The Intensity of Adolescent Substance Use Before and After the COVID-19 Pandemic

Abstract

Introduction:

Adolescent cannabis and alcohol use declined substantially following the COVID-19 pandemic onset in 2020. This study tests whether substance use “hardened,” defined as an increased proportion of heavy users resulting from overall prevalence declines driven mainly by reductions in light and moderate use.

Methods:

Data come from Monitoring the Future, which annually surveys nationally representative, cross-sectional samples of U.S. adolescents in 8^th, 10^th, and 12^th grades. The analysis includes 252,810 respondents surveyed between 2017–2024, with response rates of 86% in 8^th grade, 85% in 10^th grade, and 76% in 12^th grade. The main outcome is lifetime intensity of cannabis and alcohol use, measured as “light” (1–5), “moderate” (6–19) and “heavier” (20+) occasions of lifetime use.

Results:

Following the pandemic onset, adolescent alcohol use softened notably, while cannabis use softened slightly. For alcohol, light use increased significantly across all three grades by 3–7 percentage points, moderate use decreased in all three grades by 2–4 points, and heavy use decreased 4 points in 12^th grade and 3 points in 10^th grade. For cannabis, changes were limited to 10^th grade, where light use increased by 3 points and moderate use decreased by 2 points.

Conclusions:

These results point to a continued need for prevention policies and programs to achieve further reductions in adolescent cannabis and alcohol use. Current patterns of use are not yet at a “hardened” stage dominated by heavy users who struggle to quit, where further reductions in prevalence would be achieved primarily through cessation programs.

Introduction

The prevalence of adolescent drug use declined markedly after the onset of the COVID-19 pandemic in 2020. The reductions observed from 2020 to 2021 rank among the largest one-year declines ever recorded in the past fifty years for the two most commonly used substances among adolescents of cannabis and alcohol.¹ These declines subsequently continued and as a result of their cumulative impact, lifetime use for each of these substances dropped by as much as one third in just four years.¹ The social distancing and stay-at-home policies implemented during the pandemic² likely played a major role in initiating these declines, as these measures greatly reduced adolescents’ opportunities to use and obtain drugs, as well as their exposure to peers who engage in drug use.

Little is known about changes in the intensity of adolescent drug use since the pandemic onset in 2020. According to the “hardening hypothesis,” as population prevalence of drug use declines, the average intensity levels among remaining users increases.³ Key to this hypothesis are individuals who use substances at levels that are light or experimental, or have a proclivity for such use. The hypothesis predicts that population declines in prevalence take place because these individuals are less likely to initiate use, and/or more likely to cease use. Over time, their absence results in a relatively greater proportion of heavy users, for whom quitting requires relatively more effort.

The competing hypothesis is that average intensity levels among users either decrease or do not change as prevalence levels decline. According to this hypothesis, social and cultural forces strong enough to reduce a health outcome in an entire population have rarely been limited to an influence on only low intensity levels.⁴ As prevalence declines, the increasing number of individuals who do not initiate use can also include those who otherwise would have quickly gone on to become moderate or heavier users. Similarly, while individuals with heavier levels of use may face more psychological and behavioral obstacles to quit substance use, they may nevertheless be successful in doing so at equal or greater levels than light experimenters. To date, evidence for softening is more abundant^5–12 than is evidence for hardening,^{13, 14} at least for the outcome of cigarette smoking.

This study tests whether intensity of adolescent cannabis use and alcohol use hardened, softened, or did not change as prevalence levels for these substances fell following the pandemic onset in 2020. This study assesses intensity using lifetime prevalence, categorizing users into the groups of light, moderate, and heavier lifetime use. In the context of the hardening hypothesis—which originated in research on cigarette smoking—’intensity’ refers to frequency and cumulative extent of use over time, and not quantity of a substance consumed on a single occasion. Lifetime prevalence serves as a key metric to distinguish low-intensity users, who have used a substance only a few times, from high-intensity users, whose high levels of lifetime use can indicate more entrenched behavioral patterns and stronger identification with substance use.

