Motivations for the nonmedical use of prescription drugs in a longitudinal national sample of young adults

Abstract

Introduction.

People’s motivations for nonmedical use of prescription drugs (NMUPD) are not well studied, particularly in longitudinal representative samples. However, understanding which motivations are most popular and how these change over time for specific groups is important to inform interventions for NMUPD.

Methods.

The current study examined how young adults’ motives for NMUPD changed over young adulthood, using a nationally representative sample of 12,223 young adults in 36 cohorts (1976–2012) as part of the Monitoring the Future study across three biennial waves (waves 1, 2, 3: modal ages 19/20, 21/22, and 23/24 years). We investigated these young adults’ motivations for using stimulants, central nervous system depressants, and opioids when controlling for possible cohort effects. We included sex and college attendance as potential moderators.

Results.

Participants commonly reported recreational and self-treatment motivations over time and across drug classes, reporting four to five popular motivations in each drug class. Generalized estimating equations repeated measure analyses revealed relatively stable NMUPD motivations across young adulthood. Participants reported some reductions in experimentation and boredom as motivations for NMUPD and an increase in certain self-treatment motivations, depending on prescription drug class. Overall, men were more likely to endorse recreational motivations, whereas women were more likely to endorse self-treatment motivations, though this varied somewhat by prescription drug class. Young adults not enrolled in college courses were more likely to endorse using stimulants nonmedically for different reasons than their peers who were enrolled.

Conclusions.

NMUPD prevention and treatment efforts tailored to the young adult population should include methods to reduce both self-treatment and recreational use and need to consider prescription drug class, sex, and college attendance.

1. Introduction

1.1. Nonmedical use of prescription drugs among young adults

The nonmedical use of prescription drugs (NMUPD) among young adults is a significant and growing public health concern because it puts young adults at risk for abuse, overdose, and related negative outcomes (e.g., National Institute on Drug Abuse [NIDA], 2018). In 2018, 1.5 million young adults engaged in NMUPD for the first time in the previous year, meaning there were approximately 4,200 new young adult users daily; and young adults reported disproportionately higher rates of all types of NMUPD compared to other age groups (Substance Abuse and Mental Health Services Administration, [SAMHSA] 2019). According to NIDA, NMUPD occurs when individuals either use medications that were not prescribed to them, use their prescribed medications in higher quantities or in a different manner than prescribed; or take medications for purposes other than prescribed (e.g., to get high; NIDA, 2018). The types of prescription drugs used for nonmedical purposes most often are stimulants, central nervous system (CNS) depressants, and opioids (NIDA, 2018). Young adults report higher rates of past-year misuse of stimulants (6.5%) and opioids (5.5%) compared to CNS depressants (<5%; SAMHSA, 2019). Treatment seeking for NMUPD is rising, with rates for young adults seeking prescription opioid abuse treatment alone increasing by approximately 26% from 2002 to 2010 (SAMHSA, 2011); though there are still significant gaps in access to treatment with approximately only 15.6% of young adults with prescription opioid use disorder receiving specialty substance use treatment in the past year (SAMHSA, 2019). In young adults, NMUPD has been linked with abuse and dependence (Huang et al., 2006; McCabe et al., 2019a; SAMHSA, 2013), mental health problems (Bavarian, Flay, Ketcham, & Smit, 2013; Fischer, Lusted, Roerecke, Taylor, & Rehm, 2012; Janusis & Weyandt, 2010; Lo, Monge, Howell, & Cheng, 2013; McCauley et al., 2011; Zullig & Divin, 2012), poorer performance in collegiate settings (Arria, O’Grady, et al., 2008), more frequent sexual risk behaviors (Benotsch, Koester, Luckman, Martin, & Cejka, 2011), more sleep problems (Clegg-Kraynok, McBean, & Montgomery-Downs, 2011), more emergency room visits (SAMHSA, 2013), arrest and delinquency (Herman-Stahl, Krebs, Kroutil, & Heller, 2007), and more unintentional overdose deaths (Paulozzi, 2012), compared with those not reporting such use. Additionally, young adults who engage in NMUPD are significantly more likely than their peers to use other illicit drugs and to combine prescription drugs with alcohol and other substances, which increases the risk of potentially dangerous drug interactions (Garnier et al., 2009; Hughes et al., 2016; McCabe et al., 2014). Young adults reporting polysubstance use are at the greatest risk for future negative consequences (McCabe, Veliz, Dickinosn, Schepis & Schulenberg, 2019b). Intervention and prevention are key to reducing this public health problem and its grave costs to society.

One way to prevent substance use is to investigate why specific groups of people use and to target interventions at predictors of maintaining use over time. At the individual level, one of these factors is young adults’ motivations for engaging in NMUPD. Although it is only a small piece of the etiology for drug use and abuse (see Kendler, 2012), self-reported motivations are amenable to change (e.g., Miller, Benefield, & Tonigan, 1993), and other activities or habits may replace substance use if they address the same underlying motives. If NMUPD is driven by sensation seeking and rewards (i.e., positive reinforcement), treatments that utilize operant conditioning, such as contingency management (see Davis et al., 2016), may be effective. If individuals are motivated to escape aversive internal experiences (i.e., negative reinforcement), coping-focused treatment for substance use disorders such as cognitive behavioral therapy (Haller et al., 2016; McGovern, Lambert-Harris, Alterman, Xie, & Meier, 2011) may address these motivations. A better understanding of these motivations may lead to more effective, etiologically based strategies for prevention and treatment.

Previous research has shown that motivations for substance use in young adults are important predictors of use patterns and problems; however, much of this research has focused on alcohol use and problems (Kenney, Paves, Grimaldi, & Labrie, 2014; Kuntsche, Knibbe, & Gmel, 2005; Kuntsche, Wiers, Janssen, & Gmel, 2010; O’Neil, Lafreniere, & Jackson, 2016). Items assessing motives for NMUPD were added to the National Survey of Drug Use in Households (NSDUH) in 2015 (Hughes et al., 2016), though published research with the NSDUH data has focused on motives for prescription opioids (Han et al., 2017) and stimulants (Compton, Han, Blanco, Johnson, & Jones, 2018) across the general adult population, perhaps due to the higher rates of these two drug classes relative to others among this population (i.e., CNS depressants). Indeed, other research on motives for NMUPD is heavily focused on stimulants (Advokat, Guidry, & Martino, 2008; Barrett, Darredeau, Bordy, & Pihl, 2005) and prescription opioids as well (McCabe, Boyd, & Teter, 2009; McCabe, Cranford, Boyd, & Teter, 2007; McCabe et al., 2019a). Limited work has explored motives for CNS depressants, despite differences in motivations for CNS depressants compared to stimulants and prescription opioids (Messina et al., 2016). No work has examined the impact of motives on patterns of CNS depressant misuse over time.

Various theories explain why individuals are motivated to abuse substances. One theory is the self-medication hypothesis that Khantzian (1985, 1997) and Duncan (1974) initially articulated. This theory posits that individuals engage in drug use to treat underlying disorders or problems that have not been properly treated by other means. For some, initial use may have started appropriately, such as for pain or sleep, but then developed into addiction based on the properties of the medications themselves (Alam et al., 2012). Developmental theory suggests that individuals ages 18 to 25 may be motivated to engage in drug use for recreational reasons as part of self-exploration or instability during this developmental period.

