Help with RSD

[Fernando]

Hi everybody!

Can you help me with this: I have been determining Vitamin D3 in a tablet that contains 400 IU/1.6 g . This is a very small quantity about 6 ppm.
My question is How I consider the maximun RSD between determinations for example if I obtain: 370, 365 and 395 IU/tablet ?

Thanks for any help.

[Ann]

RSD or %CV is determined as (SD/mean value)*100

The SD of your 3 values is 16.07 to 2 d.p
Your mean value is 376.67 to 2 d.p.

Therefore, your %CV is (16.07/376.67)*100=4.27%

(However, people generally use more repetitions than n=3 (perhaps 5-7))

[DR]

(However, people generally use more repetitions than n=3 (perhaps 5-7))

If you're doing dosage uniformity, n=10.
If you're doing assay, n=1 or whatever your method says (A injections for each of B sample preps...) - for USP, anyway.

[Fernando]

Hi Ann

Do you think 4,27 % or 10 % as RSD it is too much or if not what guided you to say so.

Thanks

[Ann]

Hi Fernando

As DR said, it does depend on what you're doing. If, for example, you were trying to determine the the precision of your HPLC system by injecting 7 times from one autosampler vial (containing a standard solution), you would hope that the CV would be less than 1%. However, if you were injecting once only from 7 different vials (e.g. to test the precision of a dilution process), you would expect the CV to be slightly higher. If your analysis is more complicated (such as quantifying a drug extracted from a whole blood sample), then a CV of up to 15% may be acceptable (when injecting once from 7 extracted samples).

Hope that helps!

Ann

[Fernando]

Hi Ann, Dr

Thats the problem, I am quantifying 10 µg of Vitamin D3 in a tablet that weights 1620 mg! When the time comes I must justify the precision of the data obtained. So if I put that my RSD between determinations is 15 % what would I say?

Thank you Ann, Dr

[DR]

I would say that 5% RSD is pretty good for having a target that is such a low percentage of your total tablet weight. If your standard %RSD is low and anyone complains about the RSD of the vit. D3, have them take it up with whoever is in charge of formulation &/or production. 10µg/1.6g=0.0006% - this is not easy to do with high precision and reproducibility in solid dosage forms.

[Fernando]

Hi Dr

Did you know of a reference to justify that 5 % of RSD? I am thinking of the Horwitz formula or something like that. And about the tablet we cannot do anything because it its in the market in this concentration!

[DR]

If the tablet is on the market, there should be some specs for the tests you are doing. If you are doing dosage uniformity, the USP typically allows 6% RSD for stage 1, provided all results are between 85-115% of claim. If you are doing a potency assay (also known as composite or stability indicating assay), the USP monograph would provide only a specification with respect to claim (eg: 97-103%) and you would have to refer to your method for internal %RSD limits.

I'm not sure I'd argue the point using Horowitz. You are dealing with a, presumably, validated method and have some history with the sample type you are dealing with.

I would investigate how your samples look compared to previous runs. Are the areas and %RSDs comparable? Is the %RSD for replicate sample preps usually around 5% for Vit. D3? Is this sample any different from the previous samples (longer time on some harsh stability study condition)?

If your %RSDs are typical, I would leave it at that. If they are higher than normal, I would try to determine if this is an expected or an unexpected result. If the standards are unusually high with respect to %RSD, check the equipment. . .

[Fernando]

Hi Dr

Searching some downloads from internet I found a paper from Dr Huber published in the LCGC magazine in 1999 and in the item Precision and Reproducibility refers to a table from the AOAC that states as reference for an analyte concentration of 10 ppm the corresponding RSD is 7.3 %.
What do you think about it, doyou agree?

Thanks everybody.

[DR]

This sounds more "defensible" than the Horowitz paper.

[Alex Buske]

7.3% IS the Horwitz-Limit for 10 ppm for method precision.
Are you sure you would like to argue with that?
In the end YOU have to assay within certain limits and to determine CUT (or Uniformity of dosage Units). With a RSD of 5%, how many assay results would be out of 95-105 %, even if the real content would be 100%?
You should search for the reasons for your low precision. inject at least six (or ten) standard solutions, six (or ten) times a sample solution and six (or ten) different sample solutions. Then you might see the problem: system precision, (bad integration or poor selectivity), (poor sample solution preparation or inhomogenious CUT).
For 252I.E./g vit D3 higher precision is definitely achievable.
If you can't boost the precision, increase the number of determinations, at least for the assay.

Alex

[DR]

It is? I didn't notice that.

I only glossed over the paper and realized that, while it makes a good general point, the applicability of the ranges for UV detection will hold true for some compounds, but not others. It is all a function of your analyte's absorbtivity at the wavelength you monitor. For compounds with weak or average extinction coefficients, 7.3%RSD may be right at 10ppm, but if you're looking for something like benzaldehyde, you should be disappointed with >1%RSD at 1ppm (assuming you have a decent autosampler). Again & as you said, you have to consider your history with the product and samples. If your standard RSDs are historically a lot better than your sample RSDs, there is probably room for improvement in your production process.

[Fernando]

Hi All!

I´ll do more samples and I will be back to you with the results.

Thanks