Risk Factors for Drug Overdose in Young People: A Systematic Review of the Literature

Abstract

Objectives: Drug overdose (OD) deaths have been increasing over the past 20 years. Although risk factors for drug OD have been identified in adult populations, less is known about risk factors for OD in young people. The aim of this review is to systematically examine the literature to identify risk factors for drug OD specific to young people, including adolescents and young adults.

Methods: Our initial PubMed search identified 4001 articles. Included were cross-sectional and longitudinal cohort studies published in English that compared young people who experienced a drug OD to those who did not. Review articles, meta-analyses, case-reports, editorials, epidemiological studies, and qualitative studies were excluded. Two investigators reviewed the full texts of all relevant articles and extracted data on sample demographics, prevalence of OD, and correlates associated with OD.

Results: Twelve relevant studies were identified reflective of a sample of 5020 unique individuals with an age range of 14–30 years, and a mean age range of 20.2–26 years. The lifetime prevalence of OD in these young people ranged from 24% to 48%. Substance use characteristics most often associated with OD included injection drug, opioid, and tranquilizer use. Polysubstance use was also found to be strongly associated with OD in three studies. Other replicated risk factors for OD in young people included histories of psychopathology, incarceration, unstable housing, and witnessing an OD.

Conclusion: Opioid, tranquilizer, and injection drug use have been identified as risk factors for OD in both younger and older adult populations. Risk factors that emerged as noteworthy predictors of OD in young people specifically include polysubstance use, psychiatric comorbidity, unstable housing, and witnessing an OD. There remains a paucity of literature on drug OD risk factors in young people, with little information regarding medical and treatment history risk factors.

Introduction

Over the past 20 years, the number of drug overdoses (ODs) in the United States has increased dramatically (Rudd et al. 2016; Curtin et al. 2017). In 2017, an estimated 72,306 people in the United States died from a drug OD (National Institute on Drug Abuse 2018). Although the majority of drug OD occurs in adults, 35–44 years of age (Brady et al. 2017), younger people are also at significant risk. For example, between 2015 and 2016, the largest increase in drug OD deaths occurred in males 15–24 years of age (Seth et al. 2018). Likewise, between 1999 and 2015, the rate of drug OD deaths in adolescents 15–19 years of age nearly doubled (Curtin et al. 2017), and recent data suggest that rates are continuing to increase (National Institute on Drug Abuse 2018). In 2016, 5376 people 15–24 years of age died of a drug OD, making OD a leading cause of death in young people (Center for Disease Control and Prevention 2018; Hedegaard et al. 2017).

Since the majority of drug OD involve opioids (Seth et al. 2018), much of the effort to decrease OD deaths has focused on increasing access to medication for opioid use disorders and increasing access to naloxone (the antidote for an opioid OD) (Volkow and Collins 2017). Indeed, the use of opioids such as heroin (Hakansson et al. 2008) and/or fentanyl has been identified as risk factors of drug OD in adults (Rudd et al. 2016; Somerville et al. 2017). A less extensive, but significant literature also identifies alcohol (Coffin et al. 2007), cocaine (Coffin et al. 2007), benzodiazepines (Jones and McAninch 2015), and the method of use, such as intravenous drug use, as other substance use-related risk factors in adults (Hakansson et al. 2008). Non-substance-related risk factors for drug OD in adults have also been identified, including history of mental illness, particularly depression (Webster et al. 2011; Bohnert et al. 2013; Wilder et al. 2016).

Although there is a growing literature on risk factors associated with drug OD in adults, less research has focused on OD risk factors in young people, which includes adolescents 12–18 years of age (U.S. Department of Health and Human Services 2013), and young adults 18–30 years of age (Arnett et al. 2014). While research on older adults provides a useful starting point to examining risk factors for drug OD in young people, individuals in this age group are developmentally and psychologically unique and findings in older adults may not generalize to younger populations. For example, in young people, brain maturation, in particular frontal lobe development, which involves impulse control and executive functioning, continues into the mid-20s (Paulozzi et al. 2011). Consequently, young people may be more likely to engage is risky behavior like polysubstance use, which could increase their likelihood of experiencing a drug OD (Connor et al. 2014).

Risk factors for drug OD in younger people relative to older people may also be different in that early-onset substance use disorders (SUD) are associated with a more complicated course of illness (Sowell et al. 1999; Poudel and Gautam 2017), which may be more genetically influenced (Chen et al. 2009), and more likely to be comorbid with other psychiatric disorders (Meyers and Dick 2010). At the same time, young people may be less likely than their older counterparts to engage in care (Winters 1999), receive medication to treat their SUD (Chang et al. 2018), or to be educated on naloxone (Frank et al. 2015). Hence, it is imperative to examine risk factors for drug OD specific to young people to decrease the risk for OD in this age group. The aim of this review is to identify and summarize the extant literature that examines the demographic, substance use, psychiatric, medical, psychosocial, and treatment history characteristics associated with drug OD in young people 12–30 years of age.

