memo 03 §4 · Consolidated screen
The interaction screen
11 medication rows, ready to run against a med list: the evidence level, the documented magnitude, and the action. Including the rows where the answer is no evidence of interaction.
Up-front verdict (memo 03 §4)
the interaction risk is real, enzyme-based, and concentrated in a small set of drugs — clobazam, warfarin, tacrolimus, CNS depressants. It is NOT a diffuse "cannabis interacts with everything" problem.
GAP-14 · there is no engine behind this table
Nothing computes, matches or validates a flag. The screen is the pharmacist reading this list against what the patient reports, and then typing what she raised. Stated in full on the procedure page.
Consolidated interaction screen
11 rows. Two of them — SSRIs/most antidepressants, and the PPI/antihypertensive/metformin class — record that no signal was found. Saying so is as important as flagging one.
| Patient medication | Evidence/level | Documented magnitude | Action |
|---|---|---|---|
| Clobazam | 3 clinical trials; strongest evidence in the corpus (Antón 2022); Epidiolex label | N-desmethylclobazam Cmax/AUC ≈ 3-fold; reciprocal 7-OH-CBD Cmax +73% / AUC +47% | Dose-adjust clobazam; monitor AEs; avoid CBD-dose escalation without review |
| Warfarin | LEVEL 1 (severe/probable; CYP2C9) — case reports only but consistent; no trials | Case: INR 7.2 after 1 month of edible + smoked cannabis on stable 10 mg/day (no bleeding) | INR monitoring around any cannabis start/stop/dose change; adjust warfarin |
| Tacrolimus | LEVEL 2 (moderate/probable; CYP3A4) | Plasma increases +358%, +200%, +77% with CBD (2 case reports + one 6-person trial) | Avoid; if unavoidable, dose-adjust and monitor levels |
| Everolimus (oral) | Epidiolex label | (label-mandated) | Lower starting dose per label |
| Buprenorphine | LEVEL 2 (Antón 2022) | — | Dose-adjust and monitor |
| CNS depressants — benzodiazepines, barbiturates, opioids, alcohol, TCAs, antihistamines, lithium | Marinol label §7.1 verbatim: "Additive CNS effects (e.g., dizziness, confusion, sedation, somnolence) may occur when MARINOL is taken concomitantly with drugs that have similar effects on the central nervous system"; Cesamet label verbatim: "Additive drowsiness and CNS depression"; Cesamet: "This combination should be avoided" (corrected 2026-09-12 — "additive drowsiness" is Cesamet wording, not Marinol) | qualitative | Warn; dose-time separation; never present as safe to combine |
| Valproate (with CBD) | Epidiolex label ALT>3×ULN incidence at 10–20 mg/kg/day | 30% (valproate+clobazam), 21% (valproate), 4% (clobazam), 3% (neither) | Baseline transaminases/bilirubin; hepatic-impairment caution; note exposures are pharmaceutical-dose (not asserted as dispensary-dose risk) |
| Strong CYP3A4/CYP2C19 inducers (rifampin class) | Epidiolex label | ↓CBD plasma ~32%; ↓7-OH-CBD ~63% | Consider CBD dose increase per label |
| CYP1A2, CYP2C8, UGT1A9, oral P-gp substrates | Epidiolex label | (label-mandated) | Consider dose reduction of the substrate |
| SSRIs / most antidepressants | No corpus evidence of clinically significant PK interaction | — | Counsel on additive sedation only; do not cite specific SSRI-PK numbers |
| PPIs, most antihypertensives, metformin class | No corpus evidence of clinically relevant interaction | — | No special action beyond standard counseling |
Where no signal was found
-
SSRIs / most antidepressants
No corpus evidence of clinically significant PK interaction
Action · Counsel on additive sedation only; do not cite specific SSRI-PK numbers
-
PPIs, most antihypertensives, metformin class
No corpus evidence of clinically relevant interaction
Action · No special action beyond standard counseling
Mechanism
THC/CBD/dronabinol metabolized via CYP2C9/CYP3A (~⅓ eliminated in urine). In vitro: THC inhibits CYP1A2/2B6/2C9/2D6; CBD inhibits CYP3A4/2B6/2C9/2D6/2E1; CBN inhibits CYP2B6/2C9/2E1 — in-vitro extrapolation requires dose context; only clobazam (among the screened set) is graded at clinical-trial certainty, while warfarin is the only Level 1 severity assignment (certainty and clinical severity are different axes — do not conflate).
Do not conflate
Certainty and clinical severity are different axes. Only clobazam, among the screened set, is graded at clinical-trial certainty; warfarin is the only Level 1 severity assignment. A strong magnitude with weak certainty and a weak magnitude with strong certainty are not the same finding and must not be reported as if they were.