Multiple Substance Use Disorders from Adolescence to Midlife

Abstract

Objective:

To assess associations between symptomatic polysubstance use and symptom severity during adolescence and later substance use disorder (SUD) symptoms in midlife.

Methods:

Multi-cohorts of nationally representative US adolescents in twelfth grade (N=12025; baseline cohort years 1976–2002) were followed longitudinally to age 55 (follow-up years 1993–2019). Poisson regression models were fitted longitudinally, accounting for repeated measures, panel attrition, and covariates.

Results:

An estimated 8.9% (95% CI=8.3%–9.5%) of US adolescents had moderate-to-severe SUD symptoms involving multiple substances (ie, 4+ SUD symptoms for 2+ substances). The bivariate and multivariate relationships between SUD symptom severity and multiple SUDs at baseline and subsequent SUD symptoms in midlife (ages 35–55) indicated a positive near-linear association. Over two-thirds of adolescents with moderate-to-severe SUD symptoms involving multiple substances had multiple SUD symptoms in midlife (68.6%, 95% CI=63.0%–73.7%). Most adolescents with moderate-to-severe SUD symptoms of one substance and those with fewer SUD symptoms involving multiple substances (ie, 1–3 SUD symptoms for 2+ substances) persisted with multiple SUD symptoms in midlife (57.2%, 95% CI=53.2%–61.0% and 56.4%, 95% CI=51.4%–61.4%, respectively). Adolescents with moderate-to-severe SUD symptoms involving multiple substances had substantially greater odds of SUD symptoms in midlife relative to asymptomatic adolescents (aRR=2.80, 95% CI=2.49–3.15).

Conclusions:

Most US adolescents with symptomatic polysubstance use or moderate-to-severe SUD symptoms persisted with multiple SUD symptoms in midlife. Adolescent symptomatic polysubstance use is associated with a more severe developmental course in midlife than those with the same SUD symptom burden concentrated within a single substance, therefore, early comprehensive screening and intervention strategies are needed.

INTRODUCTION

National studies from the past decade have found that over one-third of US adults developed at least one DSM-5 substance use disorder in their lifetime.^1,2 A substantial proportion of US adults with a non-alcohol SUD (eg, cannabis, stimulant, opioid) have multiple SUDs.^3,4 For many, symptomatic substance use, polysubstance use, and SUD onset begin during adolescence.^4–7 One national study found that US adults with multiple SUDs had a more persistent and severe natural disease course versus those with one SUD over a three-year period.⁸ There is a need for more longitudinal national studies that examine whether adolescent SUD symptoms spanning multiple substances are a stronger predictor of adult SUD symptoms than symptoms from a single substance, even after accounting for SUD symptom severity. This study aims to assess associations between symptomatic use involving multiple substances and symptom severity during adolescence and subsequent SUD symptoms in midlife through age 55.

METHODS

The Monitoring the Future (MTF) Panel is a US national longitudinal multi-cohort study from ages 18 to 55.⁹ Twenty-seven cohorts of nationally representative samples of high school seniors were surveyed at baseline (modal age 18), with follow-ups through age 55 (follow-up years 1993–2019). Baseline response rates over the study period (1976–2002) were 77–86%, with most non-response due to absence (< 1% refused participation). The average unweighted retention rate was 53%. Parents received a waiver of informed consent regarding their child’s participation. This secondary analysis of deidentified data was deemed exempt by the University of Michigan’s IRB. The MTF study design, protocol, and sampling methods are described in greater detail elsewhere.⁹

Baseline covariates included cohort year, sex, race/ethnicity, geographical region, urbanicity, parental education, and school performance. SUD criteria (ie, symptoms) at baseline were measured separately for alcohol use disorder (AUD), cannabis use disorder (CUD), and other drug use disorder (ODUD, ie, opioids, stimulants, and other drugs). Consistent with SUD measurements in other large-scale surveys, the same 15 items assessed DSM-IV and DSM-5 SUD symptoms, including failure to fulfill major role obligations, continued substance use despite persistent or recurrent interpersonal or social problems, and continued substance use when physically or psychologically hazardous (Supplemental Table 4).^10–13 Reliability for these 15 items were strong (AUD, α=.813; CUD, α=.846; ODUD, α=.885). As any SUD symptoms in adolescence are linked to SUD in adulthood,⁶ the 15 items were summed and collapsed as follows: asymptomatic (0 symptoms), low/mild (1–3 symptoms), and high/severe (4+ symptoms).⁶ These data were combined with the number of SUDs to create a mutually-exclusive five-category variable indicating no symptoms; single SUD with low symptoms; multiple SUDs with low symptoms; single SUD with high symptoms; and multiple SUDs with high symptoms. The SUD outcome in midlife (35–55 years) was 2+ symptoms over the past 5 years (out of 8 symptoms), measured with similar items to those used at baseline for AUD, CUD, and ODUD (see Supplemental Table 4). We followed recommended practice for coding SUD criteria at baseline and midlife,¹⁰ resulting in estimates closely resembling other national estimates for similar age groups.^1,2,4