Methods

We conducted a systemic review of peer-reviewed literature published through December 31st, 2018, which examined risk factors for OD involving substances such as alcohol and drugs in adolescents (12–18 years old) and emerging adults (18–30 years old) (Arnett et al. 2014). We searched the PubMed electronic database using various combinations of the following search terms: “overdose,” “risk factors,” “youth,” “young people,” “adolescents,” “emerging adults,” “transitional aged youth,” and “young adults.” The titles and abstracts of each article were reviewed. Included were cross-sectional and longitudinal cohort studies published in English that compared individuals who experienced an OD to those who did not. Review articles, meta-analyses, case-reports, editorials, epidemiological studies, and qualitative studies were excluded. All relevant articles were cross referenced to ensure no additional relevant articles were overlooked.

Two investigators reviewed the full texts of all relevant articles. The following variables were extracted from all articles, including sample demographics, prevalence of OD, and correlates associated with OD. If the studies performed bivariate and multivariate logistic regression analyses, only findings from the multivariate logistic regression analyses were included in the results of this review (Fig. 1).

FIG. 1.

Flow diagram of the identification, review, and selection of articles that examined the risk factors for drug overdose in young people 12–30 years of age. OD, overdose.

Results

Search results

Our initial search yielded 4001 articles once duplicate studies were removed. After careful review, 3989 articles were excluded. We excluded 3563 articles that did not examine risk factors associated with OD, 335 that were published in a language other than English, 84 that focused on older adult populations, 1 that examined risk factors for unintentional pediatric exposure, 5 that focused on intentional self-poisoning with substances other than alcohol or illicit drugs (e.g., nonopioid paracetamols, antidepressants, and pesticides), and 1 case report. Our final review consisted of 12 articles.

There were nine cross-sectional studies and three longitudinal studies—all observational in nature. The studies were reflective of a sample of 5020 unique individuals with an age range of 14–30 years. The sample sizes ranged from 124 to 858 subjects. The majority (n = 11) of the studies included adolescents and young adults younger than 18. Mean ages reported in the studies ranged from 20.2 to 26 years, and median age ranged from 21 to 28.7 years. Most studies examined risk factors associated with OD exclusively in young people, with one study, including adults of all ages, comparing those 18–30 years of age to those >30 years of age.

All studies examined risk factors associated with nonfatal OD. Most studies (n = 9) focused on risk factors associated with OD of unspecified intentionality with multivariate logistic regression analyses (Calvo et al. 2017). Two studies examined risk factors of unintentional OD (OD without the intention of self-harm) with multivariate logistic regression analysis (Richer et al. 2013; Mitra et al. 2015). One study examined risk for both unintentional and intentional OD (ingestion of a substance that was reported as a suicide attempt) in the same sample with bivariate analyses (Yule et al. 2018). In this review, OD will refer to OD of unspecified intent unless otherwise stated.

The populations of each study varied in terms of substance use. Six studies included only individuals who reported high risk substance use (i.e., heroin, crack, cocaine, and/or injection drug use) (Ochoa et al. 2001, 2005; Burns et al. 2004; Sherman et al. 2007; Chahua et al. 2014; Riley et al. 2016). Other studies consisted of more heterogenous samples, including two studies that consisted of individuals who reported using drugs other than marijuana in the 30 days before study enrollment (Werb et al. 2008; Mitra et al. 2015) and two studies that consisted of individuals engaged in SUD treatment or harm reduction programs (Calvo et al. 2017; Yule et al. 2018). Cohorts of two studies were made up of young people who were unstably housed or homeless (Richer et al. 2013; Mitra et al. 2015). One study consisted of young people who endorsed a history of nonmedical prescription drug use (Silva et al. 2013). All studies were conducted in large metropolitan areas, including the United States (N = 7), Canada (4), Spain (1), and Australia (1) (Table 1).

Table 1.

Studies Reporting Risk Factors for Drug Overdose in Young People 12–30 Years Of Age

