Little has been done to explore the potential risks of substance use disorder (SUD) associated with psychedelic use among those with psychiatric illness in a community setting. Psychedelic use has more than doubled globally from 2015 to 2020 based on the Global Drug Survey (Winstock et al., 2021), where diverse classes of psychedelic drugs are now used to manage psychiatric conditions or emotional distress. Individuals with psychological illness are especially likely to be diagnosed with an SUD and require special attention when considering risks and benefits of pharmaceutical treatments (Buckley, 2006). Analyzed separately, psychiatric illness and problematic psychedelic use have each been shown to be associated with SUD. Both internalizing and externalizing psychiatric illnesses are epidemiologically associated with comorbid SUD (Kroll et al., 2020), while individuals in the general population with markers for psychedelic use disorder are at the greatest risk of having other comorbid SUD markers (Von Gunten & Wu, 2021). Studies have not combined analysis of psychedelic use with psychiatric illness, and it is unclear whether these cumulatively would result in more severe SUD or not.

Serotonergic and entactogenic psychedelics are increasingly studied for regulatory approval as medications. While clinical trial results in the United States (US) and United Kingdom suggest that psychedelic use in a clinical setting is largely safe and well-tolerated (Andersen et al., 2021; Belouin et al., 2022; Palhano-Fontes et al., 2019), the real-world abuse potential of any medicinal psychedelics will likely be assessed separately for each drug by US regulators (Calderon et al., 2018). For available and widely used substances, real-world evidence, survey data, and public health factors are more heavily weighted by the Food and Drug Administration (FDA) than when assessing new substances, which often rely more on clinical evidence (Henningfield et al., 2023). Research into how different types of psychedelics may influence abuse in a real-world setting will inform regulations and requirements (e.g., risk evaluation and mitigation strategies) for approved medications (Henningfield et al., 2022).

Epidemiologic, real-world data systems offer the best opportunity to study problematic drug use behavior, because it often occurs outside the clinical setting. While clinical trial information is critical, it can be limited by a lack of generalizability and narrow case definition (Lowe et al., 2022; Munafo et al., 2022) often excluding SUD comorbidities (Mitchell et al., 2021; Rucker et al., 2021), and making it impossible to study problematic drug behavior as an outcome. Most adults with psychiatric illness do not receive mental health services (Substance Abuse and Mental Health Services Administration, 2022), leading to potential selection bias for studies relying on clinical data. Further, psychedelics are shown to behave differently depending on the context in which they are used, highlighting the public health need for studying psychedelic use in a community setting to complement trial settings (Barnett, 2022).

In this study, we provide real-world evidence to assess whether psychiatric illness combined with psychedelic use was associated with more severe SUD markers in the general adult US population. We utilized a nationally representative population survey, which includes individuals with diverse drug use histories, in order to characterize potential determinants of SUD incidence and severity if psychedelic medications are made more widely available (Rose, 2001). The goals were to (1) quantify the associations between psychiatric illnesses, psychedelic use, and markers predictive of SUD, and (2) test differences in SUD marker severity when psychiatric illness and drug use were combined. Because psychiatric illness is a risk factor for SUD and potentially associated with psychedelic use, we hypothesized there would be a difference in SUD marker severity when psychiatric illness was combined with nonpharmaceutical psychedelic use, but marker severity differences would be smaller than when psychiatric illness was combined with nonpharmaceutical stimulant or opioid use. We examined four psychiatric illnesses with phase 2 or 3 clinical trials completed or ongoing involving psychedelic-assisted psychotherapy: anxiety disorder (Griffiths et al., 2016), major depressive disorder (MDD) (Goodwin et al., 2022; Griffiths et al., 2016; Gukasyan et al., 2022; Rucker et al., 2021), post-traumatic stress disorder (PTSD) (Mitchell et al., 2021), and bipolar disorder (Psilocybin Therapy for Depression in Bipolar II Disorder (BAP), 2023). Primarily serotonergic psychedelics are non-reinforcing, while others, such as MDMA, are weak/moderately reinforcing (Henningfield et al., 2023). Therefore, we also differentiated effects from primarily serotonergic psychedelics (lysergic acid diethylamide [LSD], mescaline, and psilocybin) and an entactogenic psycedelic (3,4-methylenedioxymethamphetamine [MDMA]).