Specifically, Arnett (2005) suggests that young adults use substances due to curiosity, sensation seeking, and a desire to have a wide range of experiences before settling into adult life. Also, since constructing a stable identity may be confusing and difficult, young adults may use substances to relieve negative feelings related to identity confusion. Young adults may be more likely to use substances to manage stress associated with disruptions in life (e.g., new residences, romantic partners, educational and vocational settings). As young adults’ independent decision-making increases and they transition out of their parents’ house, or, in the case of many college students, transition in and out of their parents’ house, they experience less parental monitoring and potentially engage in more deviant behaviors, including substance use. Young adults may also experience a feeling of “in-betweeness”—that they are no longer adolescents but not yet adults—which may also lead to substance use. Young adults are discovering that they have the ability to make independent choices about substance use that may be in opposition to their caregivers; however, they do not yet feel the need to be as responsible in their drug use as they believe adults should be. Finally, Arnett (2005) proposes that young adults often experience high levels of optimism and value in making dramatic life changes. As a result of this optimism, young adults may not fully consider the negative consequences that result from substance use. All of these factors may explain different motivations for engaging in NMUPD.

Research has theorized that as young adults age and their lives become more stable, both in terms of identity and life roles, they engage less in recreational drug use (Arnett, 2005); therefore, studying NMUPD during this developmental period is important. Further, evidence suggests that because of brain development, adolescent decision-making is different than that of young adults, which is different than that of older adults (for a reviews see Spear, 2013; Steinberg, 2008). Gray and white matter continue to increase into the early 20s. Additionally, brain areas responsible for emotions (e.g., amygdala) become fully developed by mid-adolescence while the frontal lobes are still developing at this point. Research has now found that the frontal lobes do not reach maturity until approximately age 25. As the frontal lobes are responsible for long-term thinking, emotion-driven decision-making compared to methodical decision-making is more common during this stage. Consequently, the motivations to engage in NMUPD, and the interventions targeting NMUPD, will likely vary by age. Therefore, this study focuses on the developmental period of young adulthood, the population considered most at-risk for NMUPD. It is important to note that some motivations for NMUPD and other substances, such as alcohol, may overlap (e.g., coping, enhancement; Cooper, 1994), while others motivations may play less of a role given that NMUPD has a relatively lower prevalence than alcohol (e.g., social, conformity).

Indeed, previous research has found that young adults appear motivated to engage in NMUPD for self-treatment (e.g., using the medications as they are clinically intended, but outside of a doctor’s oversight or in ways not prescribed), as well as for recreational purposes (e.g., to get high; for a review see Drazdowski, 2016). However, the research on NMUPD motivations in young adults to date is limited in several important ways. The preponderance of research is about the nonmedical use of prescription stimulants and opioids and is cross-sectional. Few studies have focused on other commonly abused prescription drug classes, such as CNS depressants. Further, only one study has examined motivations over time (longitudinal) in this population (Garnier-Dykstra, Caldeira, Vincent, O’Grady, & Arria, 2012). The authors of that study found that across all time points, using prescription stimulants to “improve focus/study/work” remained the most reported motive for nonmedical use and that motivation to use in this way increased over time. Additionally, curiosity/experimentation motivations significantly decreased over time. There were no significant changes in the following motives: to get high, to stay awake, to party, and “other.” However, Garnier-Dykstra and colleagues (2012) focused only on college students who used prescription stimulants nonmedically. Similarly, Arria et al. (2018) examined changes in NMUPD during and after college. These studies’ findings may not generalize to nonstudents, however. McCabe, Schulenberg, et al. (2014) and McCabe, Kloska, et al. (2016) did investigate trends in the nonmedical use of prescription opioids during young adulthood and from adolescence to young adulthood in national samples, but they did not examine motivations for NMUPD. Without longitudinal work, it is unclear how motivations for NMUPD may change over time and whether the same motivations continue to influence individuals as they progress developmentally.

Additionally, most of the research has surveyed students in higher education, with few researchers considering how motivations may be similar or different among young adults who are not attending college. Theoretically, the motivations for NMUPD among this latter group may be similar as they are experiencing parallel developmental demands (Arnett, 2005), but this has yet to be tested. A review of motivations to use different classes of NMUPD in college students found that the most common motives for stimulants were academic, while analgesics were used for pain and for partying, tranquilizers were used for self-medication and anxiety reduction, and sedatives were predominantly used for sleep (Bennett & Holloway, 2017). To date, cross-sectional work that includes nonstudent populations (Upadhyay et al, 2010) has found that motivations for nonmedical stimulant use among this population are similar to those found in exclusively college samples (e.g., Advokat et al., 2008; Dussault & Weyandt, 2013; Gallucci, Usdan, Martin, & Bolland, 2014; McCabe et al., 2009); yet college students’ motives for stimulants may have driven this finding. Academic motives may not apply to nonstudents, yet productivity motives (e.g., to stay awake, to get work done) may be similar across student and nonstudents populations. Given that research has found that educational attainment, specifically, completing a 4-year university degree, predicts less risky NMUPD patterns (McCabe et al., 2019b), research on NMUPD should consider populations in higher education.

Further, in the general adult population, overall men are more likely than women to report NMUPD, with variations depending on prescription drug class (Hughes et al., 2016). In young adults, there are similar trends, with males reporting more nonmedical use of prescription opioids initially, and females reporting more nonmedical use of prescription stimulants and CNS depressants (at age 18). Males report faster rates of decline in NMUPD across all prescription drug classes throughout young adulthood (age 26) when compared to females, though population-based studies are limited to date (McCabe et al., 2017). Studies that have investigated sex differences in NMUPD motivations for young adults have focused exclusively on college students. In these studies there has been a general trend that males are more likely to endorse recreational and mixed motivations (both recreational and self-treatment motivations) compared to females, and females are more likely to report self-treatment motivations, including academic motives, compared to males (e.g., Lord, Brevard, & Budman, 2011; Lord et al., 2009; McCabe et al., 2007, 2009; Pino, Tajalli, Smith, & DeSoto, 2017; Smith, DeSantis, & Martel, 2018; Vest, Murphy, & Tragesser, 2018). There is more evidence of this trend for CNS depressants and opioids as compared to stimulants (McCabe et al., 2009).

1.2. Current study

Therefore, the current study contributes to the literature on the motivations for NMUPD in young adults by addressing some of the limitations of previous work. Our sample consisted of a nationally representative cohort of high school seniors who were followed longitudinally across three biennial waves covering ages 19–24. With this sample, we can answer questions about how motivations for NMUPD change over time at the national population level when considering the impact of birth cohort (i.e., the year individuals were born). We also investigated whether college attendance and sex were moderators. We investigated specific motivations for the nonmedical use of prescription stimulants, CNS depressants, and opioids. Thus, this study expands our knowledge on under-studied prescription drug classes.