References	Inclusion dates	Ascertainment	N	Age [range] (years)	Substance used in OD	Definition of OD	Variables associated with increased risk for OD
Cross-sectional cohort studies
Burns et al. (2004)	2000	Lifetime history of heroin use	163	Median: 21	Heroin	Requiring an ambulance or naloxone after consuming too much heroin	Relationship dissatisfaction, mental illness, hopelessness, antisocial behavior, prescription opioid use, tranquilizer use, benzodiazepine use, and antidepressant use
				[15–30]
Calvo et al. (2017)	2014–2015	Young people using drugs engaged in harm reduction programs	257	Median: 26	All substances	Not imposed	Symptoms of psychological distress, unstable housing or homelessness, inpatient admission for substance detoxification, heroin use, methadone use (not prescribed), and novel/rare drug use
				[18–30]
Chahua et al. (2014)	2002–2005	Young people using heroin, who reported using at least 12 days in the last year and at least once in the last 3 months	452	Mean: 26 ± 3.25	Opioids	An episode occurring after opioid use and characterized by extreme difficulty in breathing, loss of consciousness and problems waking up or recovering consciousness, and possibly bluish skin or lips	Incarceration, depression, injection drug use, and tranquilizer use
				[≤30]
Ochoa et al. (2001)	1996	Young people who reported injecting drugs in the past 6 months	124	Median: 22	All substances	Not imposed	Borrowing syringes and gay or bisexual behavior
				[14–29]
Ochoa et al. (2005)	2000–2001	Young people who reported injecting heroin once or more in the past month	795	Median: 22	Heroin	Symptom scale ranging from 1 to 10:	Being tested for hepatitis B or C, witnessing an OD, incarceration, heroin use, cocaine use, heroin mixed with methamphetamine use, and injection drug use
						1. “the person is in a heavy nod,” 7. “the person stops breathing,” 10. “the person is dead”
				IQR: 20–25
				[15–29]
Sherman et al. (2007)	1999–2001	Young people who reported injection or non-injection use of heroin, crack, and/or cocaine within the past 5 years, reported using heroin, crack and/or cocaine 2 days in the past week, and/or had injected at least once in the past month	309	Median: 28.7	Opiates	Not imposed	White race, homelessness, and length of injection drug use
				IQR: 26.0–31.5
				[15–30]
Silva et al. (2013)	2009–2011	Young people with nonmedical prescription drug use that reported using at least three times in the past 90 days	596	Mean: 21	Opioids, tranquilizers	Not imposed	Lower social class while growing up, psychiatric hospitalization, witnessing a family member OD, opioid use, tranquilizer use, and injection drug use
				[16–25]
Werb et al. (2008)	2005–2006	Young people using drugs, who reported substance use other than marijuana in the last 30 days	478	Median: 22	All substances	Having a negative reaction due to the over consumption of drugs	Female gender, crystal methamphetamine use, heroin use, cocaine use, and injection drug use
				IQR: 20.0–23.9
				[14–26]
Yule et al. (2018)	2012–2013	Young people with substance use disorders seeking substance use disorder treatment	200	Mean 20.2 ± 2.8	All substances	Unintentional OD: substance use without intention of self-harm that was associated with a significant impairment in level of consciousness	Blackouts, injection drug use, history of an eating disorder, alcohol use disorder, and amphetamine use disorder
				[16–26]
						Intentional OD: an ingestion of a substance that was reported as a suicide attempt
Longitudinal cohort studies
Mitra et al. (2015)	2005–2012	Young people with illicit drug use (other than marijuana) in the past 30 days, who were living without housing for at least 6 months	615	Median: 21	All substances	A negative reaction from using too much drugs	Crystal methamphetamine use, prescription opioid use, heroin use, injection drug use, and binge drug use
				IQR: 20–23
				[14–26]
Richer et al. (2013)	2001–2005	Young people who did not have a place to sleep more than once in the past 6 months or have used services regularly	858	Mean: 20.4	All substances	Not imposed	Suicidal ideation, homelessness, injection drug use, and polydrug use
				[14–23]
Riley et al. (2016)	2012–2014	Young people who reported having injected drugs in the last 3 months	173	Median: 25	Opioids	A loss of consciousness during which at least one intervention was attempted by a third party (e.g., naloxone or rescue breathing)	Heroin use, benzodiazepine use, and alcohol use
				IQR: 23–27
				[≤30]

IQR, interquartile range; OD, overdose.

Prevalence

The prevalence of OD was evaluated over different time durations and overall ranged from 8% to 48%. Three studies examined OD in the past 6 months and found that 8%–16% of subjects experienced an OD during that time (Werb et al. 2008; Richer et al. 2013; Mitra et al. 2015). The highest prevalence of OD in the past 6 months was observed in a study of street-involved youth who reported using drugs other than marijuana (Mitra et al. 2015). Three studies examined OD in the year before study enrollment and found that 9%–22% of individuals experienced an OD during that time (Ochoa et al. 2005; Chahua et al. 2014; Calvo et al. 2017). The lowest prevalence of OD in the past year was observed in a study of individuals who reported recent heroin use (injection and non-injection use) (Chahua et al. 2014), while the highest prevalence was observed in a study of individuals who reported injecting heroin (Ochoa et al. 2005).

The lifetime prevalence of OD was reported in seven studies and ranged from 24% to 48% (Ochoa et al. 2001; Burns et al. 2004; Sherman et al. 2007; Richer et al. 2013; Silva et al. 2013; Calvo et al. 2017; Yule et al. 2018). The lowest prevalence of lifetime OD was observed in a cohort of young people who reported nonmedical prescription drug use at least three times in the past 90 days (Silva et al. 2013). The highest prevalence of lifetime OD was observed in a cohort of young people who reported injection drug use (Ochoa et al. 2001). Notably, two studies that focused on unintentional OD in street-involved young people in Canada found particularly high rates of OD in this population, ranging from 38% to 40% (Mitra et al. 2015; Richer et al. 2013).

Three of the 12 studies identified were longitudinal in nature (Richer et al. 2013; Mitra et al. 2015; Riley et al. 2016). Of these studies, Richer et al. (2013) was the only study to report data on history of OD at baseline and prevalence of OD during the follow-up period. Although there was a high prevalence of lifetime OD at baseline (38%), and many OD during the four-and-a-half-year study period (326 ODs, with 6% of participants reported experiencing >1 OD), the researchers did not report on the relationship between past and subsequent OD.

Correlates associated with OD

Demographic risk factors

Although all studies examined the relationship between demographic characteristics and OD, gender was the only demographic characteristic that was replicated. Female gender was associated with higher risk for OD in two studies. One study of 478 young people who reported using drugs other than marijuana found that females were two times more likely than males to experience an OD (adjusted odds ratio [AOR]: 2.06; 95% confidence interval [CI] 1.04–4.07, p = 0.038) (Werb et al. 2008). Likewise, in their study of 200 young people seeking treatment for SUD, Yule et al. (2018) also found that female gender was associated with increased risk OD (p < 0.05).

