The COVID-19 pandemic significantly disrupted daily life, profoundly affecting mental health (Hampshire et al., 2021a). Although overall rates of illicit drug use during the pandemic reduced, concurrently, there was a rise in their use as a coping mechanism (Bălăeţ et al., 2025; Rogers et al., 2020). However, longitudinal studies on the mental health of people who used illicit drugs spanning from before to after the pandemic restrictions are limited. In the UK, the need for such research is underscored by statistics indicating that 1 in 11 individuals aged 16–59 used drugs other than alcohol or tobacco in the past year (Office for National Statistics, 2023).

Of the few studies that specifically examined the relationship of substance use with mental health during the pandemic, most have focused on alcohol, tobacco or cannabis, often grouping other psychoactive substances and overlooking their unique effects (Zolopa et al., 2022). However, recent efforts have targeted the psychedelics and entactogens associations with mental health outcomes such as mood and resilience, distinct from other drugs (Kopra et al., 2023). This is timely given the increased interest in the potential use of psychedelics for therapeutic purposes (Carhart-Harris et al., 2021) and their use among the general population to try and self-manage well-being, including those without severe mental health issues (Keyes and Patrick, 2023; Livne et al., 2022). This interest is driven by the therapeutic potential of psychedelics like psilocybin for severe depression (Carhart-Harris et al., 2021), as well as research suggesting mood-enhancing effects in ‘naturalistic settings’, that is, as they are commonly used in the real world (Forstmann et al., 2020).

Research on the relationship between psychedelics use during the COVID-19 pandemic and mental health has produced mixed findings. Some studies, like those by Livne et al. (2022) and Keyes and Patrick (2023), cautioned against potential harms from increasing use, but have received criticisms based on reports of positive associations between psychedelics use during the pandemic with well-being (Morgan, 2023). The evidence for these positive associations comes from social media-based studies (Bouso et al., 2023; Cavanna et al., 2021; Révész et al., 2021), but contrasts with findings from large-scale surveys in the United Kingdom and the United States (Bălăeț et al., 2023b; Matzopoulos et al., 2021). These larger surveys indicate negative mental health correlations with psychedelics but a positive association for 3,4-methylenedioxymethamphetamine (MDMA; Bălăeț et al., 2023b), suggesting that assessing MDMA and psychedelics together, often employed by historic surveys, may contribute to inconsistent results. Discrepancies might also stem from recruitment bias, methodological differences and cultural factors. This highlights a need for more detailed longitudinal analyses to better understand mental health dynamics among psychedelic users, non-users, and users of other illicit drugs in the context of real-world stressors like the pandemic.

Studying drug use in the general population presents several challenges, particularly due to the prevalence of polydrug use; specifically, individuals often use multiple substances across time and may consume them simultaneously (Bălăeț et al., 2023b; de Jonge et al., 2022). This complicates efforts to isolate and model the effects of individual drugs, as it diverges from common patterns of drug use in the general population. Furthermore, drug use behaviour is dynamic, changing over time, adding another layer of complexity. Addressing these issues requires large-scale population sampling, which gathers data from real-world drug users in the context of environmental stressors and accounts for the varied and evolving drug use patterns observed in everyday life.

Here, we aim to advance our understanding of the relationship between mental health symptoms and common patterns of illicit drug use in the UK general population during crisis times. Specifically, we analyse online survey data collected from a large citizen science cohort spanning the timeframe from just prior to COVID-19 restrictions being put in place in the UK until early 2022. We test the hypothesis that those who use psychedelics had less frequent depression and anxiety symptoms across timeframe relative to their baselines, and relative to those who used other illicit drugs or who were drug naive.

The K-modes clustering algorithm was used to cluster all recontacted individuals based on their reported patterns of drug use during the COVID-19 pandemic. One participant could only belong to one cluster. Six distinct clusters were identified comprising those who used: cannabis (N = 2827), cocaine and cannabis (N = 393), cocaine only (N = 629), psychedelics and cannabis (N = 430), polydrug (N = 274) and those indicating ‘other’ (N = 283; Supplemental Figure S2). Users of ‘other’ drugs were excluded from subsequent analyses due to the inability to pinpoint what the substances were. The remaining datasets belonged to N = 32,491 drug-naive participants, N = 10,293 with a history of drug use and N = 1420 who stopped using drugs during the pandemic.