Based on developmental theory (Arnett, 2005) and longitudinal research by Garnier-Dykstra and colleagues (2012), we hypothesized that: (1) the recreational motivations of feeling good or getting high, experimentation, and boredom would decrease over time across all prescription drugs classes as young adults move toward stability in life and identity-formation; (2) self-treatment motives (e.g., to relax or relieve tension, to get sleep, to relieve physical pain), addiction motives, and motives related to drug interactions (e.g., to increase the effects of other drugs; to decrease the effects of other drugs) would increase over time across all prescription drug classes as exposure to prescription drugs and the risk for psychiatric problems and polysubstance use increases. Our investigation of the remaining motives was exploratory given the lack of literature on the following: to seek deeper insights and understanding, anger or frustration, to get through the day, as a substitute for heroin, and to control coughing. Similarly, given the dearth of literature on sex and college status differences in motivations for NMUPD, our moderation analyses were exploratory. All analyses controlled for possible cohort effects, given increases in prescription drug use over time and changes in drug trends (e.g., increases in the late 1990s; Schulenberg et al., 2019) that may be related to changes in motivation for use.

2. Materials and methods

2.1. Participants and procedures

The current study used limited de-identified national panel data from the Monitoring the Future (MTF) study, which collects longitudinal nationally representative data from the United States. Surveys have been carried out each year since 1975 by the University of Michigan Survey Research Center; the current research used data from years 1976–2012. Though individual items have been added to address the evolution of attitudes toward drugs and drugs monitored, the basic study design as well as the classes of drugs have remained largely consistent (Bachman et al., 2015). The MTF study uses a three-stage sampling procedure to gather a representative sample of students. In stage 1, geographic areas or primary sampling units are selected; in stage 2, schools within primary sampling units are selected; and in stage 3, students within schools are selected. Each year about 16,000 students in approximately 133 public and private high schools nationwide participate by completing surveys administered in classrooms. A randomly selected sample of approximately 2,400 students from each senior class is followed up biennially after high school on a continuing basis. These respondents receive the mailed questionnaire at their home, which they complete and return to MTF. The biennial follow-up surveys begin 1 year after high school for one random half of each cohort and 2 years after high school for the other half (for more information on the procedures, see Bachman, Johnston, & O’Malley, 2014; Bachman, Johnston, O’Malley, & Schulenberg, 2011). For the purpose of the current study and due to concerns about sample size, our sample consisted of all available data of NMUPD users across all waves from the 36 cohorts from years 1976–2012, and the two halves were combined (combining modal ages 19/20, 21/22, and 23/24 years). Previous research has employed similar methods and reported a lack of significant differences across the two halves on substance use measures (McCabe et al., 2014). Response rates for high school seniors ranged from 77% to 85% between 1976 and 2012 (Bachman et al., 2014). The follow-up panel data for surveys through wave 3 have been estimated to be approximately just below 50% (Bachman et al., 2014; McCabe et al., 2014). After accounting for sampling bias, our sample was 48% male and 73% white (weighted). More information about the total sample and related data can be found on the MTF website: http://www.monitoringthefuture.org.

2.2. Measures

Using six randomly distributed questionnaire forms, the MTF study assesses demographic and psychosocial characteristics and standard measures of substance use.

2.2.1. Demographics

Demographics were assessed at wave 1 and consisted of participant self-reports of age, sex (i.e., male/female), race/ethnicity (i.e., white/non-white), and if they were enrolled in an academic course (i.e., “During March of this year, were you taking courses at any school or college?”).

2.2.2. NMUPD

NMUPD was assessed at all three waves by asking respondents on how many occasions (if any) they had used prescription medications on their own, without a doctor’s order during the past 12 months. There were separate questions for each prescription medication class: (1) prescription stimulants (e.g., Dexedrine, Ritalin, Adderall, Concerta, methamphetamine); (2) prescription CNS depressants (e.g., Librium, Valium, Xanax, Soma, Serax, Ativan, Klonopin); and (3) prescription opioids (e.g., methadone, codeine, OxyContin, Percodan, opium, Demerol, Percocet, Ultram, morphine, and Vicodin). The response scale was 1 for no occasions, 2 for one to two occasions, 3 for three to five occasions, 4 for six to nine occasions, 5 for 10–19 occasions, 6 for 20–39 occasions, and 7 for 40 or more occasions.

2.2.3. Motivations for NMUPD

We asked participants who reported past-year NMUPD to indicate the most important reasons for NMUPD from a check-all-that-apply list of binary items to assess motivations for NMUPD. All prescription drug classes listed the following motives: (a) to experiment—to see what it’s like, (b) to relax or relieve tension, (c) to feel good or get high, (d) to seek deeper insights and understanding, (e) to have a good time with my friends, (f) to fit in with a group I like, (g) to get away from my problems or troubles, (h) because of boredom, nothing else to do, (i) because of anger or frustration, (j) to get through the day, (k) to increase the effects of some other drug(s), (l) to decrease (offset) the effects of some other drug(s), and (m) because I am “hooked”—I feel I have to have them. Specific motives for stimulants included: (a) to stay awake, (b) to get more energy, (c) to help me study, and (d) to help me lose weight. Both the CNS depressants and opioids listings included the motives to get sleep and to relieve physical pain. Opioids had the additional motives of as a substitute for heroin and to control coughing.

2.3. Data analysis

First, we calculated descriptive statistics to examine the frequencies and distribution properties of each variable, as well to gain a better understanding of the sample demographics. Then we used a test to determine whether the missing data were missing completely at random (MCAR), an assumption of generalized estimating equations (GEE). We ran all analyses using the panel weight to adjust for attrition due to the participant’s “drug strata” status from his/her base year responses. All the respondents were included in the analyses when possible. Sample sizes varied across analyses due to responses with missing items.

We used series of GEE repeated measure analyses to test variations in motivations of NMUPD using SPSS Version 26. To answer the question of whether motivations of NMUPD change over the developmental period of young adulthood, we developed GEE repeated measures equations for nonmedical users, in which each motive was, in turn, entered into a separate equation as the dependent variable, with wave (i.e., time) as the repeated factor in each equation. We developed separate equations for each of the following, nonmedical use of: (1) stimulants, (2) CNS depressants, and (3) opioids. Next, we conducted analyses to determine whether differences existed in the motivations for NMUPD based on young adults’ sex and college attendance. Following the bivariate analyses, we re-estimated each equation with wave as the repeated factor and the main effect of sex or college attendance and the first-order interaction of sex/college attendance x wave included in the model. For all GEE analyses, a Wald chi-square test determined whether wave, sex, college attendance, and the sex/college attendance x wave interaction were significantly associated with each dependent variable (i.e., each motivation). We translated odds ratios (ORs) into percent change with the following formula (for ORs of greater than or less than 1): OR/ 1+OR, or (for ORs of 1–1.99): OR x 100. We used cohort as a control variable in all GEE analyses. For models with significant chi-square tests, we evaluated pairwise comparisons of estimated marginal means produced from these models to determine among which waves the participants differed with respect to the dependent variable. In the pairwise comparisons, we used a Bonferroni correction to control for the likelihood of making a Type I error as a result of multiple comparisons. We chose this correction for bias from multiple comparisons over a data reduction method, as our study is the first to examine potential changes in NMUPD motivations over time

3. Results

3.1. Descriptive statistics

Table 1 reports the sample demographics and past year NMUPD for the total sample of nonmedical prescription drug users and by prescription drug class and wave. Of note, most individuals reported infrequent use (1–2 times in the past year). For stimulants, the top NMUPD motivations consistently included (in terms of high frequencies): to help me study, to stay awake, to get more energy, to feel good or get high, to experiment, and to help me lose weight. For CNS depressants, the top NMUPD motivations consistently included: to relax or relieve tension, to get sleep, to feel good or get high, to experiment, and to relieve physical pain. The frequencies of NMUPD motivations for prescription opioids were slightly more varied, but consistently included: to feel good to get high, to experiment, to relax or relieve tension, and to relieve physical pain (see Table S1 for the frequencies and percentage of users that endorsed each motivation across all waves). Across all prescription drug classes, participants commonly endorsed more than one motive, but not more than four or five motives across all waves (see Table S2 for the frequencies, means, and standard deviations of the number of motivations that users endorsed, by prescription drug class).

