program for grieving older adults. The American Journal of Hospice & Palliative Care, 35(6), 858–865. https://doi.org/10.1177/1049909117743956

Chesak

S. S.

Khalsa

T. K.

Bhagra

Jenkins

S. M.

Bauer

B. A.

Sood

(2019). Stress management and resiliency training for public school teachers and staff: A novel intervention to enhance resilience and positively impact student interactions. Complementary Therapies in Clinical Practice, 37(1), 32–38. https://doi.org/10.1016/j.ctcp.2019.08.001

Compton

W. M.

Valentino

R. J.

DuPont

R. L.

(2021). Polysubstance use in the US opioid crisis. Molecular Psychiatry, 26(1), 41–50. https://doi.org/10.1038/s41380-020-00949-3

Dackis

C. A.

O'Brien

C. P.

(2001). Cocaine dependence: A disease of the brain’s reward centers. Journal of Substance Abuse Treatment, 21(3), 111–117. https://doi.org/10.1016/s0740-5472(01)00192-1

Dar

Kaplan

Shaham

Frenk

(2007). Euphoriant effects of nicotine in smokers: Fact or artifact? Psychopharmacology, 191(2), 203–210. https://doi.org/10.1007/s00213-006-0662-2

10.

Dowell

Haegerich

T. M.

Chou

(2016). CDC guideline for prescribing opioids for chronic pain—United States, 2016. JAMA, 315(15), 1624–1645. https://doi.org/10.1001/jama.2016.1464

11.

Fendrich

Becker

Park

Russell

Finkelstein-Fox

Hutchison

(2021). Associations of alcohol, marijuana, and polysubstance use with non-adherence to COVID-19 public health guidelines in a US sample. Substance Abuse, 42(2), 220–226. https://doi.org/10.1080/08897077.2021.1891603

12.

Florence

C. S.

Zhou

Luo

(2016). The economic burden of prescription opioid overdose, abuse and dependence in the United States, 2013. Medical Care, 54(10), 901–906. https://doi.org/10.1097/MLR.0000000000000625

13.

Glass

T. A.

Prigerson

Kasl

S. V.

Mendes de Leon

C. F.

(1995). The effects of negative life events on alcohol consumption among older men and women. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 50(4), S205–S216. https://doi.org/10.1093/geronb/50b.4.s205

14.

Grant

B. F.

Goldstein

R. B.

Smith

S. M.

Jung

Zhang

Chou

S. P.

Pickering

R. P.

Ruan

W. J.

Huang

Saha

T. D.

Aivadyan

Greenstein

Hasin

D. S.

(2015). The alcohol Use disorder and associated disabilities interview schedule-5 (AUDADIS-5): Reliability of substance use and psychiatric disorder modules in a general population sample. Drug and Alcohol Dependence, 148(1), 27–33. https://doi.org/10.1016/j.drugalcdep.2014.11.026

15.

Han

B. H.

Palamar

J. J.

(2020). Trends in cannabis use among older adults in the United States, 2015-2018. JAMA Internal Medicine, 180(4), 609–611. https://doi.org/10.1001/jamainternmed.2019.7517

16.

Hartman

R. L.

Brown

T. L.

Milavetz

Spurgin

Gorelick

D. A.

Gaffney

Huestis

M. A.

(2015). Controlled cannabis vaporizer administration: Blood and plasma cannabinoids with and without alcohol. Clinical Chemistry, 61(6), 850–869. https://doi.org/10.1373/clinchem.2015.238287

17.

Hassamal

Miotto

Wang

Saxon

A. J.

(2016). A narrative review: The effects of opioids on sleep disordered breathing in chronic pain patients and methadone maintained patients. The American Journal on Addictions, 25(6), 452–465. https://doi.org/10.1111/ajad.12424

18.

Jones

J. D.

Mogali

Comer

S. D.

(2012). Polydrug abuse: A review of opioid and benzodiazepine combination use. Drug and Alcohol Dependence, 125(1–2), 8–18. https://doi.org/10.1016/j.drugalcdep.2012.07.004

19.

Kalapatapu

R. K.

Sullivan

M. A.

(2010). Prescription use disorders in older adults. The American Journal on Addictions, 19(6), 515–522. https://doi.org/10.1111/j.1521-0391.2010.00080.x

20.

