A psychedelic renaissance is happening right now; but before classic hallucinogenic drugs can be accepted into mainstream medicine, their safety must be demonstrated. To move towards this goal, Norwegian scientists published a study claiming that psychedelics are not linked to long-term mental health problems; these findings, however, were quickly challenged. Dose of Science takes a look at the debate.
Classic psychedelics – such as LSD, psilocybin and mescaline – were intensively researched during the 1960s and 70s, both by psychologists and neuroscientists.The main aim was to investigate how the psychedelic experience could help people with psychological problems (see Dose of Science’s article on the efficacy of LSD therapy in curing alcoholism). In the 1960s, LSD emerged from the lab, and the use of hallucinogens became associated with the nascent hippy counterculture. Due to the negative media hype of this era, a large section of the public tends to associate psychedelic drugs with psychological issues. To address this concern, Teri Krebs and Pal Johansen published an epidemiological paper on the relationship between the use of psychedelic drugs and long-term mental health .
The study uses data from the National Survey on Drug Use and Health 2008-2011 (NSDUH) to estimate the strength of association between lifetime psychedelic use and various mental health outcomes at the population level. The NSDUH is a representative, large-scale survey of the US population. 135,095 people responded to the survey, of whom 19,299 had used psychedelics at some point in their life. The study is a retrospective one, cross-sectional by design, and therefore has some inherent limitations in terms of investigating causal relationships.
The odds ratio is a standard statistical estimate of the association between two variables. In our case, the two variables are the ‘lifetime use of psychedelic drugs’ and various mental health outcomes during the past year (while the `odds’ is a technical term that is defined as 1/(1-p)). An odds ratio > 1 indicates an increased risk of mental health problems due to the use of psychedelic drugs, while an odds ratio <1 indicates a decreased risk. Figure 1 below shows the estimated odds ratios for various mental health outcomes based on the NSDUH survey. Note that the odds ratios are adjusted for potential confounders – more on those later.
Figure 1 shows the adjusted odds ratio between lifetime psychedelic use and various mental health outcomes. The odds ratio has been adjusted for age, gender, ethnicity, income, education, marital status, willingness to engage in risky activities, and the lifetime use of ten other illicit drugs (aOR: adjusted odds ratio; CI: confidence interval).
The most important feature of Figure 1 is that for all but one of the mental health outcomes, the adjusted odds ratio of 1 falls within the confidence intervals. This means that statistically, the use of psychedelics cannot be linked to these mental health outcomes. Based on this figure, the authors concluded that “we failed to find any associations between lifetime use of psychedelics and serious psychological distress, receiving or needing mental health treatment, depression, anxiety, or suicidal thoughts or behaviour in the past year.”
Soon after the publication of this study, another Norwegian group, led by Ragnar Nesvag, challenged the paper . In a Letter to the Editor (as it is customary in science), they expressed their concern that the original paper only presents the adjusted odds ratios. In epidemiological studies, the unadjusted estimates of odds ratios are usually presented first, and then multiple adjusted estimates are considered. The reason for adopting this practice is that it is often not a trivial question, which variables are confounders in any given study.
Nesvag and colleagues calculated these unadjusted odds ratios, concluding that “it is clear that the group of individuals who reported lifetime use of psychedelic drugs had two to three times higher prevalence of all problematic mental health outcomes when not adjusting for lifetime use of other drugs”. This statement means that if the odds ratio estimates are not adjusted for confounders (for the list of confounders see the caption of Figure 1), then there seems to be a strong link between psychedelics and mental health issues, the exact opposite result of the original Krebs paper!
So, the crucial question is whether the odds ratios should be adjusted or not. While many confounders (such as ethnicity, income, education, etc.) are likely to be accepted by both groups, the ones which really make a difference are those related to use of other illicit drugs. Whether the use of non-psychedelic drugs is a confounder in the context of this study is a technical question, which would require a lot more statistics to analyse. What it comes down to, is whether the use of psychedelics causes the use of other drugs. If yes, then the first paper is an example of ‘overadjustment’ (a common fallacy in regression analysis, see  for technical details or  for a more intuitive explanation), which could bias the odds ratios towards the observed null hypothesis.
The authors of both teams are in agreement that one easy way to block the influence of non-psychedelic drugs from the analysis is to calculate the odds ratio for those who have only used psychedelic drugs. This calculation was done in the response of Krebs and Johansen  – see the table below.
Table 1: Association between psychedelic use and mental health, excluding people with a history of use of other illicit substances. Note that the models are still adjusted, but not for the use of non-psychedelic drugs, as this calculation focusses on people who have only used psychedelics (aOR: adjusted odds ratio; CI: confidence interval).
The first important feature to note is that in this case the odds ratios often exceed 1, indicating an association between psychedelic drug use and problematic mental health outcomes; the results, however, are not consistent across the two surveys (in both cases it is the same NSDUH survey, but taken in different years). The second feature to note is that the sample size is significantly smaller (n=192 for the 2001-2004 and n=156 for the 2008-2011), because most people who use psychedelics also used other illicit drugs. A consequence of the small sample size is that the confidence intervals are much wider, making it very difficult to draw firm conclusions.
So what do these conflicting results say about the long-term safety of psychedelic drugs? Well, most importantly, they suggest that this is a complex question, where more research is needed. The Krebs analysis is based on cross-sectional data, which cannot be used to make causal inferences. The lack of temporal information is the reason why it is not straightforward to assess whether the use of non-psychedelic drugs should be treated as a confounder – with a huge impact on the estimated odds ratios.
Another important unknown factor is the dosage. At a low dosage, psychedelics are unlikely to be associated with long-term mental health problems, but there may be such an association in the case of large doses. Since most recreational users do not start with a large dose, one could argue that at the aggregated statistical level, there is no association between psychedelics and mental health issues, but a stratified analysis could show that large doses of psychedelics do show such a link.
Furthermore – stepping back from the statistical perspective – it is likely that different people will have very different risk profiles for hallucinogenic drugs. This is the reason why, in clinical trials, psychedelics are not administered to patients with pre-existing mental conditions. Here, the same argument can made as for high-dose users: individuals with mental health problems are uncommon – hence, at the population level, psychedelics appear safe, despite being dangerous for certain people (and doses).
Our best guess, therefore, is that, as our understanding of psychedelic drugs progresses, we will move away from classifying them as safe or unsafe in an absolute sense; rather, better predictors will be developed, to show for whom, when, and at what dosages it is safe to use these substances.
: Johansen, Pål-Ørjan, and Teri Suzanne Krebs. "Psychedelics not linked to mental health problems or suicidal behavior: A population study." Journal of Psychopharmacology (2015): 0269881114568039.
: Nesvåg, Ragnar, et al. "The link between use of psychedelic drugs and mental health problems." Journal of Psychopharmacology 29.9 (2015): 1035-1040.
: Schsterman, Enrique F., Stephen R. Cole, and Robert W. Platt. "Overadjustment bias and unnecessary adjustment in epidemiologic studies." Epidemiology (Cambridge, Mass.) 20.4 (2009): 488.
: http://andrewgelman.com/2005/12/27/daughters_liber/ (accessed on 2016 Aug 1)
: Krebs, Teri Suzanne, and P. Ø. Johansen. "Reply letter: Mental health of people who have used classical psychedelics and no other illicit drugs." J Psychopharmacol 29.9 (2015): 1036.