Sources for Prescription Stimulant Misuse: A Person-Centered Approach to Understanding Links to Substance Use and Psychiatric Impairment

Abstract

Prescription stimulant misuse (PSM) is most prevalent among college students and is associated with numerous negative academic and psychosocial outcomes. A large body of literature has identified predictors of PSM in this population, however few studies have utilized a person-centered approach to examine how the sources from which students procure prescription stimulants are associated with substance-related and psychiatric impairment. We used latent class analysis (LCA) to classify a geographically and racially/ethnically diverse sample of U.S. undergraduates (N=538) who misused prescription stimulants into groups based on their endorsement of nine sources of medication. We selected a five-group classification from the LCA with classes of Peer/Dealer, Given by Friend, Own Prescription, Lower Multiple Sources (i.e., relatively infrequent endorsement of multiple sources), and Any Source. Compared to the reference group (Given by Friend), the Own Prescription class was less likely to report marijuana use, simultaneous alcohol and marijuana use, alcohol or marijuana consequences, and non-oral routes of administration. On the other hand, the Own Prescription class was more likely to screen positive for anxiety, anger, and suicidality. Similarly, the Lower Multiple Sources group was more likely to screen positive for depression, anxiety, anger, and suicidality. Prevention and intervention efforts focused on PSM may be tailored differently for students who are misusing their own medication and/or endorsing multiple sources. Specifically, these students may need broader assistance with comorbid psychiatric conditions, particularly suicidality, while students who obtain stimulants from peers or a dealer may benefit more from substance-focused interventions.

Prescription stimulant (e.g., Adderall, Ritalin, Concerta) misuse (PSM), which involves use without a prescription or in ways a prescriber did not intend, has received more widespread attention in the past 15 years in college students due to substantial prevalence estimates (17%; Benson et al., 2015) and an increase in prevalence since 2003 (McCabe et al., 2014). PSM can lead to adverse physiological and psychological effects ranging from lack of appetite, insomnia, restlessness, rapid heartbeat, and irritability (Hartung et al., 2013; Rabiner et al., 2009) to more severe outcomes such as addiction and emergency room visits, particularly when prescription stimulants are combined with other substances (Chen et al., 2016; Faraone et al., 2020; Schepis et al., 2021). PSM is most often motivated by a desire to increase alertness and concentration, often to complete academic work (Benson et al., 2015). However, PSM has not been associated with improvements in grades over time (Arria et al., 2017); in fact, it has been repeatedly associated with poorer academic outcomes such as a lower grade point average and missing class (Arria et al., 2013; Benson et al., 2015).

Although there is a well-developed body of literature documenting the prevalence estimates, risk factors, and motives for PSM, there is little research examining how the sources from which people procure stimulants are associated with substance-related problems and psychiatric difficulties. Most research on PSM has reported on medication source(s) descriptively, but few studies have examined associations between medication source and broader psychiatric functioning. If medication source is a proxy for risk, this information could inform selective interventions for PSM tailored to individuals procuring from riskier sources (Compton et al., 2018).

An early study of adolescents and adults found that the most common source was friends/relatives, which included getting the medication for free, purchasing, or stealing it (Chen et al., 2014). Yet many individuals reported sourcing from physicians, which was associated with higher rates of depression, anxiety, DSM-IV dependence on one or more drugs, and history of mental health or addiction treatment compared to friend/relative sources. Other studies corroborated and expanded these findings, indicating that individuals who source from a physician, unknown others (i.e., stranger, dealer), or multiple sources reported increased frequency of misuse and likelihood of a stimulant use disorder, particularly if paying for the drug rather than obtaining it for free (Compton et al., 2018; Ford & Lacerenza, 2011; McCabe et al., 2018). Ford and Lacerenza (2011) speculated that individuals who buy from a stranger may use stimulants more for recreational purposes (e.g., to get high) as opposed to instrumental purposes (e.g., to study), which might explain their broader susceptibility to other prescription drug misuse and related problems. Taken together, these studies indicate that: (1) purchasing stimulants from a dealer or stranger or obtaining them from a physician is associated with more problematic substance use; and conversely (2) obtaining the medication for free, particularly from friends or family, may be associated with fewer substance-related problems.