Kariisa

Scholl

Wilson

Seth

Hoots

(2019). Drug overdose deaths involving cocaine and psychostimulants with abuse potential—United States, 2003–2017. MMWR. Morbidity and Mortality Weekly Report, 68(17), 388–395. https://doi.org/10.15585/mmwr.mm6817a3

21.

Lanza

H. I.

Motlagh

Orozco

(2020). E-Cigarette use among young adults: A latent class analysis examining co-use and correlates of nicotine vaping. Addictive Behaviors, 110(1), 106528. https://doi.org/10.1016/j.addbeh.2020.106528

22.

Lanza

S. T.

Rhoades

B. L.

(2013). Latent class analysis: An alternative perspective on subgroup analysis in prevention and treatment. Prevention Science: The Official Journal of the Society for Prevention Research, 14(2), 157–168. https://doi.org/10.1007/s11121-011-0201-1

23.

Lehavot

Goldberg

S. B.

Chen

J. A.

Katon

J. G.

Glass

J. E.

Fortney

J. C.

Simpson

T. L.

Schnurr

P. P.

(2018). Do trauma type, stressful life events, and social support explain women veterans’ high prevalence of PTSD? Social Psychiatry and Psychiatric Epidemiology, 53(9), 943–953. https://doi.org/10.1007/s00127-018-1550-x

24.

Lipari

R. N.

Williams

Van Horn

S. L.

(2017). Why Do Adults Misuse Prescription Drugs? In The CBHSQ Report (pp. 1-10). Substance Abuse and Mental Health Services Administration (US).

25.

Miller

M. B.

Chan

W. S.

Curtis

A. F.

Boissoneault

Robinson

Staud

Berry

R. B.

McCrae

C. S.

(2018). Pain intensity as a moderator of the association between opioid use and insomnia symptoms among adults with chronic pain. Sleep Medicine, 52(1), 98–102. https://doi.org/10.1016/j.sleep.2018.08.015

26.

Moody

Franck

Hatz

Bickel

W. K.

(2016). Impulsivity and polysubstance use: A systematic comparison of delay discounting in mono-dual-and trisubstance use. Experimental and Clinical Psychopharmacology, 24(1), 30–37. https://doi.org/10.1037/pha0000059

27.

Moos

R. H.

Brennan

P. L.

Schutte

K. K.

Moos

B. S.

(2004). High-risk alcohol consumption and late-life alcohol use problems. American Journal of Public Health, 94(11), 1985–1991. https://doi.org/10.2105/ajph.94.11.1985

28.

Mukherjee

Das

S. K.

Vaidyanathan

Vasudevan

D. M.

(2008). Consequences of alcohol consumption on neurotransmitters-an overview. Current Neurovascular Research, 5(4), 266–272. https://doi.org/10.2174/156720208786413415

29.

Myers

McLaughlin

K. A.

Wang

Blanco

Stein

D. J.

(2014). Associations between childhood adversity, adult stressful life events, and past-year drug use disorders in the National Epidemiological Study of Alcohol and Related Conditions (NESARC). Psychology of Addictive Behaviors: Journal of the Society of Psychologists in Addictive Behaviors, 28(4), 1117–1126. https://doi.org/10.1037/a0037459

30.

Nylund

K. L.

Asparouhov

Muthén

B. O.

(2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling: A Multidisciplinary Journal, 14(4), 535–569. https://doi.org/10.1080/10705510701575396

31.

Ogbu

U. C.

Lotfipour

Chakravarthy

(2015). Polysubstance abuse: Alcohol, opioids and benzodiazepines require coordinated engagement by society, patients, and physicians. The Western Journal of Emergency Medicine, 16(1), 76–79. https://doi.org/10.5811/westjem.2014.11.24720

32.

Olthuis

J. V.

Darredeau

Barrett

S. P.

(2013). Substance use initiation: The role of simultaneous polysubstance use. Drug and Alcohol Review, 32(1), 67–71. https://doi.org/10.1111/j.1465-3362.2012.00470.x

33.

Reid

M. C.

Eccleston

Pillemer

(2015). Management of chronic pain in older adults. BMJ, 350(1), h532. https://doi.org/10.1136/bmj.h532

34.

Rubio

Dumitrache

Cordon-Pozo

Rubio-Herrera

(2016). Coping: Impact of gender and stressful life events in middle and in old age. Clinical Gerontologist, 39(5), 468–488. https://doi.org/10.1080/07317115.2015.1132290

35.