Table 1.

Sample demographics of total sample and by prescription drug class for past year NMUPD users across waves

Demographic	Total sample			Stimulants			CNS depressants			Opioids
N of NMUPD users	12.223			Wave 1	Wave 2	Wave 3	Wave 1	Wave 2	Wave 3	Wave 1	Wave 2	Wave 3
Sex (n, % male)	5789 (47.5)			288 (43.6)	273 (44.7)	185 (47.5)	123 (36.1)	126 (40.9)	119 (40.8)	153 (41.9)	148 (43.1)	113 (42.1)
Race/ethnicity (n, % White)	8736 (72.5)			582 (88.6)	529 (87.6)	335 (85.8)	295 (88.3)	273 (89.1)	254 (86.6)	322 (88.6)	301 (89.6)	231 (86.1)
	Wave 1	Wave 2	Wave 3
Age (M(SD))a	19.52 (0.46)	21.49 (0.44)	23.47 (0.43)	19.43 (0.26)	21.47 (0.26)	23.38 (0.26)	19.51 (0.24)	21.42 (0.23)	23.52 (0.24)	19.49 (0.24)	21.49 (0.29)	23.47 (0.28)
Taking college classes (n, %)	5753 (68.1)	4447 (56.8)	2193 (30.7)	398 (60.7)	295 (48.6)	113 (29.0)	203 (60.1)	148(48.4)	84 (28.8)	251 (68.1)	192 (56.5)	88 (32.9)
Number of occasions used in post year (n, %)
0				7719 (92.1)	7152 (92.1)	6721 (94.5)	8023 (95.9)	7450 (96.0)	6820 (95.9)	7949 (95.6)	7396 (95.6)	6822 (96.2)
1–2				217 (2.6)	214 (2.8)	133(1.9)	183 (2.2)	160 (2.1)	147 (2.1)	189 (2.3)	174 (2.2)	132 (1.9)
3–5				131 (1.6)	133 (1.7)	78 (1.1)	78 (0.9)	57 (0.7)	57 (0.8)	79 (0.9)	77 (1.0)	63 (0.9)
6–9				98 (1.2)	77 (1.0)	50 (0.7)	36 (0.4)	46 (0.6)	39 (0.6)	46 (0.6)	35 (0.4)	29 (0.4)
10–19				95 (1.1)	78 (1.0)	51 (0.7)	27 (0.3)	28 (0.4)	24 (0.3)	27 (0.3)	24 (0.3)	18 (0.2)
20–39				52 (0.6)	51 (0.7)	38 (0.5)	9 (0.1)	9 (0.1)	9 (0.1)	14 (0.2)	14 (0.2)	14 (0.2)
40 +				73 (0.9)	61 (0.8)	44 (0.6)	10 (0.1)	11 (0.1)	18 (0.2)	14 (0.2)	19 (0.2)	14 (0.2)

Note. All descriptive statistics are weighted. All percentages are reported as valid percentages, not including participants who had missing values.

^aInformation provided by researchers at University of Michigan with access to full data set.

3.2. Missing data analyses

We conducted missing data analyses to determine if our assumption that data were missing completely at random was met. According to Little’s chi-square statistic (Little, 1988), data were missing completely at random (MCAR), χ² = 12882.59, df = 16252, p = 1.00, meeting the assumption of GEE modeling. Nevertheless, we implemented precautions given the complex sampling procedures. For example, we specified the correlation structure in the GEE analyses, even though GEE models are more flexible for missing data compared to other models (Zeger, Liang, & Albert, 1988). We conducted GEE analyses using an unstructured correlation matrix based on theory, comparison to the actual correlation matrixes, and Quasi-likelihood under Independence Model Criterion (QIC) comparisons, with very small differences observed in the QIC values.

3.3. Changes in NMUPD motivations across young adulthood

Table 2 reports motivations with significant differences across waves for all NMPUD users, and results from subsequent pairwise comparisons using Bonferroni corrections. Overall, relatively few motives evidenced a main effect of time. For stimulants, in general, participants reported less experimentation and more motivation to increase energy over time when controlling for nonsignificant cohort effects over time. For CNS depressants, participants reported more motivation to decrease the effects of other drugs over time when controlling for nonsignificant cohort effects and to relieve physical pain over time when controlling for significant cohort effects. In addition, for CNS depressants, though there was a significant effect of time on the motivation to increase the effect(s) of other drugs, there were not significant pairwise differences by time when controlling for significant cohort effects. Significant cohort effects on motivations to use CNS depressants nonmedically to relieve physical pain and to increase effects of other drugs suggest that these two motivations to use CNS depressants increased by 19% and 9%, respectively, on average for each cohort. For opioids, participants reported less experimentation and more motivation to relieve physical pain as they matured when controlling for nonsignificant cohort effects. Additionally, GEE models for addiction motivations for all three drug classes and the motivation to decrease the effect of other drugs for opioids did not converge, potentially as a result of the small number of participants endorsing these motives across waves (n = 4–25 participants for any given time point; i.e., Clegg-Kaynok et al., 2011, and Cooper et al., 2000, respectively).

Table 2.

Significant changes in NMUPD motivations across young adulthood for all users, controlling for cohort effects

NMUPD Motivation	Cohort Wald χ2	Cohort OR	Time Wald χ2	Wave 2 OR1	Wave 3 OR1	Pairwise comparisons
Stimulants (N = 1024)
To experiment, see what it’s like	2.13, p = .14	0.97, p = .26	16.85***	0.81, p = .14	0.50***	Wave 1 > Wave 3***
						Wave 2 > Wave 3**
To get more energy	1.65, p = .20	1.24, p = .11	12.12**	1.25, p = .11	1.73**	Wave 1 < Wave 3**
CNS depressants (N = 535)
To decrease/offset effects of other drug(s)	3.28, p = .07	l.11, p = .07	8.40*	5.45, p = .12	13.99**	Wave 1 < Wave 3**
To relieve physical pain	10.11***	1.09***	5.87*	1.66*	1.39, p = .16	Wave 1 < Wave 2*
To increase effects of other drug(s)	18.63***	1.19***	21.21***	1.51, p = .12	0.96, p = .92	Wave 2 > Wave 3 ns
						Wave 2 > Wave 1 ns
Opioids (N = 559)
To experiment	1.64, p = .20	1.04, p = .20	10.72**	0.57**	0.57**	Wave 1 > Wave 2**
						Wave 1 > Wave 3**
To relieve physical pain	1.06. p = .30	0.80, p = .31	8.97**	1.49*	1.85**	Wave 1 < Wave 3*
						Wave 1 < Wave 2 ns

Note. NMUPD = Non-medical use of prescription drugs. p-values are Bonferroni corrected for multiple comparisons within-model. OR = odds ratio.