Recent research utilized a person-centered approach (i.e., latent class analysis) to provide a nuanced understanding of medication source and substance-related problems. McCabe et al. (2019) identified five latent classes of participants based on ten medication sources across three prescription drug classes. Consistent with prior results (McCabe et al., 2018), the “multiple sources” class was at increased risk for prescription drug misuse and other substance use. In addition, procuring or buying from a friend was associated with more binge drinking, cigarette smoking, and marijuana use, and this pattern was more common among males. The lowest impairment was among those using their own leftover prescription, a behavior more common in females. These findings are consistent with the characterization of “active” vs. “passive” procurement of medication linked to problematic use and recreational motives (Daniulaityte et al. 2014, Ford & Lacerenza, 2011). It is unknown, however, if the findings for 12th grade students generalize to college students and to stimulants, specifically. While the two groups are close in age, the risk of peer-to-peer diversion and other forms of nonadherence may increase in college as individuals become more responsible for their own medication management.

The Current Study

We sought to improve upon and extend the prior research in several ways. First, we focused on a large, diverse sample of college students, as doing so allowed us to characterize patterns of prescription stimulant acquisition and associated substance and psychiatric impairment in those at highest risk for PSM and its negative sequelae (Faraone et al., 2020). Second, in contrast to several studies which focused on the single most recent source of prescription stimulants (Chen et al., 2014; Compton et al., 2018; Ford & Lacerenza, 2011), participants in the current study could endorse multiple sources. By expanding the response options, we were able to utilize a person-centered approach (i.e., latent class analysis) to examine whether there was patterned endorsement of multiple sources that was associated with psychiatric difficulties. Relatedly, we were able to provide more detail than prior studies since our source response options differentiated between peer and family sources. Finally, our focus on a single class of prescription drugs (i.e., stimulants) provided insights into unique patterns associated with PSM sources and more clearly informs preventive interventions for students with and without stimulant prescriptions.

Because a latent class analysis yields novel categorizations of participants, we did not advance specific hypotheses about the number of classes we would identify. Nonetheless, based on prior research, we expected elevated substance or psychiatric impairment in classes whose members endorsed obtaining prescription stimulants from multiple sources, especially buying from a dealer, friend, relative, and/or online. Conversely, we anticipated that misusing one’s own prescription (or prior prescription) or getting stimulants from a friend for free would be associated with lower levels of psychiatric and substance-related impairment.

Method

Participants and Procedure

Undergraduates at seven universities in six U.S. states (Colorado, New Mexico, New York, Virginia [two sites], Texas, and Wyoming) from Psychology Department participant pools completed an anonymous online survey for research credit between Fall 2019-Spring 2020 [see Looby et al. (2021) for more details]. The current study focused on a subset of participants (n = 538) from the larger sample (N = 4,764) who endorsed using prescription stimulants in the past year either without a prescription or, if prescribed, using in ways not intended by a prescriber. Our sample’s demographic characteristics are summarized in Table 1. This study was approved by the University of Wyoming IRB using a single-site IRB model.

Table 1.

Descriptive Statistics for Participants Endorsing Past-Year Prescription Stimulant Misuse (PSM)

Variable	n (%) or M(SD)
Sex at birth
Female	367 (68%)
Male	169 (32%)
Gender (n, %)
Female	366 (68%)
Male	168 (31%)
Gender Non-Binary/ Did Not Identify	3 (1%)
Race/Ethnicity (n, %)
Asian/Asian American/Pacific Islander	13 (2%)
African American/Black	33 (6%)
Hispanic/Latino	32 (6%)
White/non-Hispanic	329 (62%)
Other/Mixed Race/Ethnicity	128 (24%)
Age [M(SD)]	20.09 (2.74)
Route(s) of administration for PSM (n, %)
Swallow	452 (84%)
Snort	140 (26%)
Smoke	43 (8%)
Inject	0 (0%)
Days past month alcohol use [M(SD)]	7.35 (6.50)
Days past month marijuana use [M(SD)]	11.71 (11.90)
Days simultaneous alcohol and marijuana use [M(SD)]	7.66(8.71)
Alcohol consequences [M(SD)]	7.89 (5.85)
Marijuana consequences [M(SD)]	5.99 (5.02)
Other prescription drug misuse in previous year
Opioids	22%
Benzodiazepines	28%
Sleeping medications	15%
DSM-5 Symptoms (n, %)
Depression	60%
Anger	45%
Anxiety	63%
Sleep problems	41%
Suicidality	30%

Measures

Demographics.