There were no other replicated findings linking additional demographic characteristics to OD. Only 8 of the 12 studies examined the relationship between race and OD, and most studies that did report data on race consisted of cohorts that were primarily White (Ochoa et al. 2001, 2005; Sherman et al. 2007; Werb et al. 2008; Silva et al. 2013; Mitra et al. 2015; Calvo et al. 2017; Yule et al. 2018). White race was independently associated with OD history (AOR: 3.2; 95% CI 1.6–6.4) in one study of 309 young people who used crack, heroin, and/or cocaine (Sherman et al. 2007). Lower socioeconomic status in childhood was another demographic characteristic associated with increased risk for OD in a study of 596 young people who reported nonmedical prescription drug use (AOR: 1.81; 95% CI 1.15–2.83, p < 0.01) (Silva et al. 2013).

Substance use risk factors

All 12 studies found specific substance use characteristics and behaviors, such as method of use or type of drug, to be significant risk factors for OD in young people.

Injection drug use

Injection drug use is associated with more severe SUD and was the most common risk factor for OD in young people associated with substance use. Six out of the seven studies that examined risk associated with injection drug use found it was a risk factor for OD (Ochoa et al. 2005; Werb et al. 2008; Richer et al. 2013; Silva et al. 2013; Chahua et al. 2014; Mitra et al. 2015; Yule et al. 2018). For example, in a study of 795 young people who injected heroin, Ochoa et al. (2005) found both injection heroin and injection cocaine use were predictors of OD (both p < 0.01). These results were replicated by Werb et al. (2008), who found those who injected heroin and those who injected cocaine were three (p = 0.002) and five (p < 0.001) times more likely, respectively, to have experienced an OD.

The only group that evaluated the relationship between injection drug use and OD with both a bivariate and multivariate analysis and did not find a significant association between the two variables when adjusted was Calvo et al. (2017). In their study of 257 young people engaged in harm reduction programs, Calvo et al. (2017) initially found that current injection drug use was associated with past year OD (p < 0.001), but when evaluated in a multivariate logistic regression, the association no longer remained significant (p = 0.460).

Four of the five studies that did not examine injection drug use as a risk factor for OD evaluated cohorts made up entirely of individuals who reported injection drug use (Ochoa et al. 2001, 2005; Sherman et al. 2007; Riley et al. 2016). Sherman et al. (2007) did examine the relationship between length of injection drug use and OD and found that the longer the subject injected drugs, the greater their risk of experiencing an OD. In contrast, in a study of 124 young people, Ochoa et al. (2001) did not find a significant relationship between length of injection drug use and OD. Ochoa et al. (2001) did, however, find that individuals who reported sharing syringes in the past 30 days were 2.5 times more likely (AOR: 2.53; 95% CI 1.11–5.78, p = 0.03) to report a lifetime history of OD.

Opioid use

Opioid use (including oral prescription opioid use, injection heroin use, intranasal heroin use, and mixing heroin with other drugs) was strongly linked to OD history in seven distinct studies (Burns et al. 2004; Ochoa et al. 2005; Werb et al. 2008; Silva et al. 2013; Mitra et al. 2015; Riley et al. 2016; Calvo et al. 2017). One study found that those with a history of opioid use were up to 4.8 times more likely (AOR: 4.89; 95% CI 2.03–11.74, p < 0.01) to have experienced an OD than those who did not have a history of opioid use (Ochoa et al. 2005).

Tranquilizer and benzodiazepine use

Tranquilizer use (including barbiturates, benzodiazepine, and other sedative medication) was another replicated risk factor positively associated with OD in four of the eight studies that examined it (Burns et al. 2004; Ochoa et al. 2005; Sherman et al. 2007; Silva et al. 2013; Chahua et al. 2014; Riley et al. 2016; Calvo et al. 2017; Yule et al. 2018). Benzodiazepine use was specifically identified as a risk factor for OD in two studies (Burns et al. 2004; Riley et al. 2016). One study linked data from a government database on participants' prescription medication use in the last 5 years with self-report data from individuals who use heroin and observed that benzodiazepine use was positively associated with OD (p < 0.01) (Burns et al. 2004). Similarly, in their study of young people who reported injection drug use, Riley et al.(2016) found that with every five additional benzodiazepine pill-taking days, the risk of OD increased 1.2-fold (AOR: 1.22; 95% CI 1.04–1.43).

Alcohol use

Alcohol was found to be a risk factor for OD in two studies. Riley et al. (2016) found that drinking 10 or more alcoholic drinks a day increased an individual's risk of experiencing an OD 4.5-fold when compared to individuals who did not drink (AOR: 4.50; 95% CI 1.77–11.45). Yule et al. (2018) also found that individuals with an alcohol use disorder were 2.4 times more likely than those without an alcohol use disorder to have a history of OD (odds ratio [OR]: 2.39; 95% CI 1.05–5.45, p < 0.05).

Stimulant use

While all studies examined the relationship between stimulant use (cocaine, amphetamines, or methamphetamines) and OD, only 4 of the 12 studies reported a significant association between the two variables. Cocaine use was found to be associated with an increased risk for OD in three studies (Ochoa et al. 2005; Werb et al. 2008;Yule et al. 2018).