Sacco

Bucholz

K. K.

Harrington

(2014). Gender differences in stressful life events, social support, perceived stress, and alcohol use among older adults: Results from a national survey. Substance Use & Misuse, 49(4), 456–465. https://doi.org/10.3109/10826084.2013.846379

36.

Salas-Wright

C. P.

Vaughn

M. G.

Cummings-Vaughn

L. A.

Holzer

K. J.

Nelson

E. J.

AbiNader

(2017). Trends and correlates of marijuana use among late middle-aged and older adults in the United States, 2002–2014. Drug and Alcohol Dependence, 171, 97–106. https://doi.org/10.1016/j.drugalcdep.2016.11.031

37.

Salthouse

T. A.

(2009). When does age-related cognitive decline begin? Neurobiology of Aging, 30(4), 507–514. https://doi.org/10.1016/j.neurobiolaging.2008.09.023

38.

Salvatore

J. E.

Gardner

C. O.

Kendler

K. S.

(2020). Marriage and reductions in men's alcohol, tobacco, and cannabis use. Psychological Medicine, 50(15), 2634–2640. https://doi.org/10.1017/S0033291719002964

39.

Schulte

M. T.

Hser

Y.-I.

(2013). Substance use and associated health conditions throughout the lifespan. Public Health Reviews, 35(2), 3. https://doi.org/10.1007/BF03391702

40.

Seth

Scholl

Rudd

R. A.

Bacon

(2018). Overdose deaths involving opioids, cocaine, and psychostimulants—United States, 2015–2016. MMWR. Morbidity and Mortality Weekly Report, 67(12), 349–358. https://doi.org/10.15585/mmwr.mm6712a1

41.

Specht

Braun‐Michl

Schwarzkopf

Piontek

Seitz

N. N.

Wildner

Kraus

(2021). Substance use disorder and the baby boom generation: Does Berlin outpatient addiction care face a sustained change? Drug and Alcohol Review, 40(6), 979–988. https://doi.org/10.1111/dar.13245

42.

Timko

Han

Woodhead

Shelley

Cucciare

M. A.

(2018). Polysubstance use by stimulant users: Health outcomes over three years. Journal of Studies on Alcohol and Drugs, 79(5), 799–807. https://doi.org/10.15288/jsad.2018.79.799

43.

United States Census Bureau (2018). Older people projected to outnumber children for first time in US history. United States Census Bureau.

44.

Vásquez

Quiñones

Ramirez

Udo

(2019). Association between adverse childhood events and multimorbidity in a racial and ethnic diverse sample of middle-aged and older adults. Innovation in Aging, 3(2), igz016. https://doi.org/10.1093/geroni/igz016

45.

Verplaetse

T. L.

McKee

S. A.

(2017). An overview of alcohol and tobacco/nicotine interactions in the human laboratory. The American Journal of Drug and Alcohol Abuse, 43(2), 186–196. https://doi.org/10.1080/00952990.2016.1189927

46.

Verplaetse

T. L.

Moore

K. E.

Pittman

B. P.

Roberts

Oberleitner

L. M.

Smith

P. H.

Cosgrove

K. P.

McKee

S. A.

(2018). Intersection of stress and gender in association with transitions in past year DSM-5 substance use disorder diagnoses in the United States. Chronic Stress, 2(1), 2470547017752637. https://doi.org/10.1177/2470547017752637

47.

Weaver

M. F.

(2015). Focus: Addiction: Prescription sedative misuse and abuse. The Yale Journal of Biology and Medicine, 88(3), 247–256.

48.

Wei

Han

Weng

Feng

Luo

(2018). Response dynamics of midbrain dopamine neurons and serotonin neurons to heroin, nicotine, cocaine, and MDMA. Cell Discovery, 4(1), 60. https://doi.org/10.1038/s41421-018-0060-z

49.

L.-T.

Blazer

D. G.

(2011). Illicit and nonmedical drug use among older adults: A review. Journal of Aging and Health, 23(3), 481–504. https://doi.org/10.1177/0898264310386224

50.

Youdin

(2019). Old and high: A Guide to understanding the neuroscience and psychotherapeutic treatment of baby-boom adults' substance use, Abuse, and misuse. Oxford University Press.