A researcher-developed demographics measure assessed age, biological sex assigned at birth, gender identity, race, ethnicity, and year in school.

Prescription Drug Misuse.

We used a researcher-developed measure to assess use of prescription medications including opioids (e.g., Vicodin, Percocet), benzodiazepines (e.g., Xanax, Valium), sleeping medications (Z-drugs, e.g., Ambien, Sonata), and stimulants (e.g., Adderall, Ritalin). Participants first indicated if they engaged in lifetime use of each drug. Specific examples of each drug class were provided. If lifetime use was endorsed, participants indicated their procurement source(s) if they engaged in past-year misuse for each drug. Misuse was defined as use without a prescription; or if prescribed, use in ways not intended by the prescriber, such as at higher doses, more often, or not by mouth. Source questions were similar to those used in the Monitoring the Future Study (Miech et al., 2018), with the addition of “roommates” to several of the questions. Participants could select all sources that applied from the following list: took from a roommate/friend without asking; took from a relative without asking; given for free by a roommate/friend; given for free by a relative; bought from a roommate/friend; bought from a relative; bought from a drug dealer/stranger; bought on the internet; it was from my own prior prescription.

Alcohol Use and Related Consequences.

Participants were asked if they had ever used alcohol in their lifetime, and if so, the number of days in the last month on which they used alcohol. If participants reported any past-month alcohol use, they also completed the Brief Young Adult Alcohol Consequences Questionnaire (B-YAACQ; Kahler et al., 2005). The B-YAACQ is a 24-item measure that assesses a range of problems that may occur because of alcohol use. Participants dichotomously responded to each item to indicate whether each problem was experienced because of their alcohol use in the past month. Scores were summed to indicate the number of past-month alcohol-related problems. Internal consistency in the current sample was excellent (α = .90).

Marijuana Use and Related Consequences.

Participants were asked if they had ever used marijuana in their lifetime, and if so, the number of days in the last month on which they used marijuana. If participants endorsed any past-month marijuana use, they also completed the Brief Marijuana Consequences Questionnaire (B-MACQ; Simons et al., 2012). The B-MACQ is a 21-item measure that assesses a range of problems that may occur because of marijuana use. Participants dichotomously responded to each item to indicate whether each problem was experienced because of their marijuana use in the past month. Scores were summed to indicate the number of past-month marijuana-related problems. Internal consistency in the current sample was very good (α = .89).

Mental Health Indicators.

Mental health symptoms were assessed via the DSM-5 Level 1 Cross-Cutting Symptom Measure (American Psychiatric Association, 2013). This measure assesses 23 psychiatric symptoms across 13 mental health domains and has demonstrated good internal, convergent, and criterion-related validity with college students (Bravo et al., 2018). For this study, we examined the following domains: depression (2 items; α = .86), anger (1 item), anxiety (3 items; α = .87), sleep problems (1 item), and suicidality (1 item). Participants rated each item based on the severity or frequency with which each symptom was experienced in the past two weeks along a 5-point scale (0 = none: not at all to 4 = severe: nearly every day). Clinical thresholds of scores of 2 or greater on any item (or 1 or greater for suicidality) have been suggested for use for further inquiry of each domain. For the present study, we used this clinical cutoff to indicate presence or absence of a positive screen for each mental health indicator.

Data Analysis Plan

All analyses were performed in Mplus 8.6 (Muthén & Muthén, Los Angeles, CA) and Stata 16.1 (StataCorp, College Station, TX). Latent class analysis is a participant-focused technique used to uncover unique response patterns among sets of individuals. LCA attempts to use recurring patterns of similar multivariable responses to a set of variables to identify classes (Nylund-Gibson & Choi, 2018; Weller et al., 2020). For this study, the variables used to identify latent classes (also known as indicators) were responses to the procurement source questions among those engaged in past-year prescription stimulant misuse. LCA usually identifies multiple underlying classes of participants in a given sample, allowing for a personalized characterization of individual responses to both the indicators and covariates (e.g., other substance use) that goes beyond overall sample means. Details about how we created the latent classes and selected an LCA solution are described in the Supplemental materials.