In addition, Werb et al. (2008) found that individuals who reported crystal methamphetamine use were two times more likely to have a history of OD (AOR: 2.00; 95% CI 1.06–3.77, p = 0.032). These findings were supported by data from a study of 615 unstably housed young people who reported using drugs other than or in addition to marijuana (Mitra et al. 2015). In their analysis, Mitra et al. (2015) found that individuals who used methamphetamine were almost two times more likely to experience an unintentional OD (AOR: 1.70; 95% CI 1.12–2.58, p = 0.013).

Similarly, Yule et al. (2018) found young people with an amphetamine use disorder were nearly three times more likely to have a history of an OD (OR: 2.87; 95% CI 1.24–6.64, p < 0.01). In contrast to these studies, Silva et al. (2013) found individuals who reported that prescription stimulant misuse (taking a prescription stimulant that “was not prescribed for [them] or that [they] took only for the experience or feeling it caused”) was at decreased risk for OD when compared to those who did not misuse prescription stimulants (AOR: 0.60 95% CI 0.38–0.96, p < 0.05).

Novel and rare drug use

Use of other substances and medications was identified as risk factors for OD in young people, but were not replicated (Calvo et al. 2017). Calvo et al. (2017) found that those who reported using novel drugs such as phencyclidine, hallucinogens, club drugs, and cathinone, at least once in the month before enrollment, were more likely to have a history of OD than those who did not recently use novel drugs (AOR: 2.71; 95% CI 1.04–7.05, p = 0.041).

Polysubstance use

An association between polysubstance use and OD was noted in three of the seven studies that examined it. Ochoa et al. (2005) examined the risk of using heroin with other drugs and found that individuals who mixed heroin with methamphetamine were at increased risk of experiencing an OD (p < 0.01). Yule et al. (2018) found that individuals with two or more diagnosed SUDs were three times more likely than those with one SUD to have a history of OD (p < 0.05). Similarly, in a study of unstably housed young people, Richer et al. (2013) found that those who reported using three or more types of drugs in the past 6 months were more than four times likely to experience an unintentional OD (AOR: 4.06; 95% CI 2.29–5.83, p < 0.0001).

Psychiatric risk factors

Psychiatric risk factors associated with OD in young people were less commonly explored. Several studies examined psychiatric symptoms, but not diagnosis, and OD risk and three of those studies found a significant association between the two variables (Burns et al. 2004; Calvo et al. 2017; Richer et al. 2013). Richer et al. (2013) found that subjects who endorsed suicidal ideation were almost twice as likely to have an unintentional OD during follow-up periods than those without suicidal ideation (AOR: 1.88; 95% CI 1.23–2.54, p = 0.0004). Likewise, Burns et al. (2004) found that individuals who reported feelings of hopelessness, as assessed by the Beck Hopelessness Scale, were six times more likely to report a history of OD (p < 0.01). Furthermore, Calvo et al. (2017) found that those who experienced symptoms of psychological distress were nearly 10 times more likely to experience an OD (AOR: 9.71; 95% CI 1.23–76.21, p = 0.031).

Three studies found a significant association between psychiatric diagnoses and OD, and the only specific diagnosis that was replicated as a risk factor was depression (Burns et al. 2004; Chahua et al. 2014;Yule et al. 2018). Burns et al. (2004) found that history of a psychiatric disorder was associated with a history of OD (p < 0.05), but did not specify which psychiatric diagnosis. Similarly, Chahua et al. (2014) found a positive association between major depressive disorder and history of OD (AOR: 2.2; 95% CI 1.01–4.74, p < 0.05) in young people who reported injection heroin use, but did not assess for other psychiatric characteristics or diagnoses.

Yule et al. (2018) was the only group to examine a wide range of psychiatric diagnoses, including attention-deficit/hyperactivity disorder, anxiety, depression, bipolar disorder, mood disorder (not otherwise specified), psychosis, and eating disorders. They reported significant associations between depression, anxiety, and eating disorders with OD. They also found a significant association between eating disorders and unintentional OD specifically (OR: 5.41; 95% CI 1.28–24.0, p ≤ 0.01). Furthermore, Yule et al. also found that psychiatric symptoms, including a history of suicide attempt, self-injurious behavior, and abuse (emotional, physical, and/or sexual), were positively associated with OD.

Medical risk factors

Risky substance use behavior is associated with increased medical morbidity. Despite this, only 2 of the 12 studies in this review examined medical characteristics and OD risk (Ochoa et al. 2005; Chahua et al. 2014). The medical characteristics examined by the two studies involved infectious diseases, which may have been chosen as variables to examine because of the association between injection drug use and increased risk for infectious diseases. One study found a significant association between being tested for hepatitis B or C and recent opioid OD (Ochoa et al. 2005).

Psychosocial risk factors

In our review, psychosocial risk factors, including history of being unstably housed or incarcerated and ever witnessing an OD, were found to be significant risk factors for OD in young people, which were replicated. Two of the five studies that examined the association between history of incarceration and OD found it to be a significant risk factor. One study found a lifetime history of imprisonment in young people was associated with a fourfold increased risk for OD (AOR: 4.1; 95% CI 1.4–12.1, p < 0.05) (Chahua et al. 2014). The other study found that individuals who reported being incarcerated for at least 20 months were almost three times more likely to experience an OD (AOR: 2.99; 95% CI 1.52–5.88, p < 0.01) when compared to those who had never been incarcerated. No association with OD was seen between individuals who reported being incarcerated for 20 months or less (Ochoa et al. 2005).