Here, only complete datasets from participants recruited prior to restrictions in the UK and who responded to at least one recontact timepoint (thus completing another set of mental health measures alongside indicating whether they have used drugs during the COVID-19 pandemic) were used in our analyses. A total of N = 48,757 follow-up entries were retained, belonging to 30,711 unique individuals who completed at least one recontact. Among them, N = 17,354 participants engaged in just one recontact, while N = 8668 and 4689 individuals completed two and three recontacts, respectively. These included N = 15,591 drug-naive individuals, N = 5738 historic users, N = 857 individuals who stopped using drugs during the pandemic, N = 1515 cannabis users, N = 306 cocaine users, N = 228 psychedelics and cannabis users, N = 210 cocaine and cannabis users and N = 136 polydrug users. Sociodemographic characteristics and lifestyle choices of the participants who entered the longitudinal mental health analysis are reported in Supplemental Table S1, whereas the cohort sociodemographics at recruitment and cross-sectionally at follow-up have been reported in previous publications (Bălăeț et al., 2023a, 2023b; Hampshire et al., 2021a).

Full mixed effects model outputs are reported in the supplement (Supplemental File 1). In summary, at baseline, drug use clusters had generally worse mental health scores relative to drug-naive individuals for all mental health symptoms and across all timepoints. This included historic users and those who stopped using drugs during the pandemic (Figure 2).

Regarding the psychedelics and cannabis cluster specifically, there was an effect of cluster indicative of worse baseline depression scores (β = 0.23 SD, 95% CI: 0.003, 0.46, p = 0.04) and worse baseline concentration scores (β = 0.34 SD, 95% CI: 0.10, 0.57, p = 0.003) for those who responded in December 2020 relative to drug-naive individuals. Psychedelics and cannabis users who responded in June 2021 had better baseline insomnia scores (β = −0.28 SD, 95% CI: −0.56, −0.004, p = 0.04) than drug-naive individuals, while psychedelics and cannabis users who responded in January 2022 had worse depression scores (β = 0.35 SD, 95% CI: 0.16, 0.54, p < 0.001) and worse mental health composite scores (β = 0.29 SD, 95% CI: 0.10, 0.48, p = 0.002) than drug-naive individuals. Cannabis-only users, irrespective of the timepoint they engaged with a follow-up, had on average worse baseline depression, concentration, anxiety and mental health composite scores relative to drug-naïve individuals. All average mental health scores of cannabis-only users who responded in January 2022 were worse at baseline relative to drug-naïve individuals.

For depression scores, there was a significant interaction for historic users (β = 0.06 SD, 95% CI: 0.011, 0.014, p = 0.01) who had worse depression scores in January 2022 and psychedelics and cannabis users who had better depression scores in January 2022 (β = −0.26 SD, 95% CI: −0.44, −0.08, p = 0.003). In January 2022, there was also a significant interaction for psychedelics and cannabis users who had better anxiety scores (β = −0.24 SD, 95% CI: −0.41, −0.06, p = 0.007). For the composite mental health score, in January 2022, there was a significant interaction between timepoint and the psychedelics and cannabis cluster (β = −0.2 SD, 95% CI: −0.35, −0.04, p = 0.01). This was not the case for those who only used cannabis during the same period of time, who presented with no longitudinal changes in any of the variables.

ANOVA and Tukey post hoc tests were carried out to investigate differences between clusters at follow-up. The main cluster of interest, psychedelics and cannabis, showed worse concentration scores (mean difference = 0.48 SD, 95% CI: 0.11, 0.85, p = 0.002) and worse composite mental health scores (mean difference = 0.39 SD, 95% CI: 0.02, 0.76, p = 0.02) relative to drug-naive individuals in December 2020. They were no different relative to drug-naive individuals in June 2021 or January 2022 in any mental health symptoms. However, in January 2022, cannabis-only users were significantly worse than drug-naive individuals for every mental health symptom. Full results are reported in Supplemental File 2.

