Each drug-psychedelic interaction in the database is derived from primary human studies. Available literature is independently reviewed and curated, with findings summarised at the interaction-pair level. We assign a certainty rating using a GRADE-inspired approach that scores four domains capturing study quality and how directly and consistently the findings support the interaction.

Primary certainty grading is performed for subjective effects. Physiological outcomes are considered only in relation to adverse drug reactions (ADRs). An ADR certainty rating is assigned only when an ADR is reported for the interaction pair. Pharmacogenetic interactions are graded using the same four domains, with domain-specific criteria for study limitations and imprecision. Only human studies are graded.

Primary endpoint grading (subjective effects)

Domain	Penalty	Criteria
A: Design limitations and replication	0 1 2	At least two independent experimental studies with at least one well controlled (randomised or crossover) study with a clearly defined comparator and sufficient methodological detail to support the design classification One well controlled experimental study, or multiple controlled studies with incomplete control or limited methodological detail Poorly controlled study or uncontrolled observational evidence, case series or case reports
B: Indirectness	0 1 2	Evidence addresses the same clinical question and outcome, and differences in population, route or timing are unlikely to change interpretation, or the effect is consistent across these differences Some indirectness is present and could affect interpretation, for example small group differences such as age, or somewhat different timing or route that may change exposure, but the overall finding is still reasonably applicable Major indirectness that makes extrapolation uncertain, for example evidence comes mainly from a context where route, timing or population is likely to change the interaction mechanism or the meaning of the outcome
C: Inconsistency	0 1 2	Results are consistent across studies, or only one study is available and inconsistency is not assessable Some inconsistency or heterogeneity across studies but direction is broadly similar or explainable, for example by dose, timing or formulation Important inconsistency with conflicting directions and no clear explanation
D: Imprecision	0 1 2	Adequate precision, total n ≥ 30 across comparable studies, or one adequately sized study with clearly reported and tight estimates Imprecise, total n = 10-29, wide uncertainty, or only one study and precision is not clearly adequate Very imprecise, total n < 10, very wide uncertainty or sparse reporting

ADR-specific grading (only when an ADR is reported)

Domains B (Indirectness) and C (Inconsistency) are scored as above. Domains A and D use ADR-specific criteria.

Domain	Penalty	Criteria
A: Study limitations	0 1 2	Trial or prospective study with predefined adverse effects (AE) collection and monitoring Observational study with structured AE capture using defined criteria, for example a registry Case series, case reports or unclear ascertainment methods
D: Imprecision	0 1 2	ADR frequency is interpretable because the study reports events systematically with a clear number observed, for example “X of Y participants”, and the signal is not based on only a single event ADRs are reported with a clear number observed but there are few events, for example 1-4, or estimates are unstable ADR frequency cannot be interpreted because there is no clear number observed or evidence is limited to single case reports or small case series

Pharmacogenetic grading (DGI)

Domains B (Indirectness) and C (Inconsistency) are scored as above. Domains A and D use pharmacogenetic-specific criteria.

Domain	Penalty	Criteria
A: Study limitations	0 1 2	At least two independent studies with robust genotyping and appropriate analysis, including clear phenotype definitions and assessment of functionally relevant variants One strong study with robust genotyping and appropriate analysis, or multiple studies with some design limitations Poorly controlled or exploratory evidence, for example unclear genotyping methods or case reports
D: Imprecision	0 1 2	Functional assignment is reliable and the smallest key functional group is adequately sized, with clear effect estimates. At least 10 participants in the smallest functional group being compared, or a narrow confidence interval if smaller Either functional assignment is solid but key functional groups are small, or functional assignment is limited but still probably correct. Smallest functional group n = 5-10 Functional assignment is uncertain or key functional groups are very small (n < 5), or genotyping misses functionally important alleles and is likely to misclassify phenotypes

Domain penalty scores are summed (range 0 to 8) and translated into an overall certainty level:

Total score	Level of evidence	Interpretation
0 to 1	High	Very confident the interaction effect, including direction and clinically meaningful magnitude, is correct
2 to 3	Moderate	Probably correct. Further research could change magnitude but is unlikely to reverse direction
4 to 5	Low	Limited confidence. Further research is likely to change the estimate and may reverse direction
6 to 8	Very low	Uncertain. Any observed association is weak and findings may be mixed

At the study level, interaction effects are classified using a controlled vocabulary that describes the direction and qualitative impact of co-administration on the psychedelic experience. These classifications focus on changes to subjective, behavioural, physiological or pharmacokinetic outcomes.

Interaction effects are coded as one of the following values:

Value	Meaning
attenuates	Reduced intensity, duration or onset of psychedelic effects, including subjective and physiological outcomes
potentiates	Increased intensity, duration or earlier onset of psychedelic effects
no-change	No meaningful alteration of psychedelic effects observed
mixed	Different directions across outcome domains, dosing conditions or studies
unclear	Insufficient, inconsistent or uninterpretable evidence

Adverse drug reactions are recorded separately using a dedicated ADR flag and are not treated as directional interaction effects.

At the interaction-pair level, an overall interaction effect is derived by aggregating study-level classifications. When all studies report the same effect, that value is retained. When different effects are reported across studies, the overall interaction is classified as mixed. This approach avoids forcing consensus where evidence is heterogeneous.

Literature risk labelling

Each drug-drug interaction pair is assigned an interaction risk category describing the likelihood of a clinically noticeable interaction and its likely practical significance based on the available literature. This rating is distinct from the certainty grading.

Label	Definition
Probable	If the interaction occurs, it is likely to produce a clinically noticeable and meaningful change in the psychedelic response or safety profile in typical use, for example near complete blockade, large attenuation or potentiation of core subjective effects, an ADR, or a predicted ≥100% increase or ≥50% decrease in AUC
Possible	The interaction could produce a meaningful change, but the expected impact is moderate, context dependent or not consistently expressed across domains, for example partial attenuation, selective changes such as reduced anxiety with preserved positive effects, modest prolongation, a plausible but not clearly severe safety concern, or a predicted 25-100% increase or 20-50% decrease in AUC
Limited	Any interaction effect is expected to be small or mainly secondary, with minimal change to core subjective effects, for example mild shifts in physiological measures, timing of onset or offset, or a predicted 10-25% increase or 10-20% decrease in AUC
Unlikely	The interaction is not expected to meaningfully change core subjective effects or safety in typical use, even if small physiological or incidental changes occur, including predicted AUC changes within 10%

Literature labels reflect the likelihood and practical significance of a clinically meaningful interaction based on available evidence, while model-based labels reflect predicted magnitude for pharmacokinetic interactions or potent antagonism at curated psychedelic targets for pharmacodynamic interactions. These classifications may differ when evidence is indirect or when pharmacodynamic effects occur without a measurable pharmacokinetic change.