¹Wave 1 is reference group; OR for Wave 2 and 3 are compared to Wave 1.

^*p < .05.

^**p < .01.

^***p < .001.

3.4. Sex as a moderator in changes in NMUPD motivations

Table 3 reports descriptive and chi-square statistics for NMUPD motivations with main effects for sex and/or significant moderation (wave x sex). Motivations to use stimulants to get through the day, stay awake, and lose weight decreased significantly over subsequent cohorts. After controlling for cohort effects, significantly more males than females endorsed the motivations: experimentation, have a good time with friends, increase the effects of other drugs, decrease the effects of other drugs, and stay awake. More females than males endorsed using stimulants nonmedically to lose weight. After controlling for cohort effects, more males than females endorsed the following motivations for CNS depressant use: experimentation, feel good/get high, and have a good time with friends. There were significant cohort effects for experimentation and to get high. More females than males endorsed being motivated to use CNS depressants nonmedically to get sleep and to decrease effects of other drugs. After controlling for cohort effects, more males than females endorsed the motivations for opioid use: experimentation, feel good/get high, and have a good time with friends. There were significant cohort effects for motivations to use opioids to feel good/get high and to have a good time with friends that increased over subsequent cohorts. More females than males endorsed using opioids nonmedically to get sleep, to relieve physical pain, and to control coughing.

Table 3.

Frequencies and percentages of users by sex and wave for significant GEE models, Wald chi-square, pairwise comparisons, and odds ratios for main effect of sex and wave*sex controlling for cohort effects

NMUPD Motivation	Wave 1 Male: n (%) Female: n (%)	Wave 2 Male: n (%) Female: n (%)	Wave 3 Male: n (%) Female: n (%)	Cohort Wald χ2	Cohort OR	Sex Wald χ2	Sex OR1	Wave * Sex Wald χ2	Wave 1 * 1 Sex OR	Wave 2 * Sex OR
Stimulants (N = 1022)
To experiment, see what it’s like	127 (45.5)	109 (41.0)	57(32.6)	1.44	0.98	8.35 **	0.55*	ns	1.74,	1.01,
	143 (30.5)	96 (302)	42 (21.2)	P = .23	p = .23	Male > Female			P = .11	P = .97
To feel good, get high	110 (39.3)	103 (38.7)	73 (41.7)	2.86	0.97	ns	0.43**	10.22**	2.99**	2.00*
	142 (39.3)	114 (35.8)	57 (28.8)	P = .09	p = .09			Wave 1 female > Wave 3 female*;
								Wave 3 male > Wave 3 female*
To have good time with friends	88 (31.4)	70 (206)	59 (33.7)	0.01	1.00	6.20*	0.56*	ns	1.25,	1.42,
	86 (23.8)	82 (25.8)	41 (20.7)	p = .94	p = .94	Male > Female			P = .53	P = .26
To gel through the day	60 (21.4)	76 (28.5)	47 (26.9)	6.50**	94**	ns	ns	10.73**	2.82*	1.01,
	98 (27.1)	74 (23.3)	43 (21.8)					Pairwise difference ns		P = .97
								Wave 1 male < Wave 2 male, p = .099
To increase effects of other drug(s)	36 (12.9)	26 (9.8)	21 (12.0)	0.91	1.03	12.74***	0.27**	ns	1.88,	1.45,
	25 (6.9)	15 (4.7)	12 (6.1)	p = .34	p = .34	Male > Female			p = .31	p = .58
NMUPD Motivation	Wave 1 Male: n (%) Female: n (%)	Wave 2 Male: n (%) Female: n (%)	Wave 3 Male: n (%) Female: n (%)	Cohort Wald χ2	Cohort OR	Sex Wald χ2	Sex OR1	Wave * Sex	Wave 1 * Sex OR	Wave 2 * Sex OR
To decrease (offset) effects of other drug(s)	17 (6.1)	17 (6.4)	14 (8.0)	0.24	0.98	8.85**	0.22*	ns	1.62,	2.01,
	9 (2.5)	11 (3.5)	6 (3.0)	p = .63	p = .63	Male > Female			p = .63	p = .46
To stay awake	193 (68.9)	190 (71.4)	131 (74.9)	17.88***	0.93***	22.75***	0.35***	ns	1.99*	1.57,
	221 (61.0)	188 (59.1)	108 (54.8)			Male > Female				p = .17
To help lose weight	24 (8.6)	32 (12.0)	19 (10.9)	19.63***	0.90***	124.91***	12.25***	ns	1.30,	0.58,
	180 (49.7)	149 (46.9)	110 (55.8)			Female > Male			p = .58	p = .21
CNS depressants (N = 535)
To experiment	50 (44.6)	57 (47.9)	44 (40.0)	9.97**	1.09**	27.29***	0.34**	ns	1.25,	0.76,
	63 (31.5)	36 (21.7)	32 (20.0)			Male > Female			p = .69	p = .57
To feel good, get high	52 (46.8)	66 (55.5)	55 (50.0)	8.75**	1.09**	14.88***	0.42*	ns	1.80,	0.64,
	66 (33.0)	50 (30.3)	43 (26.7)			Male > Female			p = .26	p = .37
To have good time with friends	29 (26.1)	43 (36.4)	28 (25.2)	12.41***	1.30***	8.73**	0.91,	χ2 (1, N = 446) = 5.94*	0.63,	0.06*
	30 (15.0)	26 (15.7)	19 (11.8)			Male > Female	p = .87	Pairwise differences ns	p = .57
								Wave 1 female < Wave 2 male, p = .29
To get sleep	45 (40.5)	44 (37.0)	54 (49.1)	0.47	1.02	5.92*	1.24,	ns	1.57,	1.41,
	100 (50.0)	87 (52.4)	90 (56.3)	p = .49	p = .49	Female > Male	p = .47		p = .30	p = .41
NMUPD Motivation	Wave 1 Male: n (%) Female: n (%)	Wave 2 Male: n (%) Female: n (%)	Wave 3 Male: n (%) Female: n (%)	Cohort Wald χ2	Cohort OR	Sex Wald χ2	Sex OR1	Wave * Sex	Wave 1 * Sex OR	Wave 2 * Sex OR
Opioids (N = 558)
To experiment, see what iťs like	91 (65.0)	69 (51.1)	45 (44.1)	3.05	1.05	26.16***	0.48*	ns	0.67,	0.64,
	75 (41.1)	51 (30.2)	33 (23.9)	p = .08	p = .08	Male > Female			p = .41	p = .26
To feel good, get high	89 (63.6)	81 (59.6)	64 (62.1)	4.98*	1.07*	13.63***	0.49*	ns	1.02,	0.95,
	89 (48.6)	64 (38.1)	54 (39.1)			Male > Female			p = .96	p = .91
To have good time with friends	55 (39.3)	46 (34.1)	34 (33.0)	6.79**	1.09**	13.46***	0.27***	ns	2.20,	1.98,
	41 (22.4)	35 (20.8)	23 (16.7)			Male > Female			p = .12	p = .15
To get sleep	28 (20.0)	31 (22.8)	25 (24.5)	0.22	0.99	7.82**	1.73,	ns	1.28,	0.92,
	64 (35.0)	57 (33.7)	44 (31.9)	p = .64	p = .63	Female > Male	p = .13		p = .63	p = .87
To relieve physical pain	57 (41.0)	68 (49.3)	57 (55.3)	0.49	1.02	8.68**	2.42**	ns	0.50,	0.72,
	94 (51.4)	110 (65.1)	100 (72.5)	p = .48	p = .48	Female > Male			p = .12	p = .43
To control coughing	9 (6.4)	10 (7.3)	7 (6.8)	0.49	0.90**	8.68*	3.44*	ns	0.49,	0.58,
	22 (12.0)	22 (13.0)	17 (12.3)	p = .48		Female > Male			p = .22	p = .50