We determined prevalence rates of membership in the latent classes using cross-tabulations. Then, using most likely class membership as a between-subjects variable, we compared latent classes on our substance use and mental health outcomes. We calculated cross-tabulations to determine the proportions within each latent class that endorsed any substance use behavior and means and standard deviations for days of alcohol or marijuana use and alcohol or marijuana consequences.

We used negative binomial regression models to compare the classes on count-based variables (i.e., days of alcohol use in the past 30, days of marijuana use in the past 30, number of alcohol use consequences, number of marijuana use consequences, and days of simultaneous alcohol-marijuana use in the past 30), controlling for age, biological sex, and race/ethnicity. Negative binomial regression was chosen due to the overdispersion of the data (Gardner et al., 1995) and was supported by likelihood ratio tests that suggested better fit than in a comparable Poisson regression model. Except for days of alcohol use in the past 30 (which used a simple negative binomial regression model), all models used zero-inflated negative binomial regression models, given the high prevalence of zeros in the data due to non-use of marijuana or a lack of alcohol use consequences (Yao et al., 2003).

We used logistic regression analyses for dichotomous outcomes (i.e., any misuse of opioid, benzodiazepine, or sleeping medication (analyzed separately), non-oral PSM, and the selected mental health subscales of the DSM-5 Cross-Cutting Measure), also controlling for age, biological sex, race/ethnicity and site. For both the negative binomial and logistic regression analyses, the “Given by Friend” class (described below) was set as the reference, given the relatively lower risks associated with this source category in prior research (Compton et al., 2018; McCabe et al., 2018).

Transparency and Openness

To reduce fatigue and minimize burden, we utilized a planned missing data design (i.e., matrix sampling; Graham et al., 2006; Schafer, 1997), employed in similar multi-site studies with college students (e.g., Bravo et al., 2018), which allowed for the survey to be completed in one hour. Participants completed a battery of core measures on substance use and mental health, plus a random sample of 10 measures from a larger pool (22 total measures) that assessed personality, cognitions, and other health behaviors. All measures included in the present analysis were core measures. Materials and analysis code for this study are available by emailing the corresponding author. The study was not preregistered.

Results

Details about our latent model class selection and fit indices can be found in the Supplemental materials. Within the five-class solution, the latent classes were Peer/Dealer (n = 210; 39.0% of the analytic sample), Given by Friend (n = 148, 27.5%), Own Prescription (n = 89, 16.5%), Lower Multiple Sources (n = 71, 13.2%), and Any Method (n = 20, 3.7%). The prescription stimulant source endorsement patterns for each class are captured in Figure 1. The Peer/Dealer group was marked by high endorsement of prescription stimulant purchases from friends (81% of the class), given a stimulant by a friend (51%), and purchases from a dealer (40%). The Given by Friend group had 100% endorsement of being given a prescription stimulant by a friend and low endorsement of all other sources (< 10%). Similarly, the Own Prescription class had 100% endorsement of misuse from one’s own stimulant prescription and very low endorsement of all other source categories (< 5%). The Lower Multiple Sources class had at least 20% endorsement of six sources, with given by a relative as the most endorsed source (50%), followed by given by a friend (36%). Finally, the Any Method class had very high endorsement of every source category (i.e., all above 90%). Given the small size of this class, it was excluded from further inferential analyses, though we present descriptive characteristics of this group in Table 2 and Figure 1. Table 1 summarizes the overall sociodemographic characteristics of the analytic sample.

Figure 1.

Prevalence of Sources by Stimulant Source Group (5 Class Solution)

Table 2.