Descriptive statistics and unadjusted risk ratios were generated in Mplus Version 8.11 to examine the bivariate association between SUD symptom severity and type at age 18 and the subsequent prevalence of 2+ SUD symptoms at midlife (ages 35–55). Second, Poisson regression models were fitted longitudinally, accounting for repeated measures collected between ages 35–55, to assess associations between the number of SUDs (single or multiple) and severity of SUD symptoms (low or high) at age 18 and the past five-year prevalence of SUD symptoms during this 20-year period (ages 35–55) in middle adulthood when accounting for key covariates. All bivariate and multivariate analyses used attrition weights to account for respondent characteristics associated with non-response at follow-up, consistent with other MTF panel analyses.⁹ Missing data was handled using full information maximum likelihood.

RESULTS

The longitudinal panel sample consisted of 12,025 individuals; an estimated 50.6% (95% CI=49.7%–51.5%) were female. Table 1 and Supplemental Table 1 show sociodemographic characteristics and SUD symptom severity of the longitudinal panel sample. An estimated 8.9% (n=875) of adolescents had moderate-to-severe SUD symptoms involving multiple substances (ie, 4+ SUD symptoms for 2+ substances). Between ages 35–55, 39.2% (n=2,676) of adults reported 2+ SUD symptoms (Supplemental Table 2). The bivariate and multivariate relationships between SUD symptom severity and multiple SUDs at baseline and subsequent SUD symptoms in midlife indicated a positive and near-linear association (Table 2). More than two-thirds of adolescents with moderate-to-severe SUD symptoms involving multiple substances reported multiple SUD symptoms in midlife (68.6%, 95% CI=63.0%–73.7%). Most adolescents with moderate-to-severe SUD symptoms involving a single substance and most with fewer SUD symptoms involving multiple substances (ie, 1–3 symptoms for 2+ substances) reported multiple SUD symptoms in midlife (57.2%, 95% CI=53.2%–61.0% and 56.4%, 95% CI=51.4%–61.4%, respectively). Any adolescent symptomatic substance use was associated with a significantly greater adjusted risk of midlife SUD symptoms, with SUDs involving multiple substances carrying a greater risk than those involving a single substance. Adolescents with moderate-to-severe SUD symptoms involving multiple substances had nearly three times greater risk of multiple SUD symptoms in midlife relative to asymptomatic adolescents (aRR=2.80, 95% CI=2.49–3.15). There were no meaningful interaction effects/cohort effects (Supplemental Table 3).

Table 1.

Baseline Characteristics of Respondents in the Monitoring the Future Panel Data

Sample Characteristics at Age 18	n = 12,025 (%)
Cohort Year
1976–1984	4,345 (36.13)
1985–1993	3,975 (33.06)
1994–2002	3,705 (30.81)
Population Density Metropolitan Statistical Area (MSA)
Large MSA	3,194 (26.56)
Other MSA	5,454 (45.36)
Non-MSA	3,377 (28.08)
Region
Northeast	2,607 (21.68)
Midwest	3,402 (28.29)
South	3,881 (32.27)
West	2,135 (17.75)
Sex
Male	5,941 (49.41)
Female	6,082 (50.59)
Race / Ethnicity
Black	1,279 (10.78)
Hispanic	786 (6.62)
White	9,051 (76.26)
Other	752 (6.34)
Parental Education
At least one parent is a college graduate	4,807 (40.57)
Neither parent is a college graduate	6,826 (57.61)
Don’t know or does not apply	216 (1.82)
Average Grades in School
Average grade of B- or higher	8,608 (73.62)
Average grade of C+ or lower	3,085 (26.38)
Substance Use Disorder (SUD) Symptom Severity a
No SUD symptoms	4,530 (45.95)
Single substance with 1–3 SUD symptoms	1,706 (17.31)
Multiple substances with 1–3 SUD symptoms	1,075 (10.90)
Single substance with 4+ SUD symptoms	1,672 (16.96)
Multiple substances with 4+ SUD symptoms	875 (8.88)

All cell sizes (n) and percentages (%) are unweighted. At baseline, 12,025 12th graders were selected into the longitudinal sample between 1976 and 2002 who completed Form 3. It should be noted that 5,595 respondents did not complete any follow-up between ages 35 to 55. Baseline (age 18) covariates included sex (male or female), race/ethnicity (Black, Hispanic, White, or other; other defined as Asian, American Indian, those who selected multiple races or ethnicities), US Census geographical region (Northeast, Midwest, South, or West), urbanicity based on metropolitan statistical area (MSA; large MSA, 12–16 of the largest urban areas; other MSA, other cities or towns of at least 50000 inhabitants; or non-MSA, fewer than 50,000 inhabitants), parental education (neither parent graduated from a 4-year college or at least 1 parent graduated from a 4-year college), average grade in high school (C+ or lower or B− or higher), and cohort year. Race and ethnicity options were defined by the Monitoring the Future study team. Race and ethnicity were assessed in this study because of differences across behavioral and social outcomes. For the 1976–1977 cohorts, there was a high amount of missingness for all the SUD symptom questions. This was corrected by replacing missing values with 0s for respondents who answered all the lifetime substance use questions (alcohol, marijuana, and 10 other substances) but had missingness for the SUD symptom questions.