Note. NMUPD = Non-medical use of prescription drugs. p-values are Bonferroni corrected for multiple comparisons within-model. OR = odds ratio.

¹Female is reference group for sex odds ratios.

^*p < .05.

^**p < .01.

^***p < .001.

Additionally, GEE models for the following motivations were unable to converge or received an error for concerns about validity: for CNS depressants, seeking deeper understanding, fit in with a group, and boredom; and for opioids, to relax, fit in with a group, decrease the effect of other drugs, and as a substitute for heroin. Further, for the motivation to get through the day for stimulant medications, when controlling for significant cohort effects that indicate decreases in this motivation over subsequent cohorts, we found a significant Wald chi-square for a wave x sex moderation (χ² (2, N = 790) = 10.73, p = .005); however, we did not have any statistically significant findings in subsequent pairwise comparisons.

3.5. College attendance as a moderator in changes in NMUPD motivations

Table 4 reports descriptive and chi-square statistics for NMUPD motivations that had a main effect for college attendance and/or significant moderation (wave x college attendance). We found the highest number of NMUPD motives for stimulant medications. Specifically, after controlling for cohort effects, young adults not enrolled in college courses were more likely to endorse the following motivations for using stimulants nonmedically: to relax or relieve tension, to feel good/get high, to have a good time with friends, and to help lose weight; while college students were more likely to report using stimulants nonmedically to study. In college attendance analyses, we found significant cohort effects in using stimulants to relax or relive tension, which increased over subsequent cohorts, and to lose weight, which decreased over subsequent cohorts. We found no college attendance by time interactions for stimulant motivations. Young adults not enrolled in college were more likely to endorse using opioids nonmedically to manage anger and frustration, when controlling for nonsignificant cohort effects.

Table 4.

Frequencies and percentages of users by college attendance and wave for significant GEE models, Wald chi-square, pairwise comparisons, and odds ratios for main effect of college attendance and wave*college attendance controlling for cohort effects

NMUPD motivation	Wave 1 Not enrolled: n (%) Enrolled: n (%)	Wave 2 Not enrolled: n (%) Enrolled: n (%)	Wave 3 Not enrolled n (%) Enrolled: n (%)	Cohort Wald χ2	Cohort OR	College Attendance Wald χ2	College Attendance OR1	Wave * College Attendance Wald χ2	Wave 1 * College Attendance OR	Wave 2 * College Attendance OR
Stimulants (N = 1013)
To relax, relieve tension	34 (13.9)	30 (10.0)	21 (7.9)	9.88**	1.10**	10.63**	6.07	ns	0.36,	0.45,
	38 (9.7)	16 (5.7)	3 (2.8)			Not enrolled > Enrolled	p = .06		p.= .32	p = .45
To feel good, get high	110 (44.7)	141 (47.0)	98 (36.8)	0.89	0.98	10.70***	1.31	ns	1.11,	1.16,
	140 (35.9)	76 (27.0)	33 (30.8)	P = .35	P = .35	Not enrolled > Enrolled	p = .36		P = .72	p = .69
To have good time with friends	74 (30.2)	97 (32.3)	73 (27.4)	0.09	1.01	4.41*	1.42	ns	1.42,	0.91,
	99 (25.3)	63 (22.4)	26 (24.3)	p = .77	p = .77	Not enrolled > Enrolled	p = .26		p = .26	P = .81
To help lose weight	87 (35.5)	117 (38.9)	98 (36.8)	6.87**	0.95**	4.84**	1.26	ns	1.26,	0.93,
	116 (29.7)	64 (22.8)	30 (28.0)			Not enrolled > Enrolled	p = .34		p = .34	p = .83
To study	1 (16.7)	6 (40)	19 (51.4)	0.23	0.82	347.98 ***	0.11*	ns	0.0,	0.75,
	33 (84.6)	50 (84.7)	16 (88.9)	p = .63	p = .63	Enrolled > Not enrolled			p = 1.00	p = .79
Opioids (N = 558)
To manage anger and frustration	12 (11.1)	20 (15.3)	14 (8.4)	2.26	1.07	4.98*	1.28	ns	1.82,	2.45,
	18 (8.3)	11 (6.4)	5 (6.7)	P = .13	P = .13	Not enrolled > Enrolled	p = .69		p = .47	p = .24

Note. NMUPD = Non-medical use of prescription drugs. p-values are Bonferroni corrected for multiple comparisons within-model. OR = odds ratio.

¹Not enrolled in college classes is reference group for college attendance odds ratios.

^*p < .05.

^**p < .01.

^***p < .001.

GEE models for the following motivations were unable to converge or received an error for concerns about validity: for stimulants, addiction; for CNS depressants, seeking deeper understanding, to have a good time with friends, fit in with a group, to decrease the effects of other drugs, and addiction; and for opioids, to feel good/get high, seeking deeper understanding, fit in with a group, to get away from problems, to increase the effects of other drugs, to decrease the effects of other drugs, as a substitute for heroin, and addiction.

4. Discussion

The current study investigated developmental changes in NMUPD motivations during young adulthood in a nationally representative sample across 36 cohorts who reported NMUPD. While participants most commonly reported infrequent use across prescription drug class (1–2 occasions in the past year), certain motivations were popular across the three waves of data for NMUPD, and the majority of young adults endorsed multiple motivations, with four to five popular reasons for using each prescription drug class, including the clinical reasons the medications were prescribed (e.g., opioid for pain relief), self-treatment motivations (e.g., to relax; to sleep), and recreational motivations (e.g., to feel good or get high). Our results support both the self-medication hypothesis (Duncan, 1974; Khantzian, 1985, 1997) and Arnett’s theory (2005) on recreational substance use in emerging adulthood, and are similar to prior cross-sectional research of young adults (for a review, see Drazdowski, 2016) and to other population-based studies on NMUPD motives across adulthood (e.g., Compton et al., 2017; Han et al. 2017; Hughes et al., 2016). Our findings also support the use of personalized feedback as an intervention for NMUPD, similar to prior studies of alcohol interventions in college students (Blevins & Stephens, 2016).

