Abstract
Background. Marijuana is the most commonly used illicit drug in the United States. There has been a rapidly changing marijuana policy environment and increased acceptability related to marijuana in the United States. How the changing environment will potentially influence adolescents age of initiation remains unknown. While much of extant literature has primarily focused on current marijuana use, less is known about age of first use. This study examined trends in adolescents’ age of first marijuana use in the United States. Method. Data were drawn from the 1991 to 2017 Youth Risk Behavior Survey. We used linear regression for age of first use as a continuous variable, and logistic regression for marijuana use before the age of 15 years as a dichotomous variable. Joinpoint regression analysis identified where significant changes in trend occurred. Results. Results showed that the mean age of first marijuana use increased significantly between 1991 and 2017. Males had a younger age of first use than females. Between 1991 and 1997, there was an increasing trend in the prevalence of marijuana use before the age of 15 years in all adolescents and in subgroups for males, females, all races, and 9th and 10th grades. After 1997, a significant downward trend was found in all adolescents. Conclusions. Our results do not indicate an overall decreasing trend of age of first use among the general adolescent population. Results show a downward trend in the use of marijuana before the age of 15 years since 1997.
Marijuana is the most commonly used illicit drug in the United States (Centers for Disease Control and Prevention, 2018c). Frequent marijuana use has been associated with physical and mental health consequences such as impaired breathing, increased heart rate (National Academies of Sciences, Engineering, and Medicine, 2017), anxiety, and depression (Patton et al., 2002). Additionally, marijuana use can disrupt the natural process of neurodevelopment in adolescents (Wilson et al., 2000), with cognitive deficits noted in attention, memory, processing, and verbal learning (Jacobus et al., 2015; Medina et al., 2007). Marijuana use during adolescence is a significant public health concern given that earlier age of first use has been associated with worse outcomes (Jacobus et al., 2015). In 2014, around 8.4% all persons aged ≥12 years used marijuana during the past month, which has increased by 35% compared with a prevalence of 6.2% in 2002 (Azofeifa et al., 2016). A report from Monitoring the Future Surveys found that annual marijuana prevalence among Grade 8 students increased from 2007 to 2010, decreased slightly from 2010 to 2012, and stabilized in 2017 (Johnston et al., 2018). Somewhat similar patterns were found for 10th and 12th grade adolescents, with increased in annual prevalence in 2017 (Johnston et al., 2018).
Perceptions of marijuana use have changed drastically in the United States over the past three decades with support for medical marijuana use increasing widely and support for recreational use gaining ground as well. Among adults, public favor of more liberal marijuana laws has risen each year steadily since the 1990s (Felson, Adamczyk, & Thomas, 2019). Parallel to the public shift in perceptions and norms, permissive marijuana laws have been implemented in many jurisdictions. The legal use of marijuana for medicinal purposes began in 1996 when California became the first state to legalize medical marijuana use (National Conference of State Legislatures, 2019). Currently, 33 states have approved legal use of medical marijuana (National Conference of State Legislatures, 2019), and recreational use or possession of marijuana is legal in some states.
The limited extant literature has produced mixed findings on the association between recreational marijuana legalization and adolescent use (Carliner, Brown, Sarvet, & Hasin, 2017; Cerdá et al., 2017; Rusby, Westling, Crowley, & Light, 2018). For example, while adolescent marijuana use did not increase in Colorado following legalization of recreational marijuana, an increase in use was found among adolescents in Washington after legalization of recreational marijuana (Cerdá et al., 2017). Another prior study found that cannabis use among 8th and 10th graders declined after the legalization of recreational marijuana in Washington (Dilley et al., 2019). Some studies on medical marijuana laws (MMLs) found no significant impact on adolescent marijuana use (Anderson, Hansen, & Rees, 2012; Sarvet et al., 2018). While some have indicated that there has been a decline in marijuana use in high school students since 1999 (Johnson et al., 2015), other studies have indicated that states with MMLs show a higher prevalence of adolescent marijuana use, although these studies recognize that the association may not be causal (Cerdá, Wall, Keyes, Galea, & Hasin, 2012; Wall et al., 2011). An unintended result from the implementation of marijuana liberalization laws is that adolescents are likely to perceive marijuana as less risky (Cerdá et al., 2017; Wen, Hockenberry, & Druss, 2019), which may influence the normalization of marijuana (Hathaway, Comeau, & Erickson, 2011 ). For example, a study found that in the state of Washington, which has a recreational marijuana law, significantly lower risk perception of marijuana use was found among 8th and 10th grade students when compared with states without a corresponding law (Cerdá et al., 2017).