Substance Use, Substance-Related Problems, and Psychiatric Impairment by Stimulant Source Group

Variable	Given by Friend (n=148)	Peer/Dealer (n=210)	Own Prescription (n=89)	Multiple Sources (n=71)	Any Method (n=20)
Non-oral prescription stimulant use (%)	36%	38%	14%†	31%	40%
Past month substance use
Any alcohol use	90%	93%	78%	89%	70%
Any marijuana use	67%	72%	51%†	62%	50%
Days past month alcohol use [M (SD)]	7.06 (6.30)	8.34 (6.78)	6.00 (6.48)	6.42 (6.03)	5.15 (5.70)
Days past month marijuana use	10.90 (11.75)	12.12 (12.16)	7.26 (10.98)	8.20 (10.99)	8.70 (11.65)
Any simultaneous alcohol and marijuana use	53%	61%	36%†	47%	30%
Days simultaneous alcohol & marijuana use	4.07 (6.95)	6.04 (8.56)	2.94 (6.84)	3.66 (7.04)	2.60 (6.79)
Substance-related consequences
Any alcohol consequences	80%	86%	69%†	80%	70%
Any marijuana consequences	57%	65%	39%†	52%	40%
Number of alcohol consequences	5.98 (5.21)	8.13‡ (5.96)	5.40 (6.09)	7.25 (6.94)	7.85 (7.10)
Number of marijuana consequences	3.51 (4.51)	4.57 (5.03)	2.87 (4.52)	4.10 (5.70)	3.40 (5.63)
Other prescription drug misuse in previous year
Opioids	20%	24%	23%	28%	5%
Benzodiazepines	27%	31%	20%	34%	15%
Sleeping medications	14%	11%	25%	25%	0%
DSM-5 Symptoms
Depression	55%	58%	66%	69%‡	65%
Anger	35%	44%	58%‡	51%‡	60%
Anxiety	52%	61%	84%‡	65%‡	55%
Sleep problems	35%	42%	47%	42%	45%
Suicidality	22%	29%	36%‡	42%‡	50%

Note. Given by Friend was the reference group in all analyses. Any Method was excluded from all group comparisons.

^†Value was significantly lower than the reference group.

^‡Value was significantly higher than the reference group.

Substance Use Outcomes by Latent Class

Substance use outcomes are summarized in Table 2. As compared to the reference group (Given by Friend), those engaged in misuse of their own prescription stimulant medication generally had lower prevalence of past 30-day marijuana use (51% vs. 67%; B = 0.66, SE = 0.30, z = 2.19, p = .028). Similarly, the Own Prescription class had a lower prevalence rate of past 30-day simultaneous alcohol-marijuana use than those in the Given by Friend class (36% vs. 53%; B = 0.95, SE = 0.37, z = 2.56, p = .011). In both cases, mean days of either marijuana use or simultaneous alcohol-marijuana use did not differ between the Given by Friend and Own Prescription classes; also, there were no other significant differences from the Given by Friend reference group for 30-day alcohol use, marijuana use, or simultaneous use.

For alcohol use consequences, two groups differed from the Given by Friend class. The Own Prescription class had a lower prevalence rate of any consequence (69%) versus the Given by Friend class (80%; B = 0.70, SE = 0.34, z = 2.04, p = .042). On the other hand, the Peer/Dealer class had a higher count of consequences (M = 8.13) than the Given by Friend class (M = 5.98; B = 0.27, SE = 0.08, z = 3.24, p = .001). For marijuana consequences, only one difference was found: the Own Prescription class had a lower prevalence rate of any consequences (39%) compared to the Given by Friend class (57%; B = 0.74, SE = 0.29, z = 2.51, p = .012).

Regarding other prescription medication misuse, no differences were found between the Given by Friend class and other latent classes in terms of lifetime opioid, benzodiazepine, or sleeping medication misuse prevalence. Finally, the Own Prescription latent class had lower odds of non-oral stimulant misuse versus the Given by Friend class (OR = 0.29, 95% CI = 0.14–0.58).

Mental Health Outcomes

In contrast to the substance use outcomes, members of the Given by Friend class had the lowest prevalence of positive screening scores on all mental health outcomes (see Table 2). Compared to the Given by Friend class, the Lower Multiple Sources group had higher odds of positive screening scores for depression (OR = 1.95, 95% CI = 1.05–3.63), anger (OR = 2.01, 95% CI = 1.10–3.64), anxiety (OR = 1.87, 95% CI = 1.02–3.45), and suicidality (OR = 2.63, 95% CI = 1.41–4.91). Also, the Own Prescription latent class had higher odds of positive screening scores for anger (OR = 2.42, 95% CI = 1.39–4.21), anxiety symptoms (OR = 4.78, 95% CI = 2.45–9.33), and suicidality (OR = 1.89, 95% CI = 1.05–3.41). Of note, the percentage of students screening positive for suicidality ranged from 22% and 29% for the Given by Friend and Peer/Dealer latent classes, respectively, to 42% and 50% for the Lower Multiple Source and Any Method latent classes. The Own Prescription class was intermediate, at 36%.