This study expands upon the original hardening hypothesis in three ways. First, it focuses on adolescents, thereby contributing to a growing body of work in this field that focuses on young people.^{12, 13} Second, it centers on changes in the size of the light/experimenter group, which contrasts with traditional work in this area that places primary emphasis on heavy levels of use. Third, it examines cannabis use and alcohol use, which are the two most common substances that adolescents use today.

Methods

Study Sample

Data come from the Monitoring the Future study. Every year the study draws nationally representative, cross-sectional samples of U.S. 8^th, 10^th, and 12^th grade students in public and private schools in the 48 contiguous states.¹ Each of the three grades is sampled separately and the study surveys only students in the target grade at each school, resulting in three independent, nationally representative samples every year. MTF was approved by the University of Michigan Institutional Review Board, approval #HUM00131235. Informed consent (active or passive, per school policy) was obtained from parents of students younger than 18 years and from the students who were aged 18 years or older.

The analysis uses data from surveys administered from 2017 to 2024. Student response rates during this period were 86% in 8^th grade, 85% in 10^th grade, and 76% in 12^th grade. The response rate for the question on cannabis use was 94% and for alcohol it was 92%. For the demographic variables the response rate was 94% for sex and race/ethnicity, and for parental education it was 86%. All results are univariate prevalence estimates that use all available data and do not impute for missing values. The sample size is 252,810 respondents, with a breakdown by grade and substance reported in the Appendix Tables.

Three survey developments that took place during the time period from 2017 to 2024 are important to note. First, the project transitioned from paper-and-pencil to electronic data collection in 2019. A randomized controlled experiment assigned a randomly-selected half of the 2019 schools to surveys completed on electronic devices and the other half to traditional paper-and-pencil questionnaires. Drug prevalence did not differ across the two survey modes,¹⁵ and in 2020 and afterwards all surveys used electronic data collection. Second, in 2021 many students were schooling remotely and completed the survey at home, having obtained survey instructions and the survey web site from their teachers. Third, in some years the project’s annual reports did not use a randomly-selected half of the cannabis questions because experimental updates to the wording of these survey questions resulted in significantly different drug prevalence levels. The analyses use the same exclusions so that the study results are directly comparable across years; specifically, the analyses exclude the experimental cannabis questions that appear on a randomly-selected half of the questionnaires in 8^th grade in 2021, 2022, and 2024 and in 10^th grade in 2021 and 2022.

Measures

Appendix Table 1 presents the survey question text and variable coding. The question for cannabis use reads “The next questions are about marijuana or cannabis (sometimes called pot, weed, or hashish). On how many occasions (if any) have you used marijuana (smoking, vaping, edibles)… (a) in your lifetime?” (see Appendix Table 1 for slight variations in the survey text wording over the study period). The question for alcohol reads “On how many occasions have you had any alcoholic beverages to drink – more than just a few sips… (a) in your lifetime” The analyses use groups of “light”, “moderate,” and “heavier” substance use to examine the change in their relative size during the overall declines in substance use prevalence after the pandemic onset. To ensure that each of these groups had sufficient sample sizes to produce stable estimates for each year of the study, both lifetime use of cannabis and alcohol consist of a “light” grouping for students who used the response categories “1–2 Occasions” or “3–5 Occasions,” a “moderate” group for use of “6–9 Occasions” or “10–19 Occasions,” and a “heavier” group of students for use of “20–39 Occasions” or “40 or More” occasions.

Statistical Analysis

The statistical analyses consist of proportions calculated in STATA version 18.5 and accounted for the survey’s complex sample design by using weights to produce nationally representative estimates with standard errors adjusted for clustering of outcomes within schools and within regional sampling strata. Tests of differences in proportions were conducted using two-tailed Wald tests.