Methods

Data Source

The Researched Abuse, Diversion, and Addiction Related Surveillance (RADARS) System deploys an online, serial, cross-sectional drug survey of the general adult population in the US, the Survey of Non-Medical Use of Prescription Drugs (NMURx) Program, which has been shown to be valid (Black et al., 2019) against three national benchmark probability samples and reliable in a test-retest study (Black et al., 2021). Briefly, a national sample was selected from an online panel and multiple best practices implemented to control for bias within this nonprobability sampling design. Selection bias and measurement bias were proactively mitigated by calibration weighting (Deville et al., 1993) and careless response exclusions (Meade & Craig, 2012). Calibration weights included demographic variables (age, binary sex, state, race, and Hispanic ethnicity) and health domain variables (limitation in daily activity and smoking history). Domain-specific weighting variables are necessary to improve accuracy of domain-specific estimates because useful adjustment variables must be correlated to the desired estimates (Mercer et al., 2017). Careless response exclusions substantially reduce bias for low prevalence items, such as nonpharmaceutical drug use. Questions are randomized to reduce order effect bias and skip logic employed to reduce respondent burden. The NMURx Program survey is deployed nationally, with respondents from nearly all 3-digit zip codes reporting, mitigating coverage bias (Black et al., 2019). Social desirability bias was addressed by using a confidential online format (Ong & Weiss, 2000). A target sample size of 30,000 per wave was determined so that prevalence estimates would have confidence intervals of approximately 1–2 percentage points. Data were collected during 2022 in two waves from 15 April to 03 June and 09 September to 21 October (Fig. 1). Median survey completion time was 12 min, 40 s. The NMURx Program study protocol is approved by the Colorado Multiple Institutional Review Board, with the most recent certificate of exemption granted on 17 January 2019.

Fig. 1
figure 1

Survey recruitment diagram. The survey was deployed in two waves in 2022

Full size image

Exposure Definitions and Outcome Measure

Participant demographics, history of psychiatric illness, past 12-month nonpharmaceutical drug use, and the Drug Abuse Screening Test (DAST-10) are self-reported on this survey. Respondents were asked to self-report if they have had a diagnosis of psychiatric illness in their lifetime and to identify the illness. Exposure to psychedelics was defined for two psychedelic groups, as respondents reporting past 12-month use of (1) serotonergic psychedelics including LSD, mescaline, or psilocybin, and (2) MDMA. Past 12-month use of nonpharmaceutical opioids (heroin, fentanyl, or fentanyl analogs) or stimulants (cocaine, crack, or methamphetamine) were included as comparator drug classes. The DAST-10 is a validated, continuous instrument for problematic drug use behaviors with scores ranging from 0 to 10 with excellent psychometrics (Stewart et al., 2023) and was selected as the outcome variable. DAST-10 is a suitable proxy for SUD, where scores greater than 2 have high sensitivity and specificity for DSM-IV SUD clinical diagnosis (Yudko et al., 2007). Questions in the DAST-10 are not drug specific (behaviors related to alcohol are excluded), and therefore the outcome represents overall severity of drug use behavior, rather than specific abuse of a single drug class. Three example problematic behaviors the DAST-10 asks about are (1) using multiple drugs at once, (2) familial neglect due to drug use, or (3) medical problems like memory loss or convulsions. Because the questions are agnostic to drug class, the DAST-10 score is well suited when comparing differences in problematic drug use behaviors across drug classes. Race and ethnicity information was self-identified. Hispanic, White, Black, and Asian are reported separately; American Indian, Alaska Native, Native Hawaiian, Pacific Islander, multiple races, or another race were combined due to infrequent reporting of these categories.

Analytic Plan

Estimates of descriptive characteristics with 95% confidence intervals (CIs) were calculated using weighted values to generate nationally representative estimates. A weighted linear regression was used to test whether adults with a history of diagnosed psychiatric illness who also used either of the psychedelic drug classes had higher mean DAST-10 scores compared to adults with a history of diagnosed psychiatric illness but who did not use psychedelics. Because multiple drug use was anticipated, all drugs and statistical interactions were included in a single, saturated model. Therefore, estimates of differences in DAST-10 score by drug class were tested while controlling for use of other drug class covariates (including controlling for use of multiple types of psychedelics). Only main statistical parameters and 2-way interaction terms are interpreted; higher order terms were included only for statistical control of multiple drug use. Additional statistical details are provided in the supplemental material. Models for PTSD, anxiety disorder, MDD, and bipolar disorder were examined separately. All CIs were estimated using 250 bootstrap replicates. All analyses were conducted using SAS 9.4 (Cary, NC, USA) and p-values less than 0.05 were considered significant.

