Our results illustrate that recreational drug use during the March 2020–June2021 period can be framed by 10 distinct clusters defined in a data-driven way on the basis of the combinations of drugs participants reported using (Figure 1). These distinct clusters indicate preferences for certain substances over others (eg primarily cannabis or primarily cocaine users), or a preference for an associated pattern of use (eg psychedelics and cannabis, cocaine and cannabis, MDMA and cannabis, extreme polydrug use). Most importantly, they are characterised by different levels of mood and resilience profiles during the COVID-19 pandemic.

Perhaps most strikingly, all but one drug cluster had consistently negative associations with mood, the exception being the MDMA plus cannabis cluster, where we observed positive associations with mood. Conversely, while we expected to see those who chose to use psychedelics as having better mood and higher resilience than users of other drugs or participants who never used drugs, this was not the case. In fact, we observed, contrary to our hypothesis, that those who used psychedelics indicated higher levels of anxiety, worrying about different things, irritability, apathy, depression and overall mood problems relative to individuals who never used drugs. We also found that while they reported a more relaxed lifestyle as a result of the pandemic, they disagreed that the pandemic effects could be associated with an improvement in the environment and spent less time with loved ones. Considering almost all psychedelics users in our sample were polydrug users, especially in association with cannabis, we carried out the same analysis but this time calculating effect size differences relative to the cannabis only cluster. This produced similar results; specifically, relative to those who only used cannabis, those who used psychedelics plus cannabis reported higher levels of anxiety, worrying about different things, irritability and overall mood problems; though depression levels were not significantly different.

Our findings concerning the naturalistic use of psychedelics diverge from the anticipated outcomes, but in doing so help contextualize the optimistic landscape painted by prior investigations. Prior to the COVID-19 pandemic the overwhelming majority of studies found a positive association between naturalistic psychedelics use and dimensions of mental wellbeing. For example, in a prospective naturalistic study of depressed participants, reductions in depressive symptoms were observed for up to a month after using a psychedelic (43). A cohort of Indigenous people in Canada and the United States reported the incidence of fewer depressive symptoms, anxiety and stress within a month after taking a psychedelic relative to a previous baseline (44). Subjective improvements in depression and anxiety were also reported to be associated with the naturalistic use of mescaline (26). This effect has been observed to increase as a function of higher psychedelics exposure, up to a ceiling (45). Ceremonial use of psychedelics such as ayahuasca has also been linked to improvements in depression lasting for up to 6 months (46). Interestingly, significant reductions in anxiety and depression were also identified in ayahuasca-naive participants who underwent such a ceremony (29). Using psilocybin truffles in supportive group settings has been linked to a reduction in anxiety (47, 48). Those who microdosed psilocybin also report short to medium term improvements in mood and mental health (49). It is worth noting that all of these investigations have demonstrated enhancements in various dimensions of mental wellbeing relative to participants’ own baseline, that is, in a within-subjects experimental design, instead of comparing those who used psychedelics versus those who used other drugs, or never used drugs in their lifetime. While prior findings offer valuable insights into the potential of psychedelics to ameliorate low mood following experiences in naturalistic settings, it remains unclear whether the magnitude of these improvements was substantial enough to match (or go beyond) the mood levels of non-drug users or users of other drugs, thus potentially resulting in psychedelics users as having better mood than other subsets of the population.

Findings concerning naturalistic psychedelics use during COVID-19 pandemic offer a more particular basis for drawing direct comparisons with our results. A few studies did attempt to explore the relationship between use of psychedelics and mental health outcomes during the COVID-19 pandemic and reported that people did indeed use psychedelics with the intention of better coping with pandemic stresses (50, 51) and that lifetime use of psychedelics was associated with better mental health indicators (52). Psychedelics have even been proposed for the purpose of treating mental health conditions due to or aggravated by COVID-19 viral infection (53). (50) indicated two thirds of their sample claimed that pandemic use of psychedelics helped them deal with the global situation better. (51) reported those who used psychedelics (and also MDMA) used problem-focused coping strategies in response to the global crisis more often than non-users and (54) (on a different cross-sectional analysis of data from the same study as (51)) reported that users of psychedelics, especially regular ones, reported less psychological distress than non-users. However, a survey carried out in the United States of America (USA) found an association between psychedelics use, namely between past use of psilocybin mushrooms, and worse mood self-assessment scores at the time of assessment (55). Our findings agree with those of Matzopoulos et al. (55), who employed similar mood self-assessment scales as ours to survey the USA general population, but differ to those of (53) who used different mental wellbeing assessments and carried out the survey mainly in Spain and Brazil through snowball sampling on social media.