Discussion

The current study identified latent classes of college students based on how they procure stimulants and whether those classes differed on indices of substance use, substance-related problems, and psychiatric symptoms. More than 1 in 10 students from this large, diverse sample of US undergraduates endorsed procuring stimulants for misuse in the previous year, with five distinct groups emerging. Students evidenced differences in their likelihood of paying for stimulants versus getting them for free, using one source versus multiple sources, and using one’s own prescription versus someone else’s. Specifically, purchasing from peers/dealers or obtaining from friends for free were the two largest classes. Although somewhat less common, a sizable proportion misused their own medication or endorsed using a wide range of sources, albeit with low frequency. A small number cited using all sources and methods of procurement (i.e., stealing, obtaining for free, buying) at a high frequency. Compared to McCabe et al.’s (2019) latent class analysis of high school seniors, the percentage of participants misusing their own prescription in our study was nearly identical (16.9% vs. 16.5%, respectively). On the other hand, compared to McCabe et al. (2019), the class of participants in our study who most often purchased stimulants was somewhat larger (39%) than McCabe et al. (2019) (28.3%) and the percentage who largely received them for free in our study (27.5%) was somewhat smaller compared to participants in McCabe et al. (40.6%).

We found mixed support for our first hypothesis that purchasers of prescription stimulants would evidence more substance-related and psychiatric impairment. Consistent with our prediction, students who primarily purchased stimulants (Peer/Dealer group) differed from the reference (Given by Friend) group, but only on the endorsement of more alcohol consequences. Compared to previous research, the Peer/Dealer group may have exhibited less impairment than purchasers in previous research for several reasons. First, Ford and Lacerenza (2011) differentiated between purchasing from dealers/strangers and from family/friends, with the former experiencing more impairment. Since we allowed participants to endorse more than one source, ultimately we showed that students who purchased stimulants from dealers also frequently purchased from friends. The lack of emergence of a class that solely used dealers in the current study suggests that exclusive use of dealers was uncommon. Second, compared to the national US samples in Ford and Lacerenza’s (2011) and McCabe et al.’s (2018; 2019) studies, purchasers in our sample may have shown less impairment because we focused on college students. With the widespread availability of stimulants on college campuses (Faraone et al., 2020), purchasing likely is more normative than in non-college settings. Finally, very high rates of alcohol and marijuana use in emerging adults (Schulenberg et al., 2021), may have made it more difficult to differentiate participants on these substance use behaviors, particularly because participants were already misusing prescription stimulants, a behavior consistently associated with other substance use (Faraone et al., 2020).

As predicted, the larger of our two groups that reported using multiple sources, the Lower Multiple Sources group, evidenced greater impairment than the Given by Peer group (the reference group), but only with respect to psychiatric symptoms (i.e., depression, anxiety, anger, and suicidality). Contrary to our prediction, the Lower Multiple Sources group did not evidence more substance use, prescription drug misuse, or substance-related consequences. McCabe (2018) showed that using multiple sources was associated with a higher risk of having a stimulant use disorder; in another study of high school seniors, the multiple sources group reported more prescription drug misuse and other substance use (McCabe et al., 2019). Of note, these studies did not report on psychiatric symptoms, only substance-related outcomes, so psychopathology comparisons are not possible.