Adolescents have a heightened potential to internalize beliefs and norms about health risk behaviors (Jackson, 1998), and the transition from adolescence to young adulthood is a time of numerous personal and social changes, some of which pose risks for marijuana use (Bonnie et al., 2015). For example, adolescents have an increased susceptibility to peer influence compared with other developmental periods (Steinberg & Monahan, 2007) and are vulnerable to peer influence on substance use (Loke & Mak, 2013). These factors have serious implications for the passage of marijuana laws, both medical and recreational, among this population, especially as early initiation of substance use, mainly before the age of 18 years, has profound long-term impacts on the lives of young people, and dramatically influences their future ability to quit in adulthood (Breslau & Peterson, 1996; Moffitt, 1993). Previous studies have shown that early age at first use, especially before the age of 15 years, was significantly associated with multiple outcomes, including current marijuana use and more frequent marijuana consumption (Azagba & Asbridge, 2019; Ehlers, Slutske, Gilder, & Lau, 2007; El-Menshawi, Castro, Rodriguez de la Vega, Gabriel Ruiz Peláez, & Barengo, 2018; Le Strat, Dubertret, & Le Foll, 2015). Using marijuana at an earlier age has also been associated with lower levels of educational attainment, welfare dependence, unemployment, and the use of other drugs (Fergusson, Boden, & Horwood, 2015). As more states continue to implement recreational marijuana laws, which are more expansive than MMLs in terms of retail access or availability, ongoing monitoring of marijuana use is particularly needed among adolescents. How these laws will potentially influence adolescents’ age of initiation remains unknown. While much of the extant literature has primarily focused on current marijuana use, less is known about age of first use (also known as age of initiation). The present study examined national trends in age of first marijuana use among U.S. adolescents from 1991 to 2017 using data from the Youth Risk Behavior Survey.
Method
Data
We used combined data from the Youth Risk Behavior Survey (YRBS) from the Centers for Disease Control and Prevention for the years 1991 to 2017. Since 1991, the YRBS survey has been conducted every 2 years to monitor six categories of priority health-risk behaviors among youth and young adults in the United States. The three-stage, cluster sample design is used in the national YRBS surveys to present all public and private school students in Grades 9 to 12 in the 50 states and the District of Columbia. The data in our study included 203,663 nationwide survey responses from 1991 to 2017. Sample size ranged from 12,272 to 16,410. Details about the sampling strategy and psychometric properties of the 2015 YRBS have been reported elsewhere (Annor et al., 2018; Centers for Disease Control and Prevention, 2018a).
Measures
Age at first marijuana use was analyzed as a continuous variable. We assessed age of first marijuana use by the question, “How old were you when you tried marijuana for the first time?” with response options including “A = I have never tried marijuana,” “B = 8 years old or younger,” “C = 9 or 10 years old,” “D = 11 or 12 years old,” “E = 13 or 14 years old,” “F = 15 or 16 years old,” and “G = 17 years old or older.” The following continuous variable of age at first use was created: 8 years, 9.5 years, 11.5 years, 13.5 years, 15.5 years, and 17 years. We used the midpoints of C, D, E, and F. The “A = I have never tried marijuana” responses were excluded from the age of first use analysis. In addition, we created a dichotomous variable, marijuana use before the age of 15 years (early age at first use). We defined the variable as a student with a B, C, D, or E response for the first marijuana use question. Sociodemographic variables (e.g., sex, race, and grade level) were obtained from the survey. Race was a four-level variable based on the survey derived classifications: “White,” “Black or African American,” “Hispanic/Latino,” and “All Other Races.”