At all timepoints, most individuals who used drugs during the pandemic reported no change in frequency of use from their pre-restriction levels. Approximately one in four individuals reported using drugs less frequently, and one in five reported using them more frequently. Distinct patterns were observed for different drug use clusters at different timepoints, which differed from one another. Just above half of psychedelics and cannabis users reported no change from their baseline levels, with the rest being almost evenly split between using less and using more, a pattern observed at all follow-up timepoints.

Finally, we examined whether there were differences in the changes from pre-restrictions to follow-up in the different mental health symptoms deltas (difference in scores between pre-restrictions to follow-up) based on users’ self-reported change in frequency of drug use from pre-pandemic times to follow-up. Positive deltas indicate a worsening of scores, whereas negative deltas indicate an improvement in scores (Figure 3). Segregation by drug cluster is reported in Supplemental Materials.

Results from the two-way permutational ANOVA (Figure 4) showed that changes in composite mental health were associated with changes in drug use, regardless of cluster, evidenced by a significant main effect of change (F = 7.03, p = 0.001), but not of cluster or an interaction between change and cluster. For depression, there was a significant main effect of change (F = 4.78, p = 0.01), but not of cluster or interaction. For anxiety, there was a significant main effect of change (F = 3.54, p = 0.04), but the main effect of cluster and the interaction term were not significant. For concentration, the main effect of change was significant (F = 6.56, p = 0.001), but not of cluster or interaction. For tiredness, there was a significant main effect of change (F = 9.07, p < 0.001) and a significant interaction (F = 2.49, p = 0.01), but not of cluster. For insomnia, there were no significant main effects of change, cluster or interaction.

Supplemental Figure S4 illustrates the split per cluster.

There was no statistically significant difference in the number of participants who used psychedelics and cannabis during the COVID-19 pandemic and indicated a different type of change relative to their self-perceived baseline before the pandemic (Figure 5(a)). A two-way ANOVA analysis revealed a main effect of change in June 2021 (F(2, 288) = 20.54, p < 0.001) and a main effect of change in January 2022 (F(2, 624) = 6.72, p < 0.001) but no effect of mental health score, nor an interaction between the two at any timepoints. Those who used more by June 2021 had worse mental health scores, whereas those who used less by January 2022 had better scores at that time (Figure 5(b)). Longitudinally, those who used more drugs during the pandemic had worse mental health scores in December 2020 and June 2021 relative to their baseline scores, whereas those who used less had better scores in January 2022 relative to their baseline scores (Figure 5(c)).

This study examined how mental health evolved over time in relation to different patterns of illicit drug use during the COVID-19 pandemic within a large citizen science cohort. We identified six distinct clusters of drug users, with cannabis only emerging as the most frequent pattern, followed by cocaine, cocaine and cannabis, psychedelics and cannabis and polydrug use. These clusters are consistent with established naturalistic drug user behaviours (Hakkarainen et al., 2019) and have been replicated using different algorithms (Bălăeț et al., 2023b). They also match the official order of prevalence of substances being used in the UK during the period of time the study ran (Office for National Statistics, 2023). Together, these aspects underscore the ecological validity of using a data-driven clustering approach for studying associations between drug use and other variables.

Significant differences in mental health scores were observed before the pandemic between those who went on to use drugs during the pandemic and drug-naive individuals. These findings are consistent with the literature suggesting drug users have generally worse mental health than non-users (Torrens et al., 2011). We also found that those who increased their drug use frequency during the pandemic, irrespective of what drugs they used more of, had worse mental health scores during the pandemic. Previous research has shown a direct relationship between the quantity of drug use and more severe mental health issues (Ringen et al., 2008).