4.1. Changes in NMUPD motivations across young adulthood

According to GEE models, motivations to engage in NMUPD were relatively stable over time. As a result, many of the study’s hypotheses were only partially supported, and support varied by prescription drug class. Uniquely, for CNS depressants the motivation to decrease the effects of some other drug(s) increased between waves 1 and 3, with a striking effect size (OR = 13.99) at wave 3 compared to wave 1. This finding suggests that CNS depressants are more likely to be combined with other drugs, especially as young adults age. As young adults age, they may be exposed to a variety of substances and therefore are more likely to combine drugs. Indeed, in other young adult and college student samples, those who engage in NMUPD were more likely to endorse using other drugs, binge drinking, and combining prescription drugs with other substances (Advokat et al., 2008; Barrett et al., 2005; Garnier et al., 2009; Hughes et al., 2016; McCabe et al., 2007, 2009, 2019a). These behaviors increase the risk of potentially dangerous drug interactions. Specifically, the disinhibitory and aggressive effects that CNS depressants can have, as well as the risk for opioid-related deaths, increase when individuals combine CNS depressants with alcohol and other drugs (Jann, Kennedy, & Lopez, 2014; Lader, 2011; Webster et al., 2011). Additionally, if young adults are using a variety of drugs, prevention or treatment programs may not be effective if they only target NMUPD. More research is needed to understand how NMUPD fits into the larger drug culture and how and why using CNS depressants to regulate the effects of other drugs increases over time.

We predicted that recreational motivations would decrease over time across all prescription drugs classes as young adults move toward more life stability (Arnett, 2005). Our findings partially supported this hypothesis. Our findings are in line with previous findings that older adolescents report recreational motives more often than self-medication motives for NMUPD (McCabe et al., 2019c). As a result, treatment that targets these recreational motivations and provides rewards for non-NMUPD-related activities (e.g., Petry et al., 2006) may be more appropriate when individuals are entering young adulthood (when these motivations are most common), at least for stimulant and opioid medications.

On the other hand, the motivation to feel good or get high was consistent over time and participants endorsed these motivations across prescription drug classes. Since this motivation remained stable and popular across young adulthood, public health interventions for young adults in general (rather than only high school seniors, first-year college students, or 18/19 year olds), should focus on addressing the motive to feel good or get high. The popularity and consistency of this motivation calls into question parts of Arnett’s theory (2005), which suggests that young adults engage less in NMUPD recreationally as they become more stable and potentially experience less stress. Additionally, according to research on brain development, young adults should have the cognitive capacity to make methodical rather than emotion-driven decisions (e.g., Steinberg, 2008), which would decrease these emotionally related motivations. Therefore, more work needs to be conducted to see if by age 24 young adults are actually experiencing less stress and negative affect associated with disruptions in life as these theories propose. Research has found that many young adults feel that they are reaching “adulthood” later, as traditional markers of adulthood (e.g., getting married, buying a house; Hutchison, 2015) are delayed; thus, this motivation may not actually decrease until a later age than the ages that we investigated in this study (e.g., 30 years old; Arnett, 2000). This speculation could be tested using later cohorts from the MTF study when more data become available in later follow-up waves.

We also predicted that self-treatment motivations would increase over time across all prescription drug classes because exposure to prescription drugs and the risk for psychiatric problems increases (e.g., Kessler et al., 2005) under the self-medication hypothesis (Duncan, 1974; Khantzian, 1985, 1997). Our findings partially supported this hypothesis in all prescription drug classes, which suggests that outreach may be needed to counter these motivations as young adults age. Also, targeted evidence-based treatments in early young adulthood may help to prevent nonmedical self-treatment. For example, introducing and promoting ways to increase individuals’ energy through the use of exercise (Haskell et al., 2007) or to decrease physical pain through the use of nonpharmacological treatments (e.g., Garg, Joshi, Mishra, & Bhatnagar, 2012) may help to prevent young adults from engaging in NMUPD.

Other self-treatment motivations (e.g., stimulants: to stay awake; CNS depressants: to get sleep; opioids: to relax or relieve tension) were stable across the three waves of data collection. This is not to say that these motivations are not important intervention targets for prevention and treatment efforts; rather, because this population commonly reported these self-treatment motivations and they remained stable over time, mental health professionals should target these motivations throughout young adulthood. Connecting young adults to mental health professionals who can accurately diagnose and prescribe appropriate treatments, including medications, may help to alleviate the symptoms that cause self-medication, while reducing the risks associated with NMUPD; for example, different substances’ negative interactions when taken while not under the care of a medical professional (McCabe et al., 2009). Such interventions are especially important because young adults may also misdiagnose their symptoms, resulting in inappropriate use of medications, which in turn may lead their underlying problems to become worse (Holloway & Bennett, 2012).

Moreover, participants relatively commonly endorsed using prescription CNS depressants and opioid medications as a sleep aid, and reported using stimulants to get more energy and to stay awake. Since research has shown that medications to reduce sleep problems decline in effectiveness over time and there is research that supports psychosocial treatments (e.g., cognitive-behavioral therapy) as evidence-based treatments for insomnia (Morin, 2010), assessing and targeting sleep problems may be an effective prevention and intervention tool for reducing NMUPD in young adults. Additionally, evidence has shown that young adults who use prescription stimulants nonmedically report worse subjective and overall sleep quality, and more sleep disturbances, compared to their peers who do not endorse such use (Clegg-Kraynok et al., 2011). Targeting sleep problems may be beneficial to all young adults regardless of the prescription drug class they use.

4.2. Sex effects on NMUPD motivations in young adults

In general, for GEE models that were able to converge, more males endorsed recreational motivations compared to females, reinforcing what has been suggested in other studies with samples of students in higher education (Lord et al., 2009, 2011; McCabe et al., 2007, 2009; Vest et al., 2018). Conversely, for CNS depressants and opioids, more females reported self-treatment motivations, or using the medications as they are clinically intended, compared to males; this finding is also in line with previous work (Lord et al., 2011; McCabe et al., 2007, 2009). For stimulant medications, females did not report more self-treatment motivations, which another study also found (McCabe et al., 2009). Since recreational and self-medication motives were the most common motives for both sexes and sex-motivation interactions were small, treatment interventions should focus on providing or improving patients alternative recreational activities with fewer negative consequences (e.g., sports leagues) and ways to manage physical and mental health–related pain for both males and females.