Other reasons for the discrepant findings between our multiple sources group and participants endorsing multiple sources in prior research include the fact that McCabe et al. (2018) focused on all adults 18–25 years and might have captured emerging adults experiencing higher levels of impairment overall. Relatedly, given that fewer than two-thirds of high school seniors go directly to college (National Center for Education Management Statistics, 2021), the topography of PSM in our sample compared to that of the high school seniors in McCabe et al.’s (2019) sample likely differed due, in part, to differences in educational attainment, which has been associated with different patterns of sourcing prescription medication for misuse (Ford et al., 2020). Also of note, the multiple source group in McCabe et al. (2019) evidenced a different profile than our multiple sources group, in that their group reported a much higher frequency of purchasing stimulants from a friend or dealer than participants in our Lower Multiple Sources group. More frequent purchasing behavior, particularly among high school students where stimulant misuse is less normative, suggests that McCabe et al.’s (2019) multiple sources group engaged in more risky behavior than individuals in our multiple sources group.

We also found mixed support for our second hypothesis, namely that misusing one’s own prescription would be associated with less substance and psychiatric impairment. As predicted, the Own Prescription group reported less substance use (i.e., marijuana and simultaneous alcohol/marijuana use) and a lower likelihood of experiencing substance-related consequences and using non-oral routes of administration compared to the Given by Friend group. Accordingly, medication misuse was not necessarily indicative of problematic polysubstance use in this study. However, misusing one’s own prescription was associated with a greater likelihood of psychiatric difficulties, namely screening positive for anxiety, anger, and suicidality, which we did not predict. Presuming most students are prescribed stimulant medications for the treatment of ADHD, these data likely reflect the high rates of psychiatric comorbidity in college students with ADHD (Weyandt & Paul, 2013). Of concern, the higher rate of suicidality in this group suggests that prescribers should be especially attuned to nonadherence among patients who take stimulant medications and should screen for suicidality in cases where nonadherence is reported or suspected. Nonadherence may lead to poorer ADHD symptom management (Safren et al., 2007) and/or may be associated with nonadherence to other psychiatric medications (e.g., anti-anxiety, antidepressant), thereby increasing the risk of other psychiatric symptoms.

Our finding that students misusing their own prescription reported higher rates of psychiatric symptoms was consistent with Chen et al. (2014), who showed that individuals sourcing prescription stimulants for misuse from a physician had more psychiatric comorbidity. However, Chen et al. also showed that the group sourcing from a physician also was more likely to be dependent on one or more drugs. Because Chen et al. (2014) focused on adults across the lifespan, they may have captured impairment that developed over longer periods of time in individuals misusing their own medication. Consistent with this idea, a recent latent class analysis characterizing PSM trajectories between the ages of 18 and 50 showed that substance use disorder (SUD) symptoms at age 50 were most likely among individuals whose PSM peaked around age 40 (McCabe et al., 2022). Even though our findings differed, Chen et al.’s and McCabe et al.’s findings are cautionary in that they suggest SUDs could become more pronounced in the Own Prescription group if their PSM persists.

One unexpected finding was the extent to which psychiatric symptoms in our sample were elevated compared to a previous normative sample of college students. Bravo et al. (2018) reported prevalence rates among a large sample of college students with varied substance use histories of approximately 28% for both depression and anxiety symptoms, compared to rates of 60% and 45%, respectively, in the current study. Our sample also endorsed suicidality at a much higher rate (30%) than the normative sample (7%) (Bravo et al., 2018). Our findings were consistent with research showing more pronounced depressive symptoms and suicidality among college students who misused stimulants (Zullig & Divin, 2012) and higher suicidality in college students who reported a history of any prescription drug misuse (Vidourek et al., 2010). Overall, our findings suggest that regardless of the source(s) of stimulants for misuse, students with past year PSM were more psychiatrically impaired than the broader college population. Although it is impossible to determine the directionality of depressive symptoms and stimulant misuse from the extant research, some students may misuse stimulants to enhance mood (Benson & Flory, 2015; Schepis et al., 2020). Although PSM is often motivated by a desire for cognitive enhancement, a subset (15–28.5%) of college students report using to “feel better” (Rabiner et al., 2009) or to “get through the day” (Drazdowski et al., 2020), suggesting mood management likely motivates some students’ use.