Results

Adults with Psychiatric Illness and Past-Year Psychedelic Use

A total of 59,041 respondents were included, which estimated a 257 million adults in the US population in 2022 (Table 1). The national estimate of past 12-month use of either psychedelic group was 2.6% (95% CI: 2.5, 2.8). National prevalence estimates of history of diagnosed psychiatric illness were anxiety disorder 22.7% (22.3, 23.2), MDD 8.8% (8.5, 9.1), PTSD 6.0% (5.8, 6.2), and bipolar disorder 4.7% (4.5, 4.9). Adults with history of diagnosed psychiatric illness were younger and more likely to be females, identify as White, report being a healthcare worker, and have lower self-assessed general health. Adults with history of psychiatric illness were more likely to be limited in their daily work, upwards of twice as likely for those with history of MDD or PTSD. Adults with history of anxiety disorder, MDD, and bipolar disorder were less likely to be veterans; those with PTSD were more likely. Nonpharmaceutical drug use of any drug class was more common among adults with history of psychiatric illness. There was little variation in DAST-10 scores between states. The median state-aggregate score across the 50 states and District of Columbia was 0.64 (IQR: 0.11).

Table 1 Demographics and drug use for adults with history of psychiatric illness
Full size table

Psychedelic use of either class was two to four times as prevalent among adults with a history of psychiatric illnesses. Among these four disorders, the rate of serotonergic psychedelic use was highest for bipolar disorder, where an estimated 7.4% (6.3, 8.5) has used a serotonergic psychedelic in the past year. The lowest prevalence was using MDMA with anxiety disorder (1.6% [1.3, 1.9]). Prevalence of serotonergic psychedelic use among subpopulations were higher than using nonpharmaceutical opioids, but generally lower than using nonpharmaceutical stimulants. Prevalence of MDMA use was similar to use of nonpharmaceutical opioids. The DAST-10 profiles for adults with psychiatric illness using serotonergic psychedelics were less severe (i.e., higher proportions with scores 0–2) when compared to adults with similar histories who used nonpharmaceutical opioids or stimulants (Fig. 2).

Fig. 2
figure 2

Problematic drug use markers among adults with history of psychiatric illness. DAST-10, Drug Abuse Screening Test, 10 question

Full size image

Summary of Statistical Model Parameters

The statistical models included covariates for multiple drug use and demographic covariates and were repeated for each of the four psychiatric illnesses (all modeling parameters provided in supplemental material). Increasing age and being female were significantly negatively associated with DAST-10 score. Hispanic ethnicity was significantly positively associated; Asian, Black, and White race indicators were not significantly associated. Prescription opioid and stimulant use were both significantly associated with increasing scores. Region was not significantly associated. Effect sizes for demographic covariates were small (typically with magnitudes of less than a half a DAST-10 point).

Associations of drug use with DAST-10 were positive and significant for all of serotonergic psychedelics, MDMA, opioids, and stimulants, indicating that when controlling for multiple drug use, use of each drug class is independently associated with an increase in DAST-10 score. A positive association was also observed for each psychiatric illness with DAST-10, suggesting higher risk of problematic drug use with each of the four disorders, independent of drug use. Interaction terms, which test whether differences in DAST-10 had larger magnitudes when drug use was combined with psychiatric illness, were significant only for stimulant or opioid use, but not for serotonergic psychedelic use. Significant interactions indicate that DAST-10 scores statistically increased more when psychiatric illness was combined with stimulant or opioid use than either drug use or psychiatric illness alone would account for. Interactions with serotonergic psychedelic use was not significant for any psychiatric illness, indicating there was no additional significant association observed with the combination of psychiatric illness and serotonergic psychedelic use, unlike with stimulants or opioids. Notably, the interaction terms for MDMA use with BPD and MDD were significant interactions, indicating this combination does associate with additional increases in DAST-10 scores. Finally, there were significant, negative interactions between serotonergic psychedelic use and use of nonpharmaceutical opioids and stimulants. This suggests a resilience association, where DAST-10 scores were smaller in adults who use both.

