Demographic information, such as biological sex, gender, age, continent of residence, educational level, and daily occupation, was collected at baseline. Additionally, information about participants’ previous experience with psychedelics (i.e., ayahuasca, DMT, 5-MeO-DMT, LSD, novel lysergamides (e.g., 1P-LSD, ALD-52), psilocybin, Salvia divinorum, ibogaine, and mescaline) in both full and microdoses was collected at baseline.

At baseline, participants were asked if they had any experience in the practice of meditation/mindfulness. If this question was answered with ‘yes’, it was asked what meditation/mindfulness tools were used, where they could choose multiple answers from the following options: ‘I followed an online course’, ‘I use(d) a mobile application’, ‘I watch(ed) Youtube videos’, ‘I follow(ed) sessions at a retreat’, or the option to provide an answer through free text entry. Subsequently, it was asked when the last time was that the participant practiced meditation/mindfulness (i.e., ‘more than one year ago’; ‘less than one year ago, more than one month ago’; ‘less than one month ago, more than one week ago’; or ‘within the past seven days’). A variable ‘recent mindfulness’ was created to group individuals based on the recentness of their mindfulness practice (0 = no experience with meditation/mindfulness or practiced it more than 7 days ago; 1 = practiced meditation/mindfulness within the past 7 days), to differentiate respondents who recently practiced meditation/mindfulness from respondents who did not.

Participants were asked if they had been diagnosed by a medical doctor or a therapist with a psychiatric, neurological, or physical disorder, and if so, what these diagnoses were. Pre-set answer options included ‘ADHD’, ‘depression’, ‘anxiety disorder’, ‘substance use disorder’, ‘dyslexia’, ‘autism/Asperger syndrome’, ‘obsessive-compulsive disorder’, ‘bipolar disorder, chronic pain’, ‘cluster headaches’, ‘epilepsy’, ‘migraines’, ‘post-traumatic stress disorder (PTSD)’, ‘schizophrenia’, ‘I do not want to mention’, or the option to provide another answer in a textbox. A variable ‘comorbidity’ was constructed differentiating participants with at least one comorbid diagnosis alongside ADHD from participants without comorbid diagnoses alongside ADHD or without an ADHD diagnosis (0 = only ADHD or no ADHD; 1 = ADHD and at least one other diagnosis).

Respondents who indicated having an ADHD diagnosis were asked at what age they received the diagnosis and if they were currently using prescribed ADHD medication, stopped using it, or never used it. If they indicated to be using prescribed medication, it was asked what type of medication this was (i.e., ‘Adderall (amphetamine)’, ‘Concerta (methylphenidate)’, ‘Dexedrine (amphetamine)’, ‘Focalin (dexmethylphenidate)’, ‘Ritalin (methylphenidate)’, ‘Strattera (atomoxetine hydrochloride)’, ‘I do not want to mention’, or a free text entry). In the case prescribed ADHD medication was discontinued in the past, it was asked what the reasons were for this: ‘it did not relieve my symptoms’, ‘because of psychological side effects’, ‘because of physical side effects’, ‘I do not want to mention’, or the option to provide another reason through free text entry. A variable ‘medication use’ was constructed differentiating participants using conventional medication alongside MD during the study from participants without conventional medication, who were only MD during the study (0 = only MD; 1 = MD and using conventional ADHD medication).