^aThe SUD symptom severity variable is based on form 3 items assessed at baseline (age 18) and refers to individuals with symptoms consistent with alcohol use disorder, cannabis use disorder, and other drug use disorder. Multiple substances refers to at least two drug classes.

Table 2.

Adolescents’ Substance Use Severity and Later Substance Use Disorder Symptoms in Adulthood

	2+ Substance Use Disorder Symptoms in Adulthood (Ages 35–55)
	n	% (95% CI)	uRR (95% CI)	aRR (95% CI)a
			n = 6,430	n = 6,430
			Obs = 18,666	Obs = 18,666
SUD Symptom Severity at Age 18 b
No SUD symptoms	614/2322	25.80 (23.79, 27.92)	Ref	Ref
Single substance with 1–3 SUD symptoms	413/955	42.83 (39.26, 46.47)	1.63 (1.44, 1.84)	1.57 (1.40, 1.78)
Multiple substances with 1–3 SUD symptoms	348/570	56.42 (51.36, 61.35)	2.32 (2.05, 2.62)	2.21 (1.96, 2.50)
Single substance with 4+ SUD symptoms	525/885	57.17 (53.22, 61.02)	2.28 (2.04, 2.54)	2.21 (1.98, 2.47)
Multiple substances with 4+ SUD symptoms	302/428	68.62 (62.97, 73.77)	2.95 (2.61, 3.32)	2.80 (2.49, 3.15)
Age Linear Trend c				0.90 (0.88, 0.92)
SUD Linear Trend c				1.10 (1.09, 1.12)

Abbreviations: SUD, substance use disorder; CI, confidence interval; uRR, unadjusted risk ratios; aRR, adjusted risk ratios; Obs, observations; Ref, reference.

Notes: Cell sizes (n) are unweighted. CIs incorporate panel analysis weights from the Monitoring the Future Study. Models were estimated using full information maximum likelihood to account for missing data.

^aFor adjusted model, baseline covariate variables include cohort year, population density, region, sex, race / ethnicity, parental education, and average grade in school. Unadjusted and adjusted models also control for a continuous age variable (0 = 35, 1 = 40, 2 = 45, 3 = 50, 4 = 55).

^bThe SUD symptom severity at age 18 variable is based on form 3 items assessed at baseline (age 18) and refers to individuals with symptoms consistent with alcohol use disorder, cannabis use disorder, and other drug use disorder. Multiple substances refers to at least two drug classes.

^cAn additional model was fitted to estimate adjusted linear trends for age and SUD (estimated in the same model), with SUD symptoms at age 18 ranging from 0 to 15 (the maximum number of symptoms across all substances).

DISCUSSION

This study offers novel insights into the course of SUD from adolescence through midlife as a function of the number of SUD symptoms and SUD severity. We found that more than one in every 12 US adolescents had moderate-to-severe SUD symptoms involving multiple substances; over two-thirds of these adolescents reported multiple SUD symptoms in midlife. These findings extend prior work indicating that relative to individuals with a single SUD or less severe SUD symptoms, those with multiple SUDs or moderate-to-severe SUD symptom severity in adolescence have a more persistent and severe course of multiple SUD symptoms in later adulthood.^6,8 Notably, SUD symptoms from multiple substances (versus a single substance) conferred greater risk of multiple midlife SUD symptoms, even at similar severity, suggesting adolescent polysubstance use confers particular risk.

These findings underline the importance of SUD symptom screening during adolescence, as early intervention is associated with better outcomes.^14,15 Screening, clinical diagnosis, SUD pharmacotherapy, and treatment often focus on drug-specific SUDs rather than multiple SUDs, which are more complex. Accounting both for SUD symptom severity and multiple SUDs is necessary to identify adolescents at greatest risk for persistent substance-related problems. Still, even low SUD symptom severity marked elevated risk of later substance-related problems, so any SUD symptoms warrant intervention.

Strengths include the MTF panel design, which is the largest national multi-cohort prospective study to assess SUD symptom severity and multiple SUDs from adolescence to ages 35–55 with consistent methods and measures. Limitations included the omission of subgroups with higher (eg, school dropouts and incarcerated) or lower rates of substance use (eg, homeschooled).⁹ Though attrition weights were used, higher attrition rates among adults who use substances frequently may have led to underestimates of more severe SUD symptom profiles over time. The study methods cannot establish formal DSM-based diagnoses, and validation studies are needed, especially since the ODUD category combined several drug classes.

CONCLUSIONS

Symptomatic polysubstance use during adolescence signals multiple SUD symptoms in midlife for most individuals. Screening and interventions for adolescents should assess symptomatic use of multiple substances and SUD symptom severity to identify and intervene with those at greatest risk for long-term persistent substance-related problems.