Results

Table 1 presents the demographics of the sample (n=252,810), which is representative of the U.S. adolescent population from 2017–2024. The percent reporting Hispanic was 23.7, and 51.6% reported as non-Hispanic White, 13.9% as non-Hispanic Black, 4.4% as non-Hispanic Asian, 2.0% as non-Hispanic Indigenous, 0.5% as non-Hispanic Middle Eastern, and 3.9% as non-Hispanic multi-racial. The sample is 48.6% female and 48.8% male, with 2.6% reporting “other” or “prefer not to answer.” Fifty nine percent report at least one parent with a college degree.

Table 1:

Sample Descriptives

	Grades 8, 10, and 12 Combined (n = 252,810a)
Variable	Prevalence	95% C.I.
Year
2017	12.3	(10.8–13.9)
2018	12.3	(10.9–13.8)
2019	12.5	(11.2–14.0)
2020b	12.5	(10.1–15.5)
2021	12.5	(11.1–14.2)
2022	12.5	(11.1–14.2)
2023	12.5	(10.7–14.7)
2024	12.8	(10.9–14.9)
Race/Ethnicityb
Hispanic	23.7	(21.5–26.0)
Black	13.9	(12.6–15.3)
Asian	4.4	(3.8–5.0)
White	51.6	(49.3–54.0)
Indigenous Peoples	2.0	(1.8–2.3)
Middle-Eastern	0.5	(0.3–0.9)
Multi-racial	3.9	(3.5–4.3)
Sexc
Male	48.8	(48.2–49.4)
Female	48.6	(47.9–49.2)
Other or prefer not to answer	2.6	(2.4–2.8)
Parent has college education	58.0	(56.4–59.5)

^aSample size is unweighted. All other estimates use sampling weights

^bData collected 2020 was halted on March 13 due to pandemic concerns. The resulting sample size is about one-quarter the size of a regular data collection. All 2020 data collected before the implementation of social distancing policies. Respondents in 2020 are upweighted so that 2020 has the same relative influence as the other years.

^cBefore 2020 only “male” and “female” were listed as response categories to the question on sex. In 2020 and 2021 the option “other or prefer not to answer” was added, and in 2022+ “other” and “prefer not to answer’ were included as separate response categories

Figure 1 presents a bar chart of yearly estimates for both lifetime use and intensity levels for cannabis use from 2017 to 2024, with estimate details in Appendix Tables 2 and 3. In all grades overall lifetime prevalence significantly declined, and in 12^th and 10^th grades less than half of these declines resulted from a reduction in light use. In 12^th grade, a seven-point decline in lifetime use from pre- to post-pandemic levels (from 44.1% to 37.1%), consisted of a three-point decline in light use (from 16.9% to 14.2%) a two-point decline in moderate use (from 9.3% to 7.5%) and a two-point decline in heavier use (from 17.9% to 15.5%). In 10^th grade a ten-point decline in lifetime cannabis use from pre- to post-pandemic levels (from 32.5% to 22.3%) consisted of a four-point decline in light use (from 13.9% to 10.3%), a two-point decline in moderate use (from 6.7% to 4.3%), and a four-point decline in heavier use (from 11.9% to 7.7%). In 8^th grade a three-point decline in lifetime cannabis use (from 14.2% to 10.9%), consisted of a two-point decline in light use (from 8.2% to 6.3%) and a one-point decline for both moderate (from 2.8% to 2.0%) and light use (from 3.2% to 2.6%).

Figure 1:

Prevalence of Lifetime Use and Lifetime Intensity Levels for Cannabis Before and After the Pandemic Onset, by Grade (Estimate Details Presented in Appendix Table 2)