The obvious question prompted by our results is - why is it that we are not seeing, in the light of the positive effects demonstrated by clinical studies on psychedelics as well as naturalistic surveys, a positive association between naturalistic psychedelics use and mental health during times of crisis? Some of the earliest theories aimed at explaining the variability in the effects of psychedelics have converged around the concept of set and setting (56), and their potential for acting as non-specific context amplifiers (57). Set and setting refer to the mindset and intention of the individual experiencing the psychedelic, and the characteristics of the environment where the experience is taking place, respectively, (58, 59). In modern clinical studies, for example, in order to ensure an optimal “set,” careful attention is given to provide participants with psychological support before, during and after the experience, and to ensure an optimal “setting” the same attention is paid to the environment in which the psychedelic intervention takes place (60). Given the importance that this notion of set and setting is given in the clinical environment, it is potentially unsurprising that the benefits seen there are not transferred to the naturalistic setting where these variables may not be controlled optimally. Not only was the context of the experiences captured in our study not explicitly therapeutic, but it was also heavily marked by a global mental health crisis as well as significant disruptions in the immediate environment of individuals as a result of the COVID-19 pandemic. As demonstrated by our findings, lifestyle disruptions were common in the majority of drug use clusters, as was spending less time with loved ones, and in certain instances an increase in conflict at home. These factors undoubtedly had imminent effects on the set and setting of psychedelic experiences during that period, and it is reasonable to infer that the disruptive context would have likely contributed to the associations we observe between use of psychedelics and mood self-assessment scores.

Another major difference between the present study and the previous literature is the socio-cultural context in which the psychedelics have been taken. Different countries will have different cultural acceptance, cultural significance and general stigma around psychedelic drug use to that of the UK. Adding to this, local regulation due to the pandemic may have produced differences in access to drugs. Since context can influence psychedelic experiences in a naturalistic setting (61), it cannot be ruled out that these factors contributed to the outcomes we report. More specifically, it is not excluded that psychedelics use in the UK (where strict lockdown and infection control guidelines were employed) during the COVID-19 pandemic could have led to the amplification of general distress, which would have in turn influenced the outcome of psychedelic experiences. It is worth noting, in support of this perspective, that our findings align with those emerging from the USA during a similar timeframe (54), but not to findings from Spain or Brazil (53). While political climate, healthcare access and infection control guidelines differed in these countries and undoubtedly affected their population in differential ways, the specific socio-cultural nuances related to the use of psychedelics also need to be considered. Neither the USA or the UK have a recent history of general cultural acceptance of psychedelics use, nor were these substances legal for recreational use at the time of assessment. Given the above, there is a question as to what extent our findings generalize outside the borders of the UK or the USA.

Despite the issues with naturalistic psychedelic drug use studies with regards to uncontrolled set and setting, the majority of these studies (both the within-subjects design pre-pandemic studies as well as the cross-sectional study of (53) carried out during the pandemic) still report a positive association with mental wellbeing. Another key difference between our study and most of the other studies mentioned above that could explain the reported outcomes is the nature of participant recruitment methodology. With few exceptions, the above mentioned studies advertised psychedelics-related research in psychedelic-profiled social media groups. Members of these groups who respond to such advertisements may not be representative of those groups or of the people who use drugs more generally. Recruitment bias has been previously called out to be a confound in surveys specifically recruiting psychedelics users, since openly advertising such studies opens the door for self-selecting participants who, on the basis of enthusiasm, positive experiences, and/or desire to contribute to research to advance a global societal movement, might not yield objective datasets and consequently allow scientists to draw the right conclusions (62–65). In particular, biases in sampling related to positive past experiences, but also as pertains to a prospective study-related expectation that there may be a mental health benefit associated with psychedelic use, could confound the results.

There are other reasons why we might observe different outcomes. None of the past studies looking at naturalistic use of psychedelics used a data-driven approach to cluster the use of psychedelics with the use of other drugs alongside to expose effects of common drug interactions. Additionally, none directly compared the data of psychedelics users from clusters derived in this way with data from users of other drugs who do not use psychedelics. Notwithstanding the debate on what constitutes a psychedelic to begin with (66), in past studies classical and non-classical psychedelics were often assessed concomitantly under the umbrella term of ‘psychedelics’, regardless of the use of other drugs. This is an important consideration as drugs with different pharmacology and subjective effects could produce varying outcomes. Our own data, for example, reveals that within the cluster of individuals who primarily used psychedelics and cannabis during the pandemic a small proportion have also used MDMA. (53) also included MDMA users as part of their ‘psychedelics users’ group in a cross-sectional analysis of users vs. non users of psychedelics. Notably though, MDMA users were more prevalent in their sample compared to ours. The implications of this observation are significant, particularly considering that our results demonstrate a contrasting association between the use of primarily MDMA and cannabis (in the absence of psychedelics) and dimensions of mental wellbeing. Specifically, individuals within this cluster exhibited better mood relative to those who never used drugs in their lifetime at the time of assessment, drawing further attention to the question pertaining to whether the proportion of MDMA experiences captured within the ‘psychedelics’ label is what could be driving the differential outcomes.

