Pharmaceutical Calculation Suite
Mean, standard deviation and %RSD of assay results across blender locations — with the acceptance criteria sourced honestly: established industry practice descended from a withdrawn FDA draft, not current FDA guidance.
Blend uniformity analysis assesses whether the active is homogeneously distributed in a powder blend before compression or encapsulation. The core statistics are simple — mean, standard deviation, and relative standard deviation of assay results from samples taken at defined blender locations. What requires care is the status of the acceptance criteria: the widely used limits (RSD ≤ 5.0%; individuals within 90.0–110.0%; mean within 90.0–110.0%) descend from an FDA draft guidance that was withdrawn in 2002, and are best described as established industry practice, not current FDA guidance. This page states that plainly and points to what FDA currently references.
Where this is used:
Samples are drawn from multiple predefined blender locations (stratified sampling), commonly 10 or more, at a sample size typically 1–3× the dosage-unit weight per firm procedure.
s and RSD% are different quantities with different bases: s is in % LC; RSD% is s as a percentage of the mean. They are numerically close only when the mean sits near 100% LC — never report one as the other.
Sourcing — read before citing the acceptance limits. The commonly applied criteria (blend RSD ≤ 5.0%; individuals within 90.0–110.0% of target) originate in FDA's August 1999 draft guidance ANDAs: Blend Uniformity Analysis, which FDA withdrew on 17 May 2002 (67 FR 35120), citing unresolved issues including the adequacy of blend sampling techniques. FDA's November 2003 draft, Powder Blends and Finished Dosage Units — Stratified In-Process Dosage Unit Sampling and Assessment, was in turn withdrawn on 7 August 2013 (78 FR 48175). Neither is current FDA guidance. The limits remain widely used in firms' internal procedures and in post-withdrawal frameworks in the literature. Cite them as your firm's justified internal criteria, not as FDA requirements. For stratified-sampling acceptance criteria, FDA's CGMP Questions and Answers page on production and process controls states: “Please refer to ASTM E2709 and ASTM E2810 for further guidance on establishing acceptance criteria for a stratified sampling plan.”
| Symbol | Variable | Units | Description |
|---|---|---|---|
| xᵢ | Individual assay result | % LC | Assay of the sample from blender location i, expressed as percent of label claim (or percent of target). |
| n | Number of samples | count | Commonly 10 or more, from predefined blender locations (stratified sampling). |
| x̄ | Mean | % LC | Arithmetic mean of the assay results. |
| s | Sample standard deviation | % LC | n−1 (sample) form. In the same units as the data — not a percentage of the mean. |
| RSD% | Relative standard deviation (result) | % | s expressed as a percentage of the mean. The quantity the common ≤ 5.0% criterion applies to. |
| limits | Acceptance limits | % LC | Three criteria are applied: mean range, individual range, and maximum RSD. Left blank, the mean and individual ranges both default to the commonly applied 90.0–110.0% and the RSD limit to 5.0%; enter your specification's values to override. |
Scenario: ten samples from predefined locations in a V-blender qualification run (% LC): 99.2, 101.4, 98.7, 100.5, 102.1, 99.8, 100.3, 98.9, 101.0, 100.6.
PharmaCalc returns mean, %RSD and a per-location pass/fail across all your sites. It is computed server-side against the published method and written into a GMP PDF report with the inputs, formula chain, references, document control and signature pages — traceable to the software release that produced it.
Open the Blend Uniformity calculator — free demo