The decision tree
The calculator runs four branches in order: null-benefit gate first, then dose range from goal × diet, then form selection, then metformin-screen flag.
1. Null-benefit gate (block first):
- Dietary-rda goal + omnivore + no metformin → blocked. A mixed omnivore diet meets the 2.4 µg/day adult RDA without supplementation, and Mendelian randomization (PMID 40739033) confirms null effect on cognitive performance and 8 psychiatric outcomes from raising serum B12 in genetically replete adults. The calculator redirects to lab work (serum B12 + MMA) before recommending dose. If the user is vegan / vegetarian, on metformin ≥4 y, or has any other deficiency-context, they pick a different goal and proceed.
2. Daily dose range (by goal):
- Dietary gap-filling: 250–500 µg/day for vegan / vegetarian
- Metformin prevention: 500–1000 µg/day
- Repletion (PA, atrophic gastritis, post-bariatric): 1000–2000 µg/day oral
- Diabetic peripheral neuropathy (DPN): 1500 µg/day mecobalamin (split TID)
- Pregnancy (vegan / vegetarian): 50–250 µg/day, continued through lactation
The repletion range comes from the 2024 prospective cohort (Lacombe et al., PMID 38797248) — oral 1000 µg/day cyanocobalamin reversed pernicious anemia in 88.5% of patients at 1 month and 100% by 12 months, even without intrinsic factor. The Cochrane review (PMID 29543316) confirms oral and IM are equivalent for serum B12 normalization at low-quality but consistent evidence.
3. Form (cyanocobalamin vs methylcobalamin vs hydroxocobalamin):
Cyanocobalamin is first-line for non-DPN dosing. The Cochrane review found no clinically meaningful difference vs methylcobalamin for repletion, and cyanocobalamin is cheapest with the longest shelf life. The calculator forces methylcobalamin (mecobalamin) on the DPN goal because the peripheral-neuropathy meta-analysis (Sawangjit 2020, PMID 32716261) is mecobalamin-specific — the evidence base does not transfer cleanly to other cobalamin forms. Hydroxocobalamin is reserved for chronic kidney disease, Leber's hereditary optic neuropathy (where cyanocobalamin is contraindicated due to theoretical cyanide load on the optic nerve), and as a longer-half-life IM option in clinical settings.
Transdermal patches are never recommended — molecule too large for meaningful skin absorption, and severe deficiency on patch reliance is documented in case literature.
4. Metformin-screen flag (always, when applicable):
ADA 2026 standards recommend annual B12 + MMA + CBC screening after 4 years on metformin. The 2026 Diabetologia GWAS (PMID 41537778) identified CUBN rs1801222 as a pharmacogenomic modifier — AA homozygotes hit 12.84% deficiency on long-term metformin vs 6.02% in GG, and reach the 10% deficiency threshold in 11 years vs 21. The calculator surfaces the annual-screen recommendation regardless of goal whenever metformin exposure is ≥4 years, with a 6-month interval for known CUBN AA carriers.
What the calculator does NOT cover
- ALS ultra-high-dose protocol (50 mg IM 2×/week mecobalamin) — neurologist-only territory, Japan-approved indication; outside self-serve dosing scope.
- Acute SACD with neurological signs — numbness, ataxia, cognitive change in suspected B12 deficiency requires urgent clinical evaluation with IM loading, not oral self-supplementation.
- N₂O abuse / functional deficiency on normal serum B12 — emerging public health signal requires medical evaluation; the calculator does not attempt to dose for this scenario.
- CKD with severe renal impairment (GFR <30) — interpretation of B12 / MMA labs is altered; defer to nephrology.
Where the math stops being safe to extrapolate
The "more is better" intuition fails for B12. The vault's distinctive position — and what the calculator enforces — is to supplement to repletion, not past it. The MR null on cognition and 8 psychiatric disorders is the strongest evidence in the field that B12 is a deficiency-correction tool, not a performance-enhancement tool. The calculator builds the user-facing refusal directly into the goal × diet logic so the wrong answer cannot be rendered.
The metformin-screen recommendation is a screening intervention, not a treatment one. If labs already show low serum B12 / elevated MMA, the user should switch goals to "repletion" rather than "metformin-prevention" — the prevention dose (500–1000 µg/d) is conservative for active deficiency.