Our analysis focuses on modeling the choices of drugs used specifically during the pandemic timeframe, rather than the frequency or underlying motivations for these choices. It is not excluded that using psychedelics in conjunction with other substances (e.g., as was evident in our sample that the majority of those who chose to use psychedelics were polydrug users and also chose to use cannabis, whereas a minority used other substances too) during a global mental health crisis cancels out therapeutic effects that could potentially be derived from naturalistic experiences with psychedelics. The effects on mental wellbeing would be contingent on prior experiences, the frequency of use of each individual substance, on their dosage, whether they were consumed together or separately and under which circumstances (as discussed above). Within our data-driven clusters, we anticipate capturing a diverse range of such patterns. For instance, some individuals may use cannabis daily while only occasionally using psychedelics, whereas others may frequently engage in microdosing psychedelics while rarely using cannabis. These patterns might have been in place prior to the pandemic or adopted because of it. Furthermore, the dosage of these substances is expected to have varied among the individuals surveyed, in turn leading to varying acute and long-term effects mediated by different degrees neurobiological changes and the resulting intensity of those experiences. Additionally, it is reasonable to assume that the individuals we surveyed would have used psychedelics or cannabis for a variety of reasons not limited to self-medication, recreationally, as a social catalyst, or for spiritual purposes. These motivations would have differentially influenced their mood post-experience. Moreover, whether therapeutic effects even exist subsequent to naturalistic use, they might be short lived [maximum documented has been 6 months post experience by Ruffell et al. (46)] and our participants might have had their experiences months apart from answering our survey, and/or modulated by subsequent use of other substances (or even prescription medications) not limited to the ones captured in our assessment.

A core difficulty in interpreting our findings is to do with potential causal links between psychedelics use, mood self-assessment and resilience metrics in the general population. Contextual variability affecting the outcome of drug-induced experiences aside, it might also be that people with poor mental health to begin with take psychedelics, and therefore their perceived improvements in mood might be visible only in a within-subjects study design rather than something comparable across large segments of non-drug using population. It has been previously documented that people with poor mental health indeed engage in drug use to self-medicate (67), and it is possible that individuals experiencing low mood during the pandemic turned to psychedelics for this reason. Drawing upon prior research findings, psychedelics could have indeed contributed to improvements in the mood of our participants, suggesting that our results could be driven by baseline differences in mood that preceded the experiences rather than a lack of benefits derived from psychedelics use. However, this observation raises the possibility that the improvements in mood resulting from psychedelic experiences in naturalistic settings may not reach a level that equals or surpasses the mood levels of individuals with no history of drug use. This could be the case particularly during times of global crisis but also beyond. On the other hand, it is also plausible that our results are influenced by fundamental differences in dimensions other than mood between individuals who choose to use psychedelics and those who use other drugs or have no history of drug use, which were not captured in our assessment –genetics, brain and body health status, co-morbidities and psychological history to name a few. To add to the complexity, there might be differences between people who chose to use psychedelics during the pandemic versus people who chose to use psychedelics before the pandemic, or those who will choose to use psychedelics after the pandemic. These underlying differences could potentially contribute to baseline variations in mental health levels that persist regardless of any potential positive effects on mood resulting from psychedelic experiences in naturalistic settings. It is also possible that a combination of these factors – limited magnitude of psychedelic effects on mood as well as fundamental differences between user groups – might be at play in determining cross-sectional differences. Importantly, the severity of the pandemic impact highlights that it is crucial to consider these factors when evaluating survey data on psychedelic use collected as the world recovers from disruption.

Our results reflecting a positive association between MDMA/ecstasy and cannabis use are interesting and surprising, though we would like to exert caution interpreting them and their generalisability to the wider spectrum of MDMA users. The within group variation was high and the sample size of this cluster low (N = 90) relative to the other clusters, consistent with previous work reporting a decrease in typical ‘party’ drugs such as MDMA/ecstasy during the COVID-19 pandemic (68), thus the effects we report reflect a bias toward more extreme values in this particular case. Additionally, as this is a self-report sample we cannot exclude that some mood self-assessments might be inaccurate either due to dishonesty or as a function of personal awareness, and in lower samples this raises higher levels of skepticism in interpretation. However, if these results were to be replicated in a larger sample, a possible interpretation would be that since acute MDMA effects are associated with feelings of bonding, love and social connection (69), MDMA-catalyzed social interactions during the pandemic could have acted as a protective mechanism on the mood of users, relative to those who did not have these experiences.