Statistical Analysis
For each year included in this study, we generated national estimates of the mean age of first marijuana use and the percentage of students who tried marijuana for the first time before the age of 15 years. Additionally, estimates were calculated for the full sample and by sex, grade, and race subgroups separately. The absolute difference in age of first marijuana use and the percentage of students who tried marijuana before the age of 15 years were calculated to compare the difference between subgroups.
To examine whether changes occurred over time in age of first marijuana use, we conducted trend analyses. We used linear regression for age of first use as a continuous variable and logistic regression for marijuana use before the age of 15 years as a dichotomous variable. We first included cubic time, quadratic time, and the linear time variable in the linear regression model and adjusted for sex, race, and grade. If the cubic time variable was not statistically significant, we removed the cubic term from the model and tested the significance of the quadratic term in the new model. Subsequently, if the quadratic term was removed due to statistical insignificance, we then tested for the linear term. Time variables were modeled as continuous variables. The model with significant highest-order time variable was selected as the final model. If the selected model included higher-order time variable (quadratic or cubic time), we used the selected model to estimate the adjusted prevalence and standard error by year to determine the joinpoints of where a significant change in trend occurred (Centers for Disease Control and Prevention, 2018b).
We set the minimum number of joinpoints to 0 and a maximum number of joinpoints to 2 based on the number of data points (National Cancer Institute, n.d.). The analysis starts hypothesis testing from the minimum number of joinpoints to the maximum joinpoints. In each hypothesis test, a grid search method is used to estimate the parameters in the null and alternative models with a Monte Carlo Permutation method used to approximate the significance of each test. Bonferroni correction was used to adjust the overall significance level of multiple hypothesis tests (Kim, Fay, Feuer, & Midthune, 2000). Similar procedures were performed using logistic regression for the dichotomous variable, marijuana use before the age of 15 years. The same procedures were repeated for the sex, race, and grade-based subgroup analyses. All tests were two-sided and a p value of <.05 was considered significant. Joinpoint analyses were performed using the NCI joinpoint software (National Cancer Institute, n.d.). All other analyses were performed in SAS, version 9.4 (SAS Institute Inc., Cary, NC), and took into account the complex sample design of the national YRBS.
Results
Age of First Marijuana Use
Table 1 presents the estimates of age of first marijuana use and the percentage of students who tried marijuana before the age of 15 years among high school students by sex. The mean age of first marijuana use increased slightly over the study period for all adolescents (from 13.7 to 13.9). This increase was found in both males (13.5 to 13.7) and females (14.0 to 14.2). Overall, male adolescents had a younger age of first marijuana use. The difference in the mean age of first marijuana use between males and females was consistent each year (see Figure 1).
National Estimates of Age of First Marijuana Use and Percentage of Students Who Tried Marijuana Before the Age of 15 Years Among High School Students by Sex, in the United States, 1991 to 2017.
Note. Age of first marijuana use was presented in weighted mean and its 95% confidence interval, and percentage of students who tried marijuana before age 15 years was presented in weighted percentage and its 95% confidence interval.
Age of first marijuana use and percentage of students who tried marijuana before the age of 15 years among high school students, by sex—Youth Risk Behavior Surveys, United States, 1991 to 2017.
The findings from the joinpoint regression analysis over the 1991 to 2017 period for age of first marijuana use among high school students in the United States are reported in Table 2. We found a significant increasing trend in the mean age of first marijuana use in all adolescents, in male and female subgroups with no joinpoint identifiable. For White adolescents, the mean age of first marijuana use decreased from 13.8 in 1991 to 13.7 in 2003, then increased to 14.2 in 2017 (see Figure 2). The increasing trend in the mean age of first marijuana use in Hispanic adolescents is consistent with the full sample. No significant change was found in the mean age of first marijuana use from 1991 to 2017 in African American adolescents. The mean age of first marijuana use in other races was stable from 1991 to 2003 but increased from 12.8 to 14.0 between 2003 and 2017. In 9th grade adolescents, the mean age of first marijuana use decreased from 1991 to 2003, then increased from 2003 to 2017 (see Figure 3). A significant increasing trend was found for the mean age of first marijuana use between 1991 and 2017 for 10th and 11th grade adolescents. No significant difference in the mean age of first marijuana use was found in 12th grade adolescents.