The self-report, short screening version of the Conners’ Adult ADHD Rating Scale (CAARS-S:SV) (30) was used to assess ADHD symptoms at baseline. This 30-item questionnaire assesses the core ADHD symptoms (i.e., inattention and hyperactivity/impulsivity) as well as related problem areas like problems with self-concept. Participants indicated to what extent the items described them on a four-point Likert scale from 0 (not at all, never) to 4 (very much, very frequently). Nine items belong to the inattention subscale, capturing problems experienced with attention and containing items such as ‘I lose things necessary for tasks or activities (e.g., to-do lists, pencils, books, or tools)’. Nine items belong to the hyperactivity/impulsivity subscale, capturing symptoms related to both hyperactivity and impulsivity and containing items such as ‘I have trouble waiting in line or taking turns with others’. The remaining 12 items belong to the ADHD index, capturing features of ADHD that are not included in the DSM diagnostic criteria, such as ‘sometimes my attention narrows so much that I am oblivious to everything else; other times it’s so broad that everything distracts me’. A DSM-IV ADHD total symptom score can be calculated by summing the scores of the inattention and hyperactivity/impulsivity subscales. The CAARS-S:SV has good internal consistency and inter-rater reliability (31), high criterion validity and moderate concurrent validity (32).

T-scores were calculated for each subscale using the scores of the standardization sample provided in the technical manual consisting of non-clinical adults in the same age range and of the same sex. Subscale T-scores equal or above 65 indicate clinically elevated symptoms according to the technical manual (30). Therefore, participants without an ADHD diagnosis who had T-scores below 65 on all CAARS-S:SV subscales were excluded from all analyses. Although the CAARS-S:SV was included in the surveys at all three time points, only the baseline scores were used in the current study. For details about the ADHD symptom scores at all time points, see Haijen and colleagues (7).

To assess trait mindfulness, the 15-item Five Facet Mindfulness Questionnaire (FFMQ-15) was used (33). The FFMQ-15 consists of five subscales, with every three items capturing one aspect of mindfulness. The first subscale observation refers to attending to or noticing internal and external experiences and contains items such as ‘when I take a shower or a bath, I stay alert to the sensations of water on my body’. Description assesses the ability to put one’s experiences into words and contains items such as ‘I’m good at finding words to describe my feelings’. Acting with awareness captures the ability to attend to the present moment activity, without behaving automatically and while the attention is focused elsewhere. This subscale contains three reverse-phrased items such as ‘I do not pay attention to what I’m doing because I’m daydreaming, worrying, or otherwise distracted’. Non-judging of inner experience describes accepting and not evaluating emotions and thoughts as good or bad and contains three reverse-phrased items such as ‘I believe some of my thoughts are abnormal or bad and I should not think that way’. Non-reactivity to inner experiences involves detachment of emotions and thoughts allowing them to come and go without being carried away by them and contains items such as ‘when I have distressing thoughts or images, I “step back” and am aware of the thought or image without getting taken over by it’ (34). Items were rated on a five-point Likert scale ranging from 1 (never or very rarely true) to 5 (very often or always true). Subscale scores range from 3 to 15 and can be summed to achieve a total score ranging from 15 to 75. Reverse-phrased items were recoded. High scores on each mindfulness facet reflect a higher level of mindfulness. The FFMQ-15 showed adequate internal consistency and did not differ from the long form of the FFMQ in terms of convergent validity (34).

To assess personality traits, the 10-item version of the Big Five Inventory (BFI; (35)) was included. This short questionnaire contains two items, of which one is reverse-phrased, describing each of the five Big Five personality traits. Example items include for extraversion ‘I see myself as someone who is outgoing, sociable’, for agreeableness ‘I see myself as someone who is generally trusting’, for conscientiousness ‘I see myself as someone who does a thorough job’, for neuroticism ‘I see myself as someone who gets nervous easily’, and for openness ‘I see myself as someone who has an active imagination’. Items are rated on a five-point Likert scale ranging from 1 (disagree strongly) to 5 (agree strongly). Reverse-phrased items were recoded. The scores of the two items belonging to one subscale were summed and divided by two to achieve an average subscale score ranging from 1 to 5. The BFI-10 has shown to be an adequate assessment of personality, with good validity and reliability metrics (35).

Through short daily surveys starting from the day after the baseline survey until the 4-week time point, participants were asked each day if they had taken a microdose that day (yes/no), if yes, what substance (LSD, novel lysergamides (e.g., 1P-LSD), psilocybin/psilocin (magic mushrooms/truffles), mescaline (e.g., san pedro), or free text) and dose they took (free text).