There were nine studies that examined the association between unstable housing and OD. Being unstably housed was defined differently in each study, but was typically distinguished by living on the street or another place not intended for human habitation, living in a shelter or hotel, or not having a fixed address. Three studies found that unstable housing was a significant risk factor for OD (Sherman et al. 2007; Richer et al. 2013; Calvo et al. 2017). Recent homelessness (in the past 6 months) was associated with a history of OD in two studies (Sherman et al. 2007; Richer et al. 2013). Likewise, young people who report being unstably housed (living in a shelter or hotel) or homeless (living in a public space) in the past 3 months were four times more likely to report a history of OD (AOR: 4.39; 95% CI 0.96–20.02, p = 0.056) (Calvo et al. 2017).

Two of the four studies that examined the relationship between witnessing OD and personally experiencing an OD in individuals who use opioids, found a significant association between the two variables (Ochoa et al. 2005; Silva et al. 2013). Silva et al., (2013) found that ever witnessing a family member OD was independently associated with a lifetime history of OD (AOR: 1.59, 95% CI 1.02–2.50, p < 0.05). Ochoa et al. (2005) found that those who reported ever having witnessed an OD were nearly three times more likely to have experienced a heroin OD in the past year (AOR: 2.89, 95% CI 1.76–4.73, p < 0.01).

The quality of an individual's relationship was another psychosocial characteristic examined by one study. Burns et al. (2004) found that relationship dissatisfaction was significantly associated with an increased risk for OD (p < 0.01).

Sexual orientation was examined as a risk factor in three studies (Ochoa et al. 2001, 2005; Silva et al. 2013) and found to be a significant predictor of OD in one study. Ochoa et al. (2001) found that individuals who reported engaging in bisexual or homosexual behavior were 2.18 times more likely to have a history of OD compared to those who reported engaging in heterosexual behavior only (p = 0.05).

Treatment history

History of SUD or psychiatric treatment was infrequently collected. When treatment history was examined as a risk factor for OD, SUD and psychiatric treatment were considered separately. In other words, when history of SUD treatment was examined, no details on psychiatric symptoms at the time of treatment were reported, and likewise, when psychiatric treatment was examined, no details about SUD symptoms at the time of treatment were reported.

Three studies found that young people with a history of psychiatric or SUD inpatient treatment were at an increased risk for OD (Silva et al. 2013; Calvo et al. 2017; Yule et al. 2018). Silva et al.(2013) found that a lifetime history of psychiatric hospitalization was associated with OD (AOR: 1.79; 95% CI 1.12–2.85, p < 0.05). Yule et al. (2018) also found an association between lifetime history of psychiatric hospitalization and OD (OR: 3.14; 95% CI 1.66–5.93, p ≤ 0.01). Calvo et al. (2017) found young people with a history of an inpatient admission for substance detoxification were more than four times likely to have experienced an OD when compared to those with no history of inpatient substance detoxification (p = 0.001). Yule et al. (2018) also found that young people who had a history of medication treatment for an SUD were more likely than those without treatment to have a history of OD (p < 0.05).

Discussion

Our review identified opioid, tranquilizer, and injection drug use as important and replicated risk factors for drug OD in young people, which have also been associated with OD in older adults. Our review also identified that polysubstance use, psychiatric comorbidity, witnessing an OD, and being unstably housed are risk factors that are especially noteworthy in young populations. Our work also highlights the lack of information that exists on the relationship between medical and treatment history characteristics and drug OD risk, and the absence of research regarding risk factors for fatal OD in this age group.

Lifetime prevalence of study subjects who reported experiencing drug OD (n = 7 samples) ranged from 24% to 48% with a mean of 34%. Although these populations varied in terms of substance use severity, all cohorts were made up of high-risk individuals who reported opioid, stimulant, prescription, or injection drug use, were seeking treatment for a SUD, engaged in a harm reduction program for substance use, or were homeless. These numbers are slightly lower than what was reported in a systematic review of adult literature, by Martins et al. (2015) who found that the prevalence of drug OD ranged from 16.6% to 68% with a mean 45.4% and standard deviation of 14.4% (n = 26 samples). These findings are consistent with national epidemiologic data that show drug OD is more prevalent in populations of older adults (Seth et al. 2018), but nevertheless, highlights the high risk for drug OD in young people.

Substance use characteristics consistently associated with drug OD in young people included heroin, prescription opioid, tranquilizers (including benzodiazepines), polysubstance, and injection drug use. The association between opioid, tranquilizer, and injection drug use with increased risk for drug OD in young people is consistent with findings in the adult literature (Martins et al. 2015), and highlights the increased morbidity and mortality associated with riskier substance use and more severe SUD even in young people.

The association between polysubstance use and increased risk for drug OD in young people may be more specific to this age group since it was associated with drug OD in three of the seven studies that examined it as a risk factor (Ochoa et al. 2005; Richer et al. 2013; Yule et al. 2018), and has only been identified as a risk factor for nonfatal OD in one study of older adults (Betts et al. 2015). It is not surprising that polysubstance use emerged as a risk factor associated with OD in young people since this group are more likely than older adults to experiment with using multiple substances (Connor et al. 2014).