Mental health in most drug use clusters remained stable over time, except for the psychedelics and cannabis cluster. At follow-up, this cluster showed significant within-subject improvements. Comparing pre-restrictions data to January 2022, individuals in this group had significantly worse depression and mental health composite scores than drug-naive individuals at baseline, but these differences diminished over time, with no significant differences relative to drug-naive individuals remaining at follow-up. Anxiety scores also dropped significantly in this cluster, though the differences from drug-naive individuals did not reach statistical significance at either baseline or follow-up. Further analyses suggest this might be due to individuals in this cluster generally using fewer drugs by January 2022 relative to their pre-pandemic baseline. By contrast, cannabis-only users consistently showed poorer mental health across all symptoms compared to drug-naive individuals, suggesting the change in mental health scores might be related to the additional use of psychedelics within that cluster.

This observation accords with previous findings that link naturalistic use of psychedelics to improved mental health (Forstmann et al., 2020), and is consistent with Bouso et al. (2023), a longitudinal analysis carried out during the pandemic on psychedelics users. Improvements in mental health have also been reported in within-subject design studies of psychedelics users in naturalistic settings (Nygart et al., 2022). Our analysis also adds further nuance to the findings previously presented in Bălăeț et al. (2023b); specifically, in that study, we found that those who went on to use drugs during the peak of the COVID-19 pandemic, including those in the psychedelics and cannabis cluster, had worse mental health scores than drug-naive individuals during that time. Here, we illustrate that while cross-sectionally psychedelics and cannabis users indeed had worse mental health before and during the peak of the pandemic, in a within-subjects design, there were improvements, and an eventual normalisation of mental health scores to be on par with drug-naive individuals by 2022, when most restrictions were already lifted in the UK.

There are several pertinent explanations for this result. At the population level, drug users have worse mental health than drug-naive individuals – and it could be that (novel) use of psychedelics, albeit less drug use more generally, during crisis times, normalises those differences. Another possible explanation is that the context is more influential in driving the effects of psychedelics than it is in driving the effects of other drugs. We note that related to the early stages of the pandemic, when restrictions and uncertainty were more severe in the UK, the participants in the psychedelics and cannabis cluster who used more drugs had worse mental health scores at those timepoints relative to their pre-restrictions baselines. This contrasts with participants surveyed later on, when restrictions were more relaxed and a lot of things started returning to normal, in January 2022, who had similar mental health scores to their pre-restrictions baselines, despite using more drugs. This explanation is in line with previous work discussing increased suggestibility during the psychedelic experiences (Carhart-Harris et al., 2018), the importance of set and setting (Hartogsohn, 2017) and how the general societal discourse could affect users (Bălăeţ, 2024; Bunce, 1979). A potential mechanistic explanation for the improvements observed in the psychedelics and cannabis cluster may also relate to the longer interval between baseline and follow-up when these improvements were noted (approximately 2 years, compared to 1 or 1.5 years in the other analyses). This extended timeframe could imply either that psychedelic use occurred more proximally to the point of survey, coinciding with a relatively stable post-pandemic context conducive to more positive outcomes, or that sufficient time had elapsed since earlier use to allow for psychological integration, thereby supporting longer-term improvements in mental health.

Alternative explanations should also be considered. One possibility is regression to the mean – a statistical phenomenon in which extreme scores at one timepoint tend to move closer to the population average upon subsequent measurement. Given the relatively poorer mental health at baseline in the psychedelics and cannabis cluster, and the apparent normalisation of these measures at follow-up, it is plausible that some of the observed improvement reflects this artefact rather than a specific effect of psychedelic use. Another non-causal interpretation is that individuals whose mental health was already improving in the aftermath of the COVID-19 pandemic may have been more inclined to experiment with psychedelics nearer to the time of the final survey. However, as the timing of psychedelic use was not recorded, it is not possible to determine whether improvements in mental health preceded or followed psychedelic use. Additional explanations include selective engagement with research, whereby participants with less positive experiences may have been less likely to remain engaged with the study, or the influence of unmeasured third variables that independently contributed to improved outcomes.

