AccuReg: Blend Uniformity Sampling and Testing

The Regulatory Clinic - January 1997
Blend Uniformity Sampling and Testing

Dear Clinic:

At my company, there are many heated opinions regarding blend uniformity OOS results and what the appropriate sampling (resampling) plan should be. Two ideas have been bounced around:

Sample the drums at the designated locations in singlate. If an OOS result is generated and there is an identifiable laboratory cause, resample in singlate at the same designated locations. If there is no identifiable cause, then resample at the same designated locations in duplicate and test under increased supervision.
OR
Sample the drums at the designated locations in duplicate to begin with. If an OOS result is generated, then test the second sample that was pulled at the same time as the first (OOS) sample.

The discussion revolves around whether a sample pulled perhaps up to two days later constitutes the same sample. My personal problem with Option 2 is that the investigation can only compare one result vs. another result. If the two sets of results do not compare, how do I give more credibility to one set over the other? Could you please help with this problem?

Signed, OOS & OOA (Out-of-Agreement) in Baltimore, MD

Dear "OOA":

As you know, the purpose of blend uniformity testing is to determine/demonstrate that you have achieved the ultimate blend (i.e., the active ingredient(s) is/are homogenously dispersed). Keep in mind that acceptance criteria should generally be a bit tighter for blends than for finished assay units, since blends will be further manipulated and may segregate slightly more prior to compaction, due to downloading into drums, loading into hoppers, then into feed frames, etc. Also, as the blend stands, segregation may occur simply due to vibrations in the building (caused by earth tremors, trains, etc.), although the degree of segregation depends on the building and the blend's properties. Whatever you do, I strongly recommend that you do not resample!

Instead of using either of the approaches you described, consider this: your sampling and testing scheme should be statistically based much like the USP's Uniformity of Dosage Units <905> procedure, which calls for an S1 level and an S2 level. The S1 level involves the individual assay of 10 randomly-selected units. The allowable range of results is 85.0-115.0% of labeled claim, with an RSD of not more than 6.0%. If these criteria are not met, then 20 additional units are randomly selected and assayed. If no more than 1 unit in 30 is outside the 85.0-115.0% range, but inside a 75.0-125.0% range, and the RSD of all 30 assays is not more than 7.8%, the batch is considered uniform.

For mixers, I suggest you sample the blend at more than 10 spots (e.g., 20 spots). Assay 10 representative spots and set criteria much like the USP Content Uniformity. For downloaded drums, I suggest taking nine samples -- three from the top, three from the middle, and three from the bottom. Test three -- one from the top, middle and bottom, respectively. The criteria should be 90.0-110.0% of the labeled claim and RSD not more than 5.0%. If one of the three samples is outside this range but is within 85.0-115%, and/or the RSD is exceeded (while the other two results meet the criteria), then test the six additional samples and set criteria for all nine units to 90.0-110.0% of labeled claims, with no more than one unit in nine outside of 90.0-110.0% (but inside 85.0-115.0%), and the RSD not more than 6.0%.

If you test based on this sampling scheme, you can be much more confident that your results accurately reflect the homogeneity of your blend, while avoiding the problems of resampling and comparison of individual results.

Robert V. Sarrio, Senior Associate
AccuReg, Inc.

Return to The Regulatory Forum
Write to the Clinic at accureg@regulatory.com

7501 Northwest 4 Street, Suite 210
Plantation, Florida 33317, USA
Telephone: 954-641-6400 Fax: 954-641-6410
Email: accureg@regulatory.com

Contact webmaster