Drug Class Use Associations with Problematic Drug Use Markers

The mean DAST-10 scores represent severity of drug use consequences for individual subpopulations when using each drug class (Fig. 3). The mean DAST-10 scores were low for adults with a history of diagnosed psychiatric illness but without any drug use, low-to-moderate for psychiatric illness subpopulations using only serotonergic psychedelics or MDMA, and substantially higher for psychiatric illness subpopulations using only either nonpharmaceutical opioids or only stimulants. When not using any drug class, DAST-10 scores differed by less than half a point for adults with a history of psychiatric illness vs those without. While confidence intervals did not overlap, these differences are likely not clinically meaningful. Among psychiatric illness subpopulations using either of the psychedelic classes, estimated mean DAST-10 scores were less than 3, suggesting typical behavior among those only using either serotonergic psychedelics or MDMA would not qualify for SUD under DSM-IV criteria. Comparatively, psychiatric illness subpopulations using only either nonpharmaceutical opioids or stimulants had mean scores above 3, suggesting SUD would typically be present for these individuals. Notably, history of PTSD combined with only opioid use resulted in the highest mean DAST-10 score.

Fig. 3
figure 3

Least-square mean DAST-10 scores by history of psychiatric illness and drug use

Full size image

Finally, we tested for an association of increased DAST-10 scores with use of the psychedelic classes within only adults with psychiatric illness histories (Fig. 4). There were statistically higher mean DAST-10 scores for both serotonergic and MDMA use. Larger differences were observed with nonpharmaceutical opioid or stimulant use. Because the statistical model included all prescription and nonpharmaceutical drug classes together, multiple drug use by individuals was controlled when least square mean differences were estimated. Mean difference in DAST-10 for each drug class is interpreted as the difference across groups with average values for multiple drug use and demographic covariates subtracted out. Estimated score increases associated with each psychedelic class were near 1 for all psychiatric illness subpopulations, a modest increase in SUD severity. None were negative, which suggests there was no independent resilience effect of psychedelic use against higher DAST-10 scores. Estimated differences for opioid and stimulant use were positive and generally similar to each other. Differences did not vary substantially across psychiatric illnesses.

Fig. 4
figure 4

Adjusted differences in DAST-10 scores among adults with history of psychiatric illness

Full size image

Discussion

Our results indicate that adults with a self-reported history of anxiety disorder, MDD, PTSD, or bipolar disorder had higher rates of nonpharmaceutical psychedelic use compared to the general adult population in the US. The NMURx Program national estimate of past-year psychedelic use of either class was consistent with the National Survey on Drug Use and Health estimate of hallucinogen use, confirming concurrent validity of the national results (Substance Abuse and Mental Health Services Administration, 2022). Prior literature suggests psychedelic drug use may place individuals with mental health challenges at risk of both physiologic and psychological harms, and harms have not been consistently observed across all substances. Physiologic harms have been observed in a variety of substances, including seizures using serotonergic psychedelics (Simonsson et al., 2022), psychosis from psilocybin (Vollenweider et al., 1998), tachycardia using LSD or psilocybin (Leonard et al., 2018), and serotonin toxicity if serotonergic psychedelics are combined with other serotonin regulating drugs (Malcolm & Thomas, 2022). Psychologically, individuals may experience dysphoria when using serotonergic psychedelics (Bender & Hellerstein, 2022), worsening BPD disorder symptoms from psilocybin (Morton et al., 2023), increase in anxiety with MDMA (Breeksema et al., 2022), impairment due to distressing experiences with serotonergic psychedelics (Simonsson et al., 2023), or even suicidality in rare cases with psilocybin (Johnson et al., 2018). Incidence of these events may be low, but severity can vary widely. Regulatory bodies should ensure licensed, trained providers of guided experiences are prepared for these outcomes.