Our study has a number of strengths which confers us the ability to draw the present inferences with confidence. First and foremost, we present one of the few studies assessing effects of naturalistic use of psychedelics, MDMA and other drugs in the absence of recruitment bias toward social media groups discussing psychedelics, since the questions which participants were asked were not advertised either at the time of recruitment or prior to the follow-ups, and our study was never advertised on social media channels pertaining to drug use/related activities. What’s more, in our analysis we employ a diverse population inclusive of a very large control population who has never used drugs but has answered the same questions during the same timeframe. We also acknowledge that drug use behavior is dynamic, and that psychedelics are most often not used in isolation, with cannabis specifically being the substance of choice that participants might use during the same period of time, thus making it generally difficult to disentangle specific drug effects when these are being used naturalistically.

There are also certain limitations our work possesses, in part owed to the main reason our study is unique - that the study design was not drug-use specific. Recreational drug use is a prevalent lifestyle decision that covers choices of both licit and illicit drugs. In our study we only model choices of recreational drugs that are illicit in the UK, and control for the frequency of otherwise licit drugs alcohol and tobacco in the multivariate analysis. This distinction in our analysis is not to imply that licit drugs are less harmful than illicit drugs. While characterization of harm based on this dichotomy is beyond the scope of this manuscript, we note the extensive evidence to suggest that alcohol and tobacco use is at least as harmful as is the use of the other recreational drugs analysed in this study (70–77). Furthermore, due to a variety of reasons not limited to stigma, it is difficult to recruit drug users from the general population, since this category of individuals, particularly if they have a high level of polydrug use, are less likely to engage with traditional survey methods (78). Therefore, we might have undersampled the population with problematic use (as reflected by the extreme poly-drug users cluster having the lowest number (N = 35) of respondents). With regards to polydrug use, at the point when the present data were collected we have not collected extensive details about participants (who declared having used a drug at least once) drug use history prior to the COVID-19 pandemic, thus diminishing our ability to draw inferences about an accumulation of experiences throughout their lifetime potentially influencing mood and/or resilience during the pandemic timeframe. We expect that within our data-driven clusters people would have had different frequencies of specific drug use prior as well as during the pandemic. However, we have not collected extensive details about participants drug use frequencies during the pandemic timeframe itself, and are therefore not able to assess the relationship between frequency of use and effects on mental wellbeing in the present study. These dimensions of mental wellbeing, too, were studied independent of clinical diagnoses of either psychiatric or neurological disorders, and given the large dataset it is not excluded that some participants might have had clinical levels of anxiety, depression, post-traumatic stress disorder or even diagnosed substance use problems. Other factors such as living conditions, quality of interpersonal relationships, and various ramifications within the umbrella term of ‘social misery’ have not been addressed in the present analysis, and it is not excluded that there would potentially be significant differences in these aspects between the different drug use clusters, with effects on mood and resilience. Lastly, we highlight that our sample, although large, is >90% represented by members of the British population, and therefore it is possible that associations between naturalistic use of psychedelics (as well as other recreational drugs) might look different for individuals residing in other parts of the world.

The effects of naturalistic use of recreational drugs on mood, and mental wellbeing more generally, warrant further attention and research since these substances are growing increasingly more popular and available beyond clinical setups. In particular, we advise for other studies to try to replicate our findings in existent/future large datasets that also collected data on naturalistic drug use. Concerning our results on MDMA and cannabis use, this is particularly important, since unintended “hype” might arise as a result of the noted positive associations, which if it results in increased naturalistic use to self-medicate for mood disturbances is of concern due to potential toxicity of prolonged MDMA exposure (79). Concerning our results related to psychedelics use, the effect of different contexts and more granular investigations of how particular set and setting features influence the quality of the experience and its long term outcomes might prove useful in advancing our understanding of psychedelics effects in naturalistic settings and inform the development of harm reduction guidelines. We note special attention might be given to cultural contexts, as perceptions of psychedelics are tied to ethnographic backgrounds (80). Amongst key survey/experimental design approaches we highlight the importance of selecting participants as agnostically as possible in relation to the study outcome, and where possible not disclosing the study hypothesis. Aiming to recruit control-participants and even users of other drugs, too, and test them on objective mental health metrics alongside psychedelics users would also strengthen and contextualize conclusions. Perhaps even more importantly, collecting non-psychedelic drug use history, lifestyle and personality data in such studies and then adequately accounting for these variables at the analysis stage might yield more holistic insights as to what exactly psychedelics modulate and in relation to/independent of what other factors. The underlying effects on brain chemistry and activity where drug interactions of potent pharmacological agents are used in conjunction with one another is also something that might be worth exploring in association with behavioural metrics. Specifically, attention ought to be given to the intersection of different patterns of using distinct drugs in parallel, both licit and illicit.