Results of Segmented Trend Regression Analyses for All Adolescents and by Subgroups.
Note. All adolescents denotes all included adolescents in our analyses. p < .05 is considered significant and presented in bold. For all adolescent model, we adjusted for sex, race, and grade. For sex subgroup, we adjusted for race and grade. For race subgroup, we adjusted for sex and grade. For grade subgroup, we adjusted for sex and race. Segment 1 included all years from the first available year up to and including the first joinpoint year. Segment 2 included all years from first joinpoint year up to and including the last year or second joinpoint year (if existed). Segment 3 (if existed) included all years from second joinpoint year up to and including the last year.
Age of first marijuana use and percentage of students who tried marijuana before the age of 15 years among high school students, by race—Youth Risk Behavior Surveys, United States, 1991 to 2017.
Age of first marijuana use and percentage of students who tried marijuana before the age of 15 years among high school students, by grade—Youth Risk Behavior Surveys, United States, 1991 to 2017.
Marijuana Use Before the Age of 15 Years
As seen in Table 1, from 1991 to 2017, the prevalence of marijuana use before the age of 15 years increased from 18.5% to 19.2% for all adolescents and increased 3.3% for females (17.0% to 18.6%) but decreased 0.2% for males (20.0% to 19.8%). The prevalence difference in marijuana use between male and female adolescents increased from 3.0 in 1991 to 9.8 in 1999, then decreased to 1.2 in 2017.
Results indicate a significant increasing trend from 1991 to 1997 in the prevalence of marijuana use before the age of 15 years in all adolescents and subgroups for males, females, all races, and 9th to 10th grades. After 1997, a significant downward trend was found in all adolescents. In female adolescents, the prevalence of marijuana use before the age of 15 years increased from 17.0% in 1991 to 20.8% in 2013; however, no significant change was found between 2013 and 2017. The prevalence of male marijuana use before the age of 15 years increased from 20.0% in 1991 to 30.6% in 1997, then decreased to 26.4% in 2005, and further decreased to 19.8% in 2017. In race subgroups, there was no significant change between 1997 and 2001 in the prevalence of marijuana use before the age of 15 years among White students, but the rate decreased significantly from 26.0% in 2001 to 16.2% in 2017. In African American and Hispanic students, the rate of marijuana use before the age of 15 years declined between 1997 and 2013, then decreased between 2013 and 2017. The change was not significant between 2013 and 2017 among African American. In other races, rates went from 32.1% in 1997 to 15.7% in 2017. In grade subgroups, a decreasing trend was found between 1997 and 2017 for 9th and 10th grade adolescents. For 11th and 12th grade participants, the rate decreased between 1995 and 2001 and increased between 2001 and 2017. No other significant change in rates were found in grade subgroups.
Discussion
The past three decades have witnessed a historic shift in public norms around marijuana use. For example, in the United States, the perceived risk of marijuana use among students in 8th, 10th, and 12th grades has decreased overall in recent years (Johnston et al., 2018). The present study used nationally representative data to examine trends in age of first marijuana use among adolescents from 1991 to 2017.