Consistent with our hypothesis, increased severity of problematic drug use behavior among these subpopulations was associated with serotonergic psychedelic and MDMA use, and severity differences were smaller than was observed for nonpharmaceutical opioids or stimulants. There was a significant statistical interaction between use of either nonpharmaceutical opioids or stimulants with psychiatric illness, but the same does not appear to be true with either psychedelic class. This suggests that psychiatric illness combined with opioid or stimulant use cumulatively exacerbates SUD severity, while combinations with either psychedelic class did not. Several hypotheses may explain this difference. This observation may be due to lower physiologic addictive properties compared to other drugs with higher known abuse potential (Nichols, 2016). Lower scores may represent a more functional group of people using psychedelics to cope with life stressors or may reflect psychedelic use improving function for those struggling with psychiatric illness. Many who report past year serotonergic psychedelic use may be microdosing, and individuals who microdose LSD or psilocybin are less likely to have SUD than those using larger doses (Rosenbaum et al., 2020). The lack of microdosing information as a covariate likely obscured meaningful differences in DAST-10 scores between different dosage schedules. Additional confounders may also be contributory and should be examined in future work.

Surveillance monitoring of drug use consequences in the broader community is challenging. It should control for the complex interplay of illness and use of multiple drugs for results to be actionable by the public health community. The high order model presented here allowed for an interpretable measure of the potentially complex interaction of non-pharmaceutical drug use and psychological illness in the general adult population. The DAST-10 measures behavioral markers that are not drug specific, meaning results here are interpreted as the broader behavioral consequences associated with drug use. This differs from past research that focused on the individual abuse potential (Bender & Hellerstein, 2022). Separately, both classes of psychedelic use and history of psychiatric illness were each associated with elevated DAST-10 scores in a real-world setting, supporting prior research showing these are risk factors for SUD (Kroll et al., 2020; Von Gunten & Wu, 2021). Additionally, the interaction of psychedelic use with nonpharmaceutical opioid use resulted in significantly lower DAST-10 scores, consistent with serotonergic psychedelics being associated with lower odds of opioid use disorder (Jones et al., 2022). Understanding drug-specific real-world interactions would apply to all harms, including SUD, and is vital to assessing the feasibility, safety, and proper diagnostic indication of potential treatments involving psychedelics. This type of information cannot be derived from clinical data alone.

Providers of psychedelic-assisted psychotherapy should be cognizant and screen for escalating generalized SUD symptoms throughout the course of treatment. However, these data do suggest that providers may not need to take additional precautions with patients using serotonergic psychedelics, and instead may rely on traditional screening for SUD (Stewart et al., 2023). For providers not utilizing psychedelic-assisted therapy, SUD remains a concern, given our results show elevated rates of nonpharmaceutical psychedelic use among adults with histories of psychiatric illness. It is unknown whether federal approval of pharmaceutical versions of psychedelic drugs for medical use would cause a concurrent increase in nonpharmaceutical use in the general population. Both providers and public health officials should be prepared for this possibility.

Limitations

The results here did not examine all potential risks from introducing a new psychoactive chemical treatment. Among other medicinal and public health consequences already outlined (Johnson et al., 2018), there is likely a subpopulation who could attempt to self-treat, and consequences of these actions remain uncharacterized. In addition, psychiatric illnesses cannot be treated as homogeneous group, and we studied only four disorders. A limitation is that only the respondents’ self-report of diagnosed psychiatric illness was captured on the survey, and current mental health burden was not assessed. However, these results do represent associations among adults who have psychiatric illness etiologies. The temporal order of events could not be discerned, such as whether psychiatric illness symptomatology preceded the initiation of psychedelic use. Finally, while core demographics and multiple drug use was controlled, other confounding effects were not. There could be a difference in DAST-10 score based on whether a person was using psychedelics for recreation or as a self-treatment option; if these effects oppose each other, it would confound the estimate of least square mean differences toward zero. Other factors, such as dosing schedule, frequency of use, or setting of use, were uncharacterized.

Conclusion

In summary, adults with a history of psychiatric illness had a higher rates of serotonergic psychedelic and MDMA use compared to the general population, and more severe problematic drug use markers in these subpopulations were observed that associated with use. If approval for psychedelic-assisted therapy is granted, it is unclear how nonpharmaceutical psychedelic use might change in the general population. Harms associated with psychedelic use, including increased risk for SUD, should be proactively managed, and a unique approach to managing these risks that differs from past approaches for other drugs of abuse should be developed. Longitudinal studies that address potential confounding and severity of symptomatology are necessary to determine the durability of these results.