The relationship between stimulant use and drug OD risk in young people was not consistent across studies, with some studies demonstrating increased risk associated with stimulant use and other studies showing no association. This finding is an area of difference in drug OD risk factors between young people and older adults, where an association between cocaine use and increased risk for OD has been more consistently demonstrated (Martins et al. 2015). The equivocal relationship between stimulant use and drug OD risk in young people may also reflect heterogeneity in the stimulant class and patterns of stimulant use since the relationship between use and OD risk was evaluated in range of stimulants from prescription stimulant misuse to cocaine and methamphetamine use. Further research is needed to evaluate the relationship between stimulant use and drug OD risk in young people, in light of increased crystal methamphetamine use across all age groups (Degenhardt et al. 2017), and a recent 44% increase between 2015 and 2016 in drug OD deaths from cocaine in individuals 15–24 years of age (Seth et al. 2018).

The association found between drug OD in young people and psychiatric comorbidity, including depression and associated symptoms of mood dysregulation, is also consistent with findings in older adult populations, and a broader literature on risks associated with self-harm and suicidality in young people. A meta-analysis of the association between depression and drug OD risk in older adults found that adults with depression were 50% more likely to have an OD when compared to adults without depression (Bartoli et al. 2014). One large registry study from 2017 showed children and adolescents 10–19 years of age, who reported a history of self-harm, were much more likely than those who did not report self-harm to die from fatal alcohol or drug poisoning (AOR: 34.3, CI: 10.2–115.7).(Morgan et al. 2017) Another study of adolescents psychiatrically hospitalized for suicidality found 53% of later deaths in this sample were due to drug OD (King et al. 2019).

The associations observed between depression and associated symptoms of mood dysregulation and drug OD highlight the need for increased research that examines risk factors for unintentional and intentional OD in the same sample. The intentionality associated with drug OD is less likely binary, unintentional versus intentional, and more likely exists on a continuum of intent for self-harm associated with high-risk substance use.

Our findings on psychosocial risk factors associated with drug OD further confirm what has been reported in the adult literature. Incarceration has been hypothesized to increase risk for drug OD since individuals who were using substances regularly lose their tolerance for substances while incarcerated, and are at high risk of relapse when released (Binswanger et al. 2013). Our review found that being in jail or prison was significantly associated with drug OD, and one study found an association between longer duration of incarceration and OD risk, suggesting generalizability of this risk factor across age groups.

Unstable housing also appears to place young people at an increased risk for drug OD. For instance, a third of the studies found significant associations between being recently unstably housed or homeless and having experienced a drug OD, and the lifetime prevalence of OD in this population was particularly high. Although homelessness has been identified as a risk factor in populations of adults, elevated rates of homelessness among young people and the high prevalence of drug OD in this group indicate that unstable housing may be a particularly important risk factor to consider in young people.

Interestingly, four studies examined the relationship between witnessing a drug OD and personally experiencing an OD, and two found a significant association between the two variables. To our knowledge, such findings have not been identified in adult literature. Witnessing a drug OD may be a risk factor that has to date only emerged in young populations since young people are more likely than older adults to engage in risky behavior with their peers, including risky substance use (Gardner and Steinberg 2005).

Several important gaps in the literature were identified in this review. First, the studies included in this review only focused on nonfatal drug OD. It is possible that there are differences between risk factors for nonfatal and fatal drug OD in young people and the findings in this review may not generalize to all types of OD. Likewise, only 3 of the 12 studies examined and reported on the intentionality of drug OD. It is possible that the studies that did not assess drug OD intentionality included samples that contained individuals who experienced an intentional OD (suicide attempt) and those who experienced an unintentional OD. More research is needed to discern differences in risk between unintentional and intentional drug OD.

Next, since all of the studies focused on high-risk samples of young people who regularly use substances or are homeless, the findings in this review may not generalize to all groups of young people, including younger adolescents, with periodic substance misuse. In addition, because all of the studies took place in urban areas, these findings may not be applicable to young people who live in suburban or rural communities. Such considerations are particularly important when considering opioid OD, as rural areas have been particularly impacted by the opioid crisis (Rigg et al. 2018).

Furthermore, although the relationship between race and drug OD was commonly examined, all but one of the studies that considered race as a risk factor was primarily White cohorts. Although most drug OD deaths occur in White populations, recent data show that OD deaths are increasing most rapidly in Black and Hispanic populations (Seth et al. 2018). Future research should continue to examine the relationship between race and drug OD to identify populations that may be particularly vulnerable to OD.

Also, history of drug OD has been identified in the adult literature as a risk factor for subsequent OD, yet little longitudinal research examining risk factors for OD in young people exists. For instance, only 3 of the 12 articles in our review include longitudinal data, and none of these studies examined the association between history of drug OD and future OD, despite the high prevalence of OD history found at baseline. Considering that nonfatal drug OD is a known risk factor for fatal OD in adult populations (Caudarella et al. 2016), it is important for future longitudinal studies of young people to examine this relationship.