Figure 2 presents a parallel bar chart for lifetime alcohol use, which declined significantly in all three grades across the pre- and post-pandemic time periods. Notably, very little of this decline resulted from changes in light use. In 12^th grade a five-point decline in lifetime use from pre- to post-pandemic levels was driven by a three-point decline in moderate use (from 19.7% to 16.5%) and a four-point decline in heavier use (from 18.5% to 15.0%). Light use after the pandemic onset significantly increased from pre- to post-pandemic levels, by two points (from 21.5% to 23.4%). Much of this increase was driven by a one-year spike in lifetime use in 2021 that is apparent in all grades and at all intensity levels, although in 12^th grade all levels of light use in each of the post-pandemic years 2021–2024 are near or above the yearly levels in 2017–2020. In 10^th grade a seven-point decline in lifetime use from pre- to post-pandemic levels (from 43.2% to 36.0%), consisted of a four-point decline in moderate use (from 14.0% to 10.4%) and a three-point decline in heavier use (from 8.6% to 6.1%); levels of light use did not significantly change over this period. In 8^th grade a three-point decline in lifetime use from pre- to post-pandemic levels (from 23.8% to 21.1%) consisted of a one-point decline in each of the three intensity levels of light use (from 14.4% to 13.3%), moderate use (from 6.5% to 5.3%) and heavier use (from 2.9% to 2.4%).

Figure 2:

Prevalence of Lifetime Use and Lifetime Intensity Levels for Alcohol Before and After the Pandemic Onset, by Grade (Estimate Details Presented in Appendix Table 2)

Figure 3 and Appendix Table 4 present the distribution of light, moderate, and heavy use for lifetime cannabis use and lifetime alcohol use in the pre- and post-pandemic time periods. These three intensity levels are exhaustive and within each time period the percentages add up to 100%. For cannabis use, lifetime intensity levels were little changed in the pre- and post-pandemic time periods. In 12^th and 8^th grade no significant changes in intensity levels took place. In 10^th grade and light use increased by three points (from 42.8% to 46.2%) while moderate use decreased by two points (from 20.7% to 19.2%).

Figure 3:

Prevalence of Intensity Levels among Ever Users, By Substance, Grade, and Time Period

* p<.05, significant difference across pre- and post-pandemic estimates

** p<.01, significant difference across pre- and post-pandemic estimates

Substantial changes took place in the distribution of intensity levels for lifetime alcohol use. In all three grades among youth who had ever used alcohol, levels of light use significantly increased while levels of moderate and/or heavier use significantly decreased. In 12^th grade light use increased seven points (from 36.1% to 42.6%), while moderate use decreased three points (from 33.0% to 30.0%) and heavier use decreased four points (from 31.0% o 27.3%). In 10^th grade light use also increased seven points (from 47.5% to 54.3%) while moderate use decreased four points (from 32.5% to 28.8%) and heavier use decreased three points (from 20.0% to 16.9%). In 8^th grade light use increased three points (from 60.5% to 63.4%) while moderate use decreased two points (from 27.2% to 25.1%).

Discussion

This study examined if intensity levels of adolescent cannabis and alcohol use changed after the onset of the pandemic in 2020 and prevalence levels of these substances fell. Contrary to the “hardening hypothesis,” the results do not support increasing intensity of use. Instead, cannabis intensity levels were little changed. Alcohol intensity levels “softened” notably; specifically, among youth who had ever used alcohol, light levels of alcohol use increased while moderate and heavier levels of use decreased.

From a policy perspective, these results indicate prevention will play a key role to further reduce adolescent cannabis and alcohol use. Adolescent use of these substances remains high, with post-pandemic lifetime use among 12th graders exceeding one in three for cannabis and reaching a majority for alcohol. Current patterns of use have not yet reached the stage where the primary focus to reduce these prevalence levels should be on cessation, which could be the case if substance use had “hardened” and primarily involved adolescents with a substantial history of high frequency use who struggle to quit. Instead, the proportion of adolescents with light levels of use remained the same or even increased following the declines after the pandemic, suggesting an ongoing need for programs and policies that aim to reduce the size of this group. These prevention efforts include state-mandated drug education prevention programs in primary and secondary schools,^16–18 media campaigns,^{19, 20} as well as more localized prevention efforts in regions and cities.^{21, 22}

Of course, cessation programs should remain a vital component of any comprehensive strategy to reduce adolescent substance use.⁸ Interventions are needed to help adolescents who regularly use substances and consequently impair their social and physical health. The size of this group is substantial, with 627,000 U.S. youth aged 12–17 receiving substance use treatment in 2023, and the number with unmet needs estimated to be much higher.²³ However, the study results suggest that prioritizing this group for interventions is not yet the most effective approach for achieving further declines in adolescent drug use at the population level.