Compound symmetry was used as a covariance structure for this LMM. A main effect of time was found on the total score of the FFMQ-15 [F_{(2, 120.9)} = 19.29, p < 0.001, η_p² = 0.24]. Pairwise comparisons showed that total mindfulness scores were higher at both 2 W (Δ2W–0 W = 3.62, p < 0.001) and 4 W (Δ4W–0 W = 5.97, p < 0.001) compared to baseline. Scores were also higher at 4 W compared to 2 W (Δ4W–2 W = 2.35, p = 0.043) (see Figure 1A). Further, a significant interaction between time and medication use was found [F_{(2, 120.8)} = 3.77, p = 0.026, η_p² = 0.06]. The estimates of fixed effects showed that scores were lower at 2 W in respondents using conventional ADHD medication compared to respondents who did not use conventional medication (β = −5.16, p = 0.009). The difference in scores was not significant at baseline (β = −0.25, p = 0.847) or 4 W (β = −1.41, p = 0.544). No interaction between comorbidity and time was found [F_{(2, 121.3)} = 0.35, p = 0.704, η_p² = 0.01].

When including the variable recent mindfulness in the model, the effect of time [F_{(2, 120.9)} = 11.18, p < 0.001, η_p² = 0.16], including the pairwise comparisons, and the interaction between time and medication use [F_{(2, 121.2)} = 3.30, p = 0.040, η_p² = 0.05] remained significant. There was no interaction between time and recent mindfulness [F_{(2, 121.1)} = 0.18, p = 0.833, η_p² = 0.00].

Compound symmetry was used as the covariance structure for this model. The LMM showed an effect of time on the FFMQ-15 observation scores [F_{(2, 119.1)} = 3.10, p = 0.049, η_p² = 0.05]. Pairwise comparisons showed that scores were higher at 4 W compared to baseline (Δ4W–0 W = 0.75, p = 0.008). Scores did not differ between 2 W and baseline (Δ2W–0 W = 0.40, p = 0.158) and 2 W and 4 W (Δ4W–2 W = 0.35, p = 0.545). Furthermore, no interaction effect between time and medication use [F_{(2, 119.5)} = 1.80, p = 0.170, η_p² = 0.03] or time and comorbidity [F_{(2, 119.8)} = 0.01, p = 0.989, η_p² = 0.00] was found.

When including the variable recent mindfulness in the model, the effect of time was no longer significant [F_{(2, 117.1)} = 0.33, p = 0.718, η_p² = 0.00; see Figure 1B]. No interaction between time and recent mindfulness was found [F_{(2, 117.2)} = 1.96, p = 0.146, η_p² = 0.02].

The First-order autoregressive (AR1) covariance structure was the best fit for this model. A main effect of time was found on the FFMQ-15 scores [F_{(2, 117.1)} = 7.75, p < 0.001, η_p² = 0.12; see Figure 1C]. Pairwise comparisons revealed that scores were higher at 2 W (Δ2W–0 W = 0.81, p = 0.002) and 4 W (Δ4W–0 W = 1.42, p < 0.001) compared to baseline. The scores did not differ between the 2 W and 4 W (Δ4W–2 W = 0.61, p = 0.121). No interaction between time and medication use [F_{(2, 119.4)} = 1.86, p = 0.161, η_p² = 0.03] or time and comorbidity [F_{(2, 119.6)} = 0.69, p = 0.505, η_p² = 0.00] was found.

When including the variable recent mindfulness in the model, the effect of time remained significant [F_{(2, 117.1)} = 6.87, p = 0.002, η_p² = 0.11] and the results of the pairwise comparisons remained unchanged. Additionally, no interaction effect between time and recent mindfulness was found [F_{(2, 117.2)} = 2.32, p = 0.103, η_p² = 0.02].