Our review also highlights the dearth of information regarding the relationship between SUD treatment history and drug OD in young people. The two studies that reported on relationships between SUD treatment and drug OD (Calvo et al. 2017; Yule et al. 2018) identified a history of admission for substance detoxification and a history of medication for SUD as risk factors for OD. While history of admission for substance detoxification is a risk factor for drug OD that has been replicated in adult populations (Green et al. 2009), treatment with medication for opioid use disorders has been found to be a protective factor against OD and death (Volkow et al. 2014; Larochelle et al. 2018). Since Yule et al. (2018) looked at lifetime history of treatment with medication for SUD and did not link the timing of medication treatment with drug OD, the association seen between medication and OD in this study may reflect a higher severity of SUD since medication for SUD is less commonly used in young people. Furthermore, since young people with an opioid use disorder who are treated with buprenorphine are less likely to be abstinent from opioids when compared to older people (Schuman-Olivier et al. 2014), it is important to continue to evaluate the relationship between medication and drug OD, as well as ways to improve adherence to medication and engagement in treatment in young people with SUD.

Similarly, in this review, only two studies examined the relationship between medical characteristics and drug OD, and these studies only considered HIV and hepatitis B and C diagnosis as potential risk factors. Medical characteristics such as chronic pain and liver, renal, vascular, or pulmonary impairments have been identified as risk factors for drug OD in adult populations (Wilder et al. 2016; Nadpara et al. 2017), and it is possible that similar risk factors exist for young people. Future research in young people should examine the relationship between medical characteristics and drug OD.

There are several limitations to our review. First, differences in study design must be considered when drawing comparisons across studies. For example, there is no standard definition of drug OD and each study imposed its own definition of OD that varied in type of substance used and severity of OD symptoms. Similarly, not all studies examined drug OD within the same time frame—while some studies collected information on lifetime history of OD, others recorded data on OD history in the previous 6 or 12 months. In addition, the substantial variation in the patient populations of the studies must be considered. Six of the 12 studies included in this review consisted of subpopulations of young people who used substances associated with greater risk for drug OD like opioids or injected drugs, and these subpopulations may have higher prevalence of OD compared to more heterogenous populations of people who use drugs.

Despite methodologic differences between studies, and existing knowledge gaps, our systematic review of risk factors for drug OD in young people reveals trends that are important for clinicians and public health officials to consider when creating targeted intervention strategies for young people. Use of specific substances, including opioids, tranquilizers, and injection drug use, emerged as the strongest risk factors for drug OD in young people. Polysubstance use, psychiatric comorbidity, recent incarceration, unstable housing, and history of witnessing an OD should also be considered when assessing for drug OD risk in young people. Although further research is needed to clarify and expand on these risk factors, this work is crucial to understanding drug OD in young people and has important implications for prevention and risk reduction strategies. Considering the high prevalence of psychiatric comorbidity and rising prevalence of drug OD in young people, further research is needed to assess risk factors for unintentional and intentional drug OD as well as fatal drug OD that are specific to this population.

Conclusion

In our systematic review of risk factors for drug OD in young people, we found polysubstance use, psychiatric comorbidity, unstable housing, and witnessing an OD are noteworthy risk factors specific to young people. Similar to findings in older adults, opioid, tranquilizer, and injection drug use are associated with increased risk for OD in young people.

Clinical Significance

Drug OD is one of the current leading causes of death in young people and recent data suggest that the rate of drug OD is continuing to increase. Although literature has identified risk factors for drug OD in adult populations, there is a paucity of research on young people specifically. Young people may be uniquely at risk due to differences in brain development, increased prevalence of psychiatric comorbidity, stronger genetic influence for SUD, and limited access to SUD treatment, including medication treatment. This review summarizes the existing literature on risk factors for drug OD in young people, highlighting risk factors specific to this age group, which can be used to inform targeted interventions for this vulnerable population.

Disclosures

Dr. A.M.Y., MD, received grant support from the Massachusetts General Hospital Louis V. Gerstner III Research Scholar Award from 2014 to 2016. Dr. A.M.Y. is currently receiving funding through the American Academy of Child and Adolescent Psychiatry Physician Scientist Program in Substance Abuse K12DA000357-17. She was a consultant to the Phoenix House from 2015 to 2017 and is currently a consultant to the Gavin House (clinical services). Dr. T.E.W., MD, receives or has received grant support from the following sources: NIH(NIDA). Dr. T.E.W. is or has been a consultant for Alcobra, Neurovance/Otsuka, and Ironshore. Dr. T.E.W. has a published book: Straight Talk About Psychiatric Medications for Kids (Guilford Press); and co/edited books ADHD in Adults and Children (Cambridge University Press), Massachusetts General Hospital Comprehensive Clinical Psychiatry (Elsevier) and Massachusetts General Hospital Psychopharmacology and Neurotherapeutics (Elsevier. Dr. T.E.W. is co/owner of a copyrighted diagnostic questionnaire (Before School Functioning Questionnaire). Dr. T.E.W. is Chief, Division of Child and Adolescent Psychiatry and (Co) Director of the Center for Addiction Medicine at Massachusetts General Hospital. He serves as a clinical consultant to the US National Football League (ERM Associates), U.S. Minor/Major League Baseball; Phoenix/Gavin House and Bay Cove Human Services. Dr. D.S., MD, MSc, is currently supported by NIDA: K12DA043490. Dr. S.M.B., MD, MSc, received grant support from NIDA, 1K23DA044324-01. In addition, she is a consultant to the Center for Evidence in Health at Brown University School of Public Health on an AHRQ grant, and to WBGH for an educational curriculum about opioids. R.M.L., BS: nothing to disclose at this time.