Limitations

Our findings should be considered in the context of several limitations. First, given the cross-sectional nature of the data, we were not able to discern the nature of the relation between the source(s) for stimulant misuse, substance use and consequences, and psychiatric problems. As aforementioned, misusing one’s own prescription could lead to more dysregulated mood and behavior, and consequently, more psychiatric symptoms. Or, experiencing more psychiatric symptoms could lead students to misuse stimulants in an attempt to manage their mood. Another possibility is that there are common liabilities for psychiatric symptoms and medication misuse. Externalizing traits such as impulsivity and sensation-seeking, which may be more prominent among students with anger and suicidality, have been associated with PSM in college students (Chinneck et al., 2018). Or, there may be unknown confounding factor(s) not assessed in the current study that could account for the higher rates of psychiatric problems and suicidality in our sample compared to the normative sample.

A second limitation relates to our measure of prescription drug misuse. Although we inquired about sources for past-year medication misuse, we did not assess the frequency or quantity of past-year prescription misuse. Accordingly, we were not able to compare the classes on the amount of past-year prescription medication misuse, although we were able to compare the latent classes based on any past-year misuse of the three other classes of prescription drugs (sleeping, pain, and anti-anxiety). A third limitation of this study is its use of self-report substance use data, which could introduce bias due to underreporting. With that said, past studies strongly indicate that substance use self-report data is both reliable and valid (Darke, 1998; Napper et al., 2010; O’Malley et al., 1983). A final limitation relates to the generalizability of our findings, which is limited to students attending four-year colleges and further, to students who take Psychology courses in college. Recent research focused on sources for misused prescription opioids and tranquilizers showed that source patterns differed by educational attainment. Specifically, emerging adults who attend or graduated from college were more likely to misuse opioids prescribed by a physician and less likely to purchase opioids and tranquilizers (Ford et al., 2020). Although it is unknown if those findings apply to prescription stimulants, it suggests that the current findings may not apply to non-college attending emerging adults.

Future Directions & Implications

Future research that examines motivations for PSM (e.g., to focus/concentrate, to socialize, to get high) may be able to determine the extent to which certain source patterns co-occur with motivations for misuse. This information could inform the development of targeted interventions for specific groups similar to those identified in the current study. For example, if the Own Prescription group cited misusing to focus/concentrate, that might signal that their ADHD symptoms are not being treated sufficiently. On the other hand, if this group reported misusing primarily to feel more energetic or better about themselves, they might benefit from more support with mood management. Given that the Own Prescription group evidenced less substance use but more psychiatric problems than the Given by Friend group, it may be interesting for future research to examine if these groups differ with respect to social support and/or isolation. Without question, students’ reasons for engaging in misuse are varied; nonetheless, identifying patterns around how students procure their medication and misuse motives can improve our ability to tailor prevention messaging and intervention content. Future research also should explore how educational attainment relates to sources for stimulant misuse and substance and psychiatric impairment by primary source(s) of procurement. Finally, given that our analysis yielded two “multiple source” classes, one with a high frequency of using all sources, and one that utilized multiple sources with a lower frequency, future research should explore potential heterogeneity in multiple source groups with respect to their quantity and frequency of stimulant misuse and their motivations.

In conclusion, the current study showed that students who misuse their own stimulant prescription or those who utilize multiple sources for procuring stimulants are at greater risk for psychiatric symptoms than students who obtain these medications free from friends. These findings suggest that it would be beneficial for prescribers to assess adherence carefully and, in cases of nonadherence, screen for mood disorders and suicidality and inquire about the efficacy of their treatment, if applicable. This study also showed that irrespective of source, students who engaged in past-year PSM reported higher rates of depression, anxiety, anger, and suicidality compared to a normative sample. These findings suggest that preventive interventions and screening for PSM in college settings should inquire about psychiatric symptoms, particularly suicidality, in addition to substance misuse, as it may be an opportunity to connect students with additional needed support. Continued research on medication sources, particularly studies that examine how source patterns may change longitudinally, and how they are associated with motivations for misuse, are critical next steps in curbing PSM and its negative sequelae.

Public Health Significance Statement.

Findings from the present study showed that more than 1 in 10 college students misused prescription stimulant medication in the previous year and these students experienced more psychiatric difficulties, including suicidality, compared to a normative sample of college students. Students who misused their own stimulant medication or who procured the medication from multiple sources were especially vulnerable to these psychiatric difficulties, suggesting that these subgroups of students may benefit from comprehensive interventions that address comorbid psychiatric problems, in addition to their prescription drug misuse.