An unstructured covariance structure was used for this model. A main effect of time was found on the FFMQ-15 acting with awareness scores [F_{(2, 51.3)} = 4.70, p = 0.013, η_p² = 0.16]. Pairwise comparisons showed that scores were higher at both 2 W (Δ2W–0 W = 0.64, p = 0.029) and 4 W (Δ4W–0 W = 1.01, p = 0.005) compared to baseline. Scores did not differ between 2 W and 4 W (Δ4W–2 W = 0.37, p = 0.490). Furthermore, the interaction effects between time and medication use [F_{(2, 51.7)} = 1.09, p = 0.343, η_p² = 0.04] and time and comorbidity [F_{(2, 51.4)} = 0.23, p = 0.792, η_p² = 0.01] were not significant.

When including the variable recent mindfulness in the model, the time effect was no longer significant [F_{(2, 50.7)} = 2.88, p = 0.066, η_p² = 0.10; see Figure 1D]. Further, the interaction effect between time and recent mindfulness was not significant [F_{(2, 50.3)} = 0.49, p = 0.616, η_p² = 0.10].

Compound symmetry was used as a covariance structure. A main effect of time was found on the FFMQ-15 non-judging of inner experience scores [F_{(2, 115.3)} = 13.40, p < 0.001, η_p² = 0.19; see Figure 1E]. Pairwise comparisons showed that scores were higher at 2 W (Δ2W–0 W = 1.04, p < 0.001) and 4 W (Δ4W–0 W = 1.94, p < 0.001) compared to baseline. Scores were also higher at 4 W compared to 2 W (Δ4W–2 W = 0.90, p = 0.036). Further, a significant interaction between time and medication use was found [F_{(2, 115.9)} = 4.82, p = 0.010, η_p² = 0.08]. Estimates of fixed effects showed that non-judging of inner experience scores were lower at 2 W (β = −2.32, p = 0.002), not at baseline (β = −0.23, p = 0.643) or 4 W (β = −0.90, p = 0.306) for respondents using conventional medication alongside MD compared to respondents not using conventional medication. No interaction between time and comorbidity was found [F_{(2, 116.4)} = 0.79, p = 0.456, η_p² = 0.01].

When including the variable recent mindfulness in the model, the main effect of time [F_{(2, 115.0)} = 8.46, p < 0.001, η_p² = 0.13], including the pairwise comparisons, and the interaction between time and medication use [F_{(2, 115.2)} = 4.55, p = 0.013, η_p² = 0.07] remained significant. Additionally, no interaction effect between time and recent mindfulness was found [F_{(2, 115.2)} = 0.00, p = 0.999, η_p² = 0.00].

First-order autoregressive (AR1) was used as a covariance structure for this LMM. A main effect of time was found on the FFMQ-15 non-reactivity to inner experience scores [F_{(2, 131.0)} = 5.24, p = 0.006, η_p² = 0.07]. Pairwise comparisons showed higher scores at both 2 W (Δ2W–0 W = 0.73, p = 0.011) and 4 W (Δ4W–0 W = 1.05, p = 0.007) compared to baseline. Scores did not differ between 2 W and 4 W (Δ4W–2 W = 0.32, p = 0.960). No interaction between time and medication use [F_{(2, 132.6)} = 1.86, p = 0.159, η_p² = 0.03] or time and comorbidity [F_{(2, 133.4)} = 0.06, p = 0.945, η_p² = 0.00] was found.

After including the recent mindfulness variable in the model, the main effect of time was no longer significant [F_{(2, 132.0)} = 2.43, p = 0.092, η_p² = 0.04; see Figure 1F]. No interaction between time and recent mindfulness was found [F_{(2, 132.2)} = 0.22, p = 0.806, η_p² = 0.00].