This study contributes to existing work on “hardening” in two ways. First, its focus on adolescents differs from the field’s traditional emphasis on adults.²⁴ This focus on adolescents would be expected to inherently favor results in support of softening. Adolescents have had relatively less time than adults to develop entrenched habits of substance use or to reach cumulative use levels that lead to dependence and addiction. Consequently, they are less likely to exhibit the “hard core” use patterns that are central to the hardening hypothesis; that is, individuals with a prolonged history of substance use who find it difficult to quit.

Second, in a substantial departure from most work on the hardening hypothesis, the analyses of this study focus on light use instead of the traditional emphasis on heavy, “hard core” use. This departure offers two advantages. First, it facilitates the expansion of research beyond cigarette smoking to include a broader range of substances. Light use, characterized as lifetime use limited to handful of occasions, is an intuitive concept that readily applies to use of all substances. In contrast, heavy, “hard core” use likely differs across substances and it would take considerable time to forge a separate consensus on its definition for each specific substance, assuming a consensus is possible.

A second advantage of the focus on light use is its direct policy implications, although in a way that is 180 degrees different than the original intention of the hardening hypothesis. Evidence that the proportion of light use has increased or has not changed directly supports the importance of sustaining prevention efforts after a decline in prevalence. This perspective shifts away from advocating for increased cessation efforts, which is a primary policy take-away associated with the hardening hypothesis. Ultimately, the concepts of softening and hardening are at the opposite ends of the same continuum. Evidence for either one – either with a focus on light use or a focus on heavy use – is important to provide empirical, scientific evidence to inform policy on adolescent substance use.

Limitations

Two limitations/caveats for this study are important to note. One limitation is that in 2021 a substantial number of students completed the MTF survey while schooling at home, where they were sheltering during the COVID-19 pandemic.² The sudden and unexpected onset of the pandemic precluded a randomized-controlled test to examine if taking the survey at home affected MTF estimates. However, the analysis shows that the study’s main findings on intensity of use after the pandemic onset persisted in 2022 and afterwards, when 95%+ of students took the survey in their school buildings. This suggests minimal, if any, impact of at-home administrations of the MTF survey on the substantive conclusions of this study.

A second limitation is that the study does not analyze nicotine vaping, which is a recent addition to the top three most commonly used substances by adolescents.^{1, 25, 26} Two considerations suggest vaping may warrant its own study. First, the survey measure to assess lifetime nicotine vaping employs qualitatively different response options that are not comparable to those for lifetime cannabis and alcohol use. Specifically, for lifetime nicotine vaping the response options include “occasionally but not regularly,” “regularly in the past” and “regularly now,” whereas the options for lifetime cannabis and alcohol use focus on the number of use occasions. Second, the FDA enacted new federal policies specifically intended to reduce adolescent vaping during the study period,²⁷ and disentangling the independent effect of these policies on vaping intensity would require substantial attention in a study’s Introduction, analysis, and Discussion, hence underscoring the need for a dedicated study.

Conclusions

Adolescent use of cannabis and alcohol did not “harden” as prevalence levels of these substances fell following the onset of the COVID-19 pandemic. Intensity of use remained largely unchanged for cannabis use, while alcohol use substantially softened and shifted toward lighter use and away from heavier use. These results point to a need for prevention policies and programs to achieve further reductions in adolescent cannabis and alcohol use. Current patterns of use are not yet at a “hardened” stage dominated by heavy users who struggle to quit, where further reductions in prevalence would be achieved primarily through cessation programs.

DATA AVAILABILITY STATEMENT

The data used in this study are available to researchers without charge at the National Addiction & HIV Archive Program (NAHDAP @ https://www.icpsr.umich.edu/web/pages/NAHDAP/index.html), housed at the Institute for Inter-university Consortium for Political and Social Research (ICPSR) at the University of Michigan.