The mean age of first marijuana use for all adolescents increased moderately from 13.7 to 13.9. Male and female adolescents followed a relatively similar pattern regarding trends in age of first use. Between 1991 and 2017, age of first marijuana use for males increased from 13.5 to 13.7, and females with a rise from 14.0 to 14.2. Male adolescents consistently had a younger age of first marijuana use than females. Prior studies have shown that adolescent males have a heightened risk of marijuana use when compared with females (Johnston, O’Malley, Bachman, & Schulenberg, 2008), which may help explain the difference in the mean age of initiation between males and females. While the mean age of first marijuana use increased from 1991 to 2017 for 10th and 11th grade adolescents, it decreased for 9th grade adolescents from 1991 to 2003 followed by an increase through 2017. No significant difference in the mean age of first marijuana use was found in 12th grade adolescents. One possible explanation for the inconsistent trend among for 9th grade adolescents is that norms about substance use may not be fully established at a younger age (Keyes et al., 2016).
Multiple races and grade groups also displayed an increase in age of first marijuana use. Hispanic adolescents’ age of first use rose from 1991 to 2017, while both White adolescents and other races experienced a similar increase from 2003 to 2017. A similar increase was not seen in African American adolescents, as no significant change was found in the mean age of first marijuana use from 1991 to 2017. Although not directly comparable, prior studies have shown an increase in the prevalence of marijuana use among African American adolescents in recent years (Johnson et al., 2015). While it remains unknown the reason for the disparate increase in marijuana use among African Americans, one reason could be the higher accessibility to marijuana among African Americans. One study showed that while the perception of very easy access to marijuana among adolescents declined in all race/ethnicity between 2002 and 2014, it remained higher among African Americans compared with Hispanic and White adolescents (Salas-Wright, Oh, Goings, & Vaughn, 2017). Another potential factor could be the changing perception that marijuana is a no-risk or low-risk behavior, which occurred between 2004 and 2012 (Wen et al., 2019). While some have found that legalization of medical marijuana was associated with an increase in probability of young adults perceiving marijuana use as no or low health risk (Wen et al., 2019), it does not necessarily indicate that use will increase as past research has found overall declines in high school student marijuana use despite increasing legalization (Johnson et al., 2015).
The prevalence of marijuana use before the age of 15 years appeared to have peaked for all subgroups between 1997 and 1999, followed by decreases in subsequent years. Prior literature has suggested that the substantial increases of marijuana use among U.S. youth in the 1990s could be attributable to increasing popularity of blunts (a cigar wrapper filled with marijuana; Golub, Johnson, & Dunlap, 2005; Golub, Johnson, Dunlap, & Sifaneck, 2004). The declining trend found from 1999 is promising given that early age of marijuana onset is associated with several negative outcomes (Azagba & Asbridge, 2019; Fergusson et al., 2015; Jacobus et al., 2015; Medina et al., 2007; Wilson et al., 2000). This has implications for population health as younger age of initiation may hinder future attempts to quit in adulthood (Breslau & Peterson, 1996; Moffitt, 1993), perhaps reducing overall marijuana use in adults and therefore lessening negative outcomes. Males consistently had a higher prevalence of marijuana use before the age of 15 years than females; however, the difference narrowed recently with males and females having similar rates in 2017. This may suggest that the gap between male and female use before the age of 15 years is closing, which is in keeping with prior studies showing that the difference in the prevalence of current use has been narrow in recent years (Johnson et al., 2015; Palmer et al., 2009).
This study had some limitations. First, the YRBS is subject to the limitations of self-report surveys, including the possibility of inaccurate recollection of events or experiences. Second, due to the nature of the study, we were not able to assess the reasons behind significant changes in trend. Another limitation is that YRBS data are collected through school-based surveys and therefore exclude adolescents that are not enrolled in schools, such as home-schooled students and those who have dropped out. Additionally, the survey is nationwide and does not provide state identifiers; thus, we were unable to determine age of first use for each state. Future research should consider examining state differences in age of marijuana initiation.
Conclusion
This study investigated trends in age of first marijuana use using data from a nationally representative survey. Results showed the mean age of first marijuana use increased significantly between 1991 and 2017. Males had a younger age of first use than females. The prevalence of marijuana use before the age of 15 years has decreased since 1999 in the whole population and in subgroups. The decreasing prevalence of marijuana before the age of 15 years is potentially beneficial for public health given that early exposure to marijuana is associated with a number of negative outcomes.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