A First-order autoregressive (AR1) covariance structure was used for this model. A main effect of time on BFI-10 conscientiousness scores was found [F_{(2, 136.3)} = 3.77, p = 0.025, η_p² = 0.05]. Bonferroni-corrected pairwise comparisons showed that scores were higher at 4 W compared to baseline (Δ4W–0 W = 0.40, p = 0.002). Scores did not differ between baseline and 2 W (Δ2W–0 W = 0.16, p = 0.159) or between 2 W and 4 W (Δ4W–2 W = 0.24, p = 0.060; see Figure 2A). No interactions between time and medication use [F_{(2, 137.9)} = 0.69, p = 0.502, η_p² = 0.01] and time and comorbidity [F_{(2, 138.3)} = 0.10, p = 0.909, η_p² = 0.00] were found. After correcting for multiple testing, the effect found on conscientiousness scores was no longer significant.

An Ante-Dependence: First-Order covariance structure was the best fit for the model. A significant effect of time on BFI-10 neuroticism scores was found [F_{(2, 60.8)} = 7.19, p = 0.002, η_p² = 0.19]. Scores were lower at 4 W compared to baseline (Δ4W–0 W = −0.60, p < 0.001) and 2 W (Δ4W–2 W = −0.34, p = 0.014; see Figure 2B). Scores were not significantly lower at 2 W compared to baseline (Δ2W–0 W = −0.26, p = 0.056). No interactions between time and medication use [F_{(2, 60.2)} = 1.08, p = 0.350, η_p² = 0.034] and time and comorbidity [F_{(2, 60.4)} = 0.23, p = 0.796, η_p² = 0.01] were found. The variable comorbidity did have a main effect on the neuroticism scores [F_{(1, 103.6)} = 6.65, p = 0.011, η_p² = 0.06], showing higher neuroticism scores for respondents with comorbid diagnoses alongside the ADHD diagnosis compared to respondents without comorbidities alongside ADHD at baseline (β = 0.38, p = 0.007) and 2 W (β = 0.51, p = 0.048), but not at 4 W (β = 0.58, p = 0.064).

A compound symmetry covariance structure was the best fit for this model. A main effect of time was found on the BFI-10 extraversion scores [F_{(2, 123.8)} = 3.58, p = 0.031, η_p² = 0.06]. Corrected pairwise comparisons showed that scores were higher at 2 W compared to baseline (Δ2W–0 W = 0.24, p = 0.039) (see Figure 2C). Scores did not differ between baseline and 4 W (Δ4W–0 W = 0.20, p = 0.249), or between 2 W and 4 W (Δ4W–2 W = −0.04, p > 0.999). No interaction between time and medication use [F_{(2, 124.6)} = 1.55, p = 0.217, η_p² = 0.02] or time and comorbidity [F_{(2, 125.0)} = 0.43, p = 0.652, η_p² = 0.01] was found. Comorbidity had an overall effect on extraversion scores [F_{(1, 316.8)} = 8.49, p = 0.004, η_p² = 0.03]. Respondents with at least one comorbid diagnosis alongside ADHD had lower scores at baseline (β = −0.48, p = 0.001) and 4 W (β = −0.62, p = 0.019), but not at 2 W (β = −0.38, p = 0.103). After correcting for multiple testing, the effect found on extraversion was no longer significant.

Compound symmetry was chosen as the covariance structure for this LMM. No effect of time was found on BFI-10 agreeableness scores [F_{(2, 129.5)} = 2.05, p = 0.133, η_p² = 0.03; see Figure 2D]. Further, no interaction between time and medication use [F_{(2, 130.3)} = 2.16, p = 0.120, η_p² = 0.03] or time and comorbidity was found [F_{(2, 130.8)} = 0.28, p = 0.755, η_p² = 0.00].

An unstructured covariance matrix was the best fit for this model. The time effect on BFI-10 openness scores was not significant [F_{(2, 51.4)} = 0.42, p = 0.662, η_p² = 0.02; see Figure 2E]. No interaction between time and medication use [F_{(2, 51.8)} = 0.27, p = 0.766, η_p² = 0.01] or time and comorbidity [F_{(2, 51.7)} = 1.07, p = 0.350, η_p² = 0.04] was found.