As our data-driven clustering reveals, virtually all people who used psychedelics also used cannabis during the pandemic, making it impractical to study the association of naturalistic psychedelics use with mental health in isolation. Looking at the cannabis cluster during the pandemic’s peak revealed notably worse mental health scores compared to drug-naive individuals across multiple symptoms, including anxiety, concentration, insomnia and tiredness, though the average mental health of this group did not significantly change over time. The associations between cannabis use and mental health are widely debated – most notably, there is evidence that it can amplify anxiety and paranoia in users while it is also being clinically prescribed for treating anxiety (Van Ameringen et al., 2020). Furthermore, there is evidence to suggest that heavy cannabis use may increase the risk of developing depression (Lev-Ran et al., 2014). Conversely, experimental findings show conflictual evidence that cannabis users typically exhibit blunted reactivity to negative emotions (Somaini et al., 2012) and stress (Cuttler et al., 2017), which, in turn, might help them cope with such symptoms. Evidence suggests consuming cannabis at the same time as psychedelics may make psychedelic experiences more intense in a dose-dependent fashion (Kuc et al., 2022), which may explain potential negative associations between combined use and worse mental health. However, since we do not have information on whether the use of psychedelics and cannabis occurred simultaneously or asynchronously across our cohort, it is not possible to firmly draw such conclusions. Intense experiences may provide grounds for improved mental health with adequate integration (Gashi et al., 2021), or benefits from psychedelics use may be independent of cannabis use during the same timeframe. Given the prevalence of combined psychedelics and cannabis use, a timely challenge for future studies is to determine how these drugs interact with each other and their effects on mental health, which may only be achievable under prospective study designs with sufficient power to map out nuanced drug use patterns.

Observational studies of naturalistic drug use present a cost-effective opportunity for understanding the associations between drug use and mental health outcomes and could help inform studies aimed at developing these substances as therapies, which are otherwise logistically challenging to carry out (Bălăeţ, 2025; Barnett, 2025; Barnett et al., 2022, 2025). Strengths of our study include that we addressed the relationship between recreational drug use and mental health in unsupervised settings while minimising the selection bias common when recruiting in drug-related social media forums. This was achieved through a recruitment approach that did not mention drug-related questions and avoided promotion on drug-centric social media channels. The participant group was diverse, including drug-naive individuals, historic users, and those who stopped using drugs during the COVID-19 pandemic; although diversity in sociodemographic characteristics, expectedly, dwindled with repeated follow-up. The methodology was adapted to recognise the complexity of drug use behaviours, including the concurrent use of cannabis with other substances, and to reflect the dynamic nature of drug use patterns where participants increased or decreased their use. Our study also has limitations. As data collection was fully automated online, we did not conduct the sorts of interviews that are sometimes used to provide comprehensive baseline data on participants’ drug use histories, which limits our ability to assess the influence of prior drug use on mental health. For instance, we did not collect data pertaining to dosage, frequency or context of drug use, which are likely important in determining mental health outcomes, nor have we gathered specific information on other drugs individuals may use in the UK, such as amphetamines. We also did not control for potential confounding effects of psychiatric medications that participants might have taken. Furthermore, participants who engaged in the study at multiple timepoints may be less likely to represent all patterns of naturalistic drug use as some sub-populations, for example, those who have addictions or are otherwise marginalised, may be less likely to engage in research. The focus of the analysis was solely on illicit drugs (in the UK), omitting licit substances like alcohol and tobacco – those data will form the basis of future analyses. Finally, the sample predominantly comprised British participants, which may limit the generalisability of the findings.

Taken together, our findings demonstrate that, on average, individuals who use drugs report poorer mental health compared to drug-naïve individuals, and these average differences remained relatively stable from before the COVID-19 restrictions to the years following their implementation. Notably, the cluster comprising psychedelics and cannabis use exhibited distinct temporal changes, with average anxiety and depression scores improving from pre-restriction baselines to the final follow-up. Additionally, we observed that increases in drug use were associated with declines in mental health over time, regardless of the specific substances involved. It is important to recognise that these results reflect population-level trends and do not account for individual variations in mental health trajectories. Future research should investigate whether the observed changes in mental health within the psychedelics and cannabis cluster are driven by alterations in the use of cannabis, psychedelics or their combined effects, particularly given their prevalent concurrent use; or whether they are a product of other synergistic or independent factors (such as the quality of interpersonal relationships, concurrent treatment for mood disorders or lifestyle changes).