Potency Factor

Potency Factor (PF) is the dimensionless scalar that lets a Master Manufacturing Record written for a reference potency be executed against real lots whose certificates of analysis report different measured potencies. The arithmetic is straightforward — PF = reference potency ÷ measured potency, evaluated on a like-for-like basis — but PF is more than the arithmetic. It is the named lot attribute, persistent across the lot lifecycle, that drives every assay-adjusted dispense the lot is used in. A lot's PF can be updated as the lot ages, as it is retested, as repacks happen, or as in-house assay supersedes the supplier CoA. Every dispense cites the PF that was active at the moment the kiosk computed the adjusted target.

A lot assaying at 98.5% against a 100% reference has PF = 1.0152 — meaning the operator must weigh 1.52% more mass than the nominal target to deliver the same active quantity. A lot assaying at 102.0% has PF = 0.9804 — meaning the operator weighs 1.96% less. In both cases the finished batch lands at label claim; without PF, the batch under- or over-delivers silently.

PF is only meaningful when reference and measured potency are on the same basis. There are five conventions in regulated practice; the MMR must specify which one applies to each active, and the lot's CoA must report on the same basis.

A lot's PF is established at one of four moments and can be re-established as the lot ages.

1. Supplier CoA import — the most common case. CoA arrives with shipment, the assay value is entered (or auto-parsed from a PDF / CSV / supplier API), PF = reference ÷ measured is computed automatically and stored on the lot record. Reference is taken from the item master.
2. In-house assay — required under 21 CFR 211.84(d)(2) for at least one identity test per supplier per lot; required for assay if the supplier is not qualified under §211.84(d)(3). The lab result overrides any CoA-derived PF; the override is recorded under change control with the lab's signature.
3. Pharmacopeial reference — when no CoA exists and no in-house assay can be run (e.g. compounding from a USP-grade powder whose certificate is a pharmacopeial general specification), PF defaults to 1.0 with a deviation note. This is rare and disfavoured.
4. Re-assay during stability or retest — if the lot is held past its expiry / retest date and an extension assay is performed, the new assay supersedes; PF is updated; the prior PF remains in lot history. Every dispense that used the old PF is unaffected (the record is immutable) but new dispenses use the new PF.

PF rarely acts alone. For most APIs the dispense correction is a composite — PF for assay variability, (1 − LOD)⁻¹ for moisture, and SBF for salt-vs-base — applied in a specific order so rounding doesn't compound.

1. Normalise salt-to-base (SBF) first — if applicable. Brings the target into salt-mass terms when the label claim is the free base.
2. Apply anhydrous-basis convert ((1 − LOD)⁻¹) next — brings the target from anhydrous mass to as-is mass.
3. Apply assay PF last — the final correction for batch-to-batch potency.
4. Recompute the counter-balance excipient — total batch mass holds.

For a botanical or marker-compound active, the marker assay typically already accounts for moisture (because the marker % is itself on a defined basis declared in the monograph), so LOD compensation may not apply. For a salt-form active sold on the as-is anhydrous basis, only SBF and assay PF apply. The MMR must declare which factors apply per item; the engine applies them in the canonical order; the operator never re-types the math.

For botanicals, PF is the most variable and the most operationally important factor. Marker-compound assays (USP <2030>, USP <2750>) report a single marker concentration that stands as a quantitative handle for an otherwise complex matrix. The marker target in the MMR is the planned percentage; the lot marker percentage from the CoA drives PF. Lots from different harvests, different geographies, different ages, and different extraction methods can produce marker drift of 10–20% lot-to-lot. Ignoring that drift produces silent sub-potency in the finished product.

21 CFR 111.70(b)(1) requires the manufacturer to establish identity, purity, strength, and composition specifications for each dietary ingredient — including the marker compound and its target range. 21 CFR 111.75(b) requires verification that the finished batch meets product specifications, which closes the loop. PF is the operational bridge: it converts the MMR's marker-target into the actual mass of botanical to dispense for this particular lot.

1. PF stored on item, not on lot — every lot inherits the same factor, batch-to-batch variability is invisible. Architectural failure mode; needs schema fix, not a procedure fix.
2. PF averaged across multi-lot dispense — when two lots are blended, the engine averages PF instead of applying each lot's PF to its own portion. Silent error.
3. PF computed against the wrong reference — item master shows 100% reference but the formula was actually validated against a 98% reference (legacy MMR). PF math is right; the target it's correcting toward is wrong.
4. Basis mismatch between reference and measured — CoA reports anhydrous, item reference is as-is, no LOD compensation applied. PF off by the moisture fraction.
5. Salt-form confusion — label claim is free base, formula and dispense weigh the salt, PF computed against salt-form assay. Need SBF; PF alone is not enough.
6. PF override without change control — supervisor overrides the computed PF at the kiosk because the operator 'doesn't trust the CoA'; no documented basis. Part 11 audit failure.
7. PF not re-baselined after re-assay — extension assay run, new assay value stored on the lot, but PF still showing CoA-receipt value. Engine uses stale PF.
8. Botanical PF set to 1.0 because marker assay was 'expected' — operator skipped the recalculation step because they assumed marker would meet target. Silent sub-potency.

• PF currency % — fraction of active lots whose PF has been confirmed (CoA-import or in-house assay) within the last X days (target ≥99%).
• PF spread across lots — variance of PF across active lots of the same item; high spread signals supplier inconsistency and a possible supplier-development action.
• PF override rate — fraction of dispense events where operator/supervisor overrode the computed PF (target ≤1%; every override is a deviation).
• In-house vs CoA agreement — fraction of in-house assays that agree with supplier CoA within tolerance (target ≥95%; disagreement triggers supplier scorecard hit).
• Re-assay coverage % — fraction of lots held past initial retest date that have had an extension assay (target 100%; missing = lot must not dispense).
• Botanical marker drift — average % drift between lot marker and item target over a rolling window; trend signals harvest variability or supplier change.
• PF-driven deviation rate — number of deviations opened because computed PF fell outside the formula's allowed window (e.g. PF outside 0.90–1.10).

1. Item master holds reference potency, basis (anhydrous / as-is / marker / activity), salt form, free-base MW, salt MW, marker compound, expected marker %, PF tolerance window (e.g. 0.90–1.10), and designated counter-balance partner.
2. Lot master holds assay %, LOD %, salt-form override, marker %, activity unit value, CoA reference, in-house assay reference, PF (computed), and full PF change history with reason codes.
3. CoA import (supplier portal upload, CSV, or API) auto-populates assay, LOD, and computes PF — the lot is not released for use until QA reviews and approves.
4. In-house assay (lab module) can override CoA-derived PF; override is captured under change control with lab signature and reason.
5. Kiosk dispense screen shows the active PF, its source ('CoA receipt 2026-04-12' or 'in-house re-assay 2026-05-22'), and the adjusted dispense target.
6. Out-of-tolerance PF (outside the item's allowed window) blocks dispense until QA reviews and explicitly releases or rejects the lot.
7. Multi-lot dispense applies each lot's own PF to its own portion — no averaging.
8. BMR/BPR records the PF used on each dispense and the CoA / in-house assay reference, so a regulator can reconstruct the math years later.
9. PF currency dashboard surfaces lots whose PF is stale (CoA older than threshold) or whose assay is approaching retest expiry.
10. Botanical-aware: items flagged as botanicals get marker-compound fields, marker-drift trending, and stricter PF tolerance defaults.