Notably, however, more females endorsed using stimulants nonmedically to lose weight. This contradicts previous work on this topic, which found that males and females reported using stimulants nonmedically for weight-loss purposes (Jeffers & Benotsch, 2014; Jeffers, Benotsch, & Koester, 2013). However, these studies were conducted at one university, which may limit their findings, and our results, from our nationally representative sample, indicate that this finding may not be true for the broader young adult population who engage in NMUPD. More females reporting weight loss as a motive is not surprising though, given that in the general population, young adult females tend to report more concerns about weight loss, using multiple methods to achieve weight loss, and are diagnosed with more eating disorders than males (Wilson, Grilo, & Vitousek, 2007). That 31% to 35% of nonmedical stimulant users reported using the drug for weight loss is a significant motivation to include weight-loss motivations in prevention and intervention efforts, specifically for those aimed at treating young adult female populations. Programs that promote a healthy lifestyle and positive body image in young adults, specifically with weight loss as an outcome (e.g., Turk et al., 2009), could be effective at reducing young adults’ (particularly females’) nonmedical use of prescription stimulants for weight loss.

Sex moderated the changes in two NMUPD motivations over time: to feel good/get high for stimulant medications and to have a good time with friends for CNS depressants. However, given the paucity of moderation findings, even though we implemented protections against Type I errors within models, it is possible that these results are the consequence of chance and should be interpreted with caution. Nevertheless, because some of our findings highlight recreational motivations, they are in line with previous work that indicates that as young adults age, they may become more mature and engage in NMUPD less frequently for recreation (Arnett, 2005). Further, our results support research on sex differences in risk-taking behaviors in adolescence and young adulthood, in which females have been found to mature more quickly in decision-making processes than males (e.g., Harris, Jenkins, & Glaser, 2006). Future studies need to replicate these results, nevertheless.

4.3. College attendance effects on NMUPD motivations in young adults

Our results from the CNS depressant and opioid models indicate that college attendance did not moderate change in motives over time, in line with theory that young adults both in and out of college experience the same development and life stage–associated changes (Arnett, 2005). We found no general differences in young adults’ motivations for the nonmedical use of prescription stimulants, although we found no moderation. One study that included a sample of young adults both enrolled and not enrolled in college found that, in general, the sample reported using stimulants nonmedically both as the medications were intended and for recreational purposes, similar to studies of college students only (Upadhyaya et al., 2010). However, this study did not specifically examine differences in NMUPD motivations by college attendance. The current study found that young adults not enrolled in college courses were more likely to endorse using stimulants nonmedically to relax or relieve tension, to feel good/get high, have a good time with friends, and to help lose weight. This finding highlights that young adults not in college may be at higher risk for using stimulants nonmedically for unique reasons. Future research should continue to include young adults not in college to replicate these findings. Also, since we were unable to reliably run a variety of motivations, research needs to look more closely at some of the less-often reported motivations and those specifically in CNS depressant and opioid prescription drug classes.

4.4. Cohort effects

Given the changes in NMUPD frequency (e.g., Lipari et al., 2015; Schulenberg et al., 2019) and the increasing availability of prescription drugs over time (Yu, 2012), we controlled for cohort effects in the current study to examine changes in motivation over time and by gender and college attendance above and beyond historical trends in motivation for NMUPD. Over time, cohorts were more likely to report using CNS depressants to relieve physical pain and to increase the effects of other drugs. This finding may coincide with increases in CNS depressant use that began in the early 2000s but had decreased by 2013; however, in 2013, the rate of use was still almost double the 1975 rate of CNS depressant use (Schulenberg et al., 2019). This is the first study to show that population-based increases in motivations to use may coincide with prevalence of use over time. Nonsignificant cohort effects on motives for nonprescription opioid misuse are similar to lack of cohort effects in adolescent longitudinal research (McCabe et al., 2019a). All other significant cohort effects were significant only in sex or college attendance analyses, suggesting that these changes may coincide or interact with changes in motivations by sex and college attendance. Overall, changes in motivations over subsequent cohorts were small in size; however, these changes are large on a public health scale, particularly, since they may account for increases in use over time. Current public health interventions should aim to decrease motivations for NMUPD to get high, have a good time with friends, relax, relieve pain, and increase the effects of other drugs; and replace NMUPD with other coping skills, recreational activities, and pain management.

4.5. Limitations and future research

As with all research, this study has limitations that need to be acknowledged and ideally addressed in future work. First, this study was a secondary data analysis of a reduced dataset, which limited the types of questions and confounding variables that we could assess. Future work needs to consider other factors that may influence NMUPD motivations in this study’s population, including other demographic factors (e.g., geographic location, GPA, race/ethnicity) and other known risk factors for drug use (e.g., stress, trauma, other drug use). Second, the current study assessed only frequency of NMUPD and does not include measures of problematic NMUPD. Since most of the sample reported infrequent use, investigating problematic use could likely lead to more information on which motivations may be most appropriate to target in prevention and treatment efforts to make the most impact. Additionally, even with a large dataset, NMUPD frequencies were low, and so were some of the reported motivations for use, therefore, it was difficult for us to gain a thorough understanding of less frequently endorsed motivations. Research with clinical samples of individuals who misuse prescription drugs could investigate the importance of less frequently reported motivations for use. Spending energy and resources on these motivations may not be as useful on a larger scale compared to focusing on NMUPD motivations that are more common in young adulthood. Future work using factor analyses or other methods of data reduction may be helpful in determining how these motivations may group together in a clinically relevant way. Since this is the first study to examine motivations of NMUPD across drug classes longitudinally and because prior work has not found homogeneity of motivations across stimulants, opioids, and CNS depressants using data reduction methods with existing formulations of motives for use (Messina et al., 2016), we examined changes in motives in an exploratory manner. In addition, we could not examine changes in use and misuse of the same drug and drug class in this study; for example, if an individual were prescribed a medication at wave 1, but were misusing the same medication at following waves, there was not a way to match these data. Understanding how individuals begin misusing (e.g., from originally prescribed medications compared to in social contexts or illicit purchasing) will inform prevention and intervention efforts. This study also relied on self-reports, which are subject to social desirability and call into question the accuracy of participants’ reports of stigmatized behaviors, such as drug use (Kazdin, 2016). Therefore, multi-informant research should be considered in the future. Last, some research has examined interventions specifically aimed at increasing knowledge of coping motivations to decrease drinking to cope and drinking problems (Blevins & Stephens, 2016). Future work should explore if interventions for NMUPD benefit from including personalized feedback to increase knowledge of NMUPD motivations.

5. Conclusion

Even with these limitations, the current study had several strengths. It is the first work to investigate the changes in NMUPD motivations across young adulthood in a nationally representative sample. Results from this study need to be replicated but can be generalized to the larger American young adult population who engage in NMUPD. As NMUPD motivations were generally stable over time, the best approach to prevention and intervention programs for NMUPD based on motivations is to address the most popular motivations. Since endorsing more than one motivation was common, interventions need to be multifaceted and address both recreational and self-treatment motivations. Research should also explore sex differences further, including sex-specific treatment interventions. Young adults both enrolled and not enrolled in college classes need to be targeted; although, those not enrolled seem to have different motivation patterns for stimulant medication use. Broad scale interventions and public health education efforts should address motivations for use, and research should further explore differential patterns by sex and college enrollment.

Highlights.

• This study tested motives for non-prescription drug use in young adults over time

• Recreation and self-treatment motives were common over time and across drug classes

• Men were more likely to endorse recreational motivations

• Women were more likely to endorse self-treatment motivations

• College course enrollment impacted motivations for stimulant use