I just got some tests done at a lab I hadn’t used in a while because I need up-to-date labs for my site.
This lab reported all of my edibles at 20% to 64% short of target D8 contents.
All of my recipes call for between 11% to 19% more D8 than the target potency before cooking / evap. We use 96% D8 from a very reputable vendor. We already fine-tuned these recipes to hit target potency with analytics from other labs long ago.
What boggles my mind most is the variability in results… If I’m calculating recipe qty’s virtually the same way for every SKU – as tightly as 11% to 19% and always OVER target – how in the world am I getting results across the board from 20% to 64% short? Even if we assume everything is short, I can’t imagine anything that would explain that variability.
I asked the lab to re-test. A week passed with no follow-up. I called to ask for an update – lab mgr says they tested basically the same. Ok… little suspect that they never sent the re-test results.
Guess I’m just feeling buyer’s remorse. Not the first time we’ve gotten mathematically impossible COA’s on edibles. Probably shouldn’t have sent ~20 samples to a lab I hadn’t used before.
Anyone else have the same issues with COA’s for edibles?
guessing they simply don’t have a clue how to get cannabinoids out of gummies…and nobody has called them on it (yet).
I would suggest that asking them how they did on spike recovery from your matrix?
unfortunately, that will likely go right over their heads…doesn’t mean you shouldn’t ask.
insisting that they’ve got it wrong, and pointing out how much state regulators dislike labs without a clue (yeah, you’ll have to exaggerate on that one) might bring joy, but chances are you’re just gonna have to tell them fuck and go elsewhere.
They take your matrix & spike with cannabinoid standard.
Then see how much they get back.
Mission critical if lab is clearly missing cannabinoids
YOU know what you put in there.
You’re pretty damn sure they’re only recovering a fraction (less than 50% in some cases).
Having them prove they can(t) recover 100% from your matrix usually involves $$, but a reasonable lab director will waive that if it’s clear they are failing miserably (good luck on finding one of those)
Testing recovery from spiking a known amount into a matrix is pretty common for any quantitative analytical method validation. Of course you will never create a universal matrix, every single gummy recipe out there is going to most likely have some slight differences, but for all intents and purposes they should have performed a spike and recovery analysis from a generic gelatin and a generic pectin gummy recipe.
I don’t think it would be unfair for you to ask to see their method validation.
As is asking for that validation on any specific matrix when it’s clear the numbers don’t add up.
In OP’s case, it would appear they are not even close…
Which is where “ok, now try MY matrix…” comes into play
Asking a lab to attempt spike recovery from MY matrix is something I’ve only had to do once. Turned out the “MCT” we were using was NOT MCT. their “MCT method” did not in fact recover more than 80%.
We reformulated once we figured out the primary issue.
I agree with @cyclopath here as a great step forward as well. If you want to give the lab the benefit of the doubt, then approach them in an amicable way and suggest you provide them with a blank formulation of your gummy for them to do two things with:
run the blank as-is, which may allow them to find any potential interferences where THC should be eluting on the chromatogram.
perform a spiked recovery with the exact matrix being used
Of course, if they never did a proper method validation or are unable to cooperate in this fashion, maybe that isn’t the lab to use going forward.
a spiked canna standard on top of a gummy sample will likely show perfectly good recovery bc the cannas are not embedded in the matrix
In this scenario the (possibly inaccurate, likely inexperienced) testing lab will have more data backing up they performed the extraction sufficiently > which they did not
Id guess these are water soluble gummies that should be extracted in water to dissolve the gummy , then diluted in an organic solvent to move the sample cannas into solution. Skipping the water extraction will give very low recovery on sample cannas
I have used Septer Analytics in NM for years. Might as well plug them, we go 10% over in all our recipes numbers come back on point. Labeling your samples well always helps. Makes for better COAs too. As I write this, the free info in this thread alone for an upcoming gummy tycoon is pretty valuable.
If we want to explore homogeneity, and determine whether the current method of homogenizing is or isn’t doing enough, I would suggest 10 individual samples from the same batch and request an individual potency test on each one for 10 individual potency assay’s. This is very common in pharma for ensuring accurate dosage between individual pills of the same batch, often called Content Uniformity.
It is obvious after these results that there is certainly interlaboratory variability but what we don’t know yet is if that variability is due to the labs themselves, or the product. I’m not going to say the mixing is or isn’t sufficient, because I don’t know, but before I would go to the labs and start questioning the difference in the data, I might suggest running this type of content uniformity test at one or even multiple labs if you have not done this type of study before. At the very least it is another data point for you to examine.
This will at least let you get an idea of what your single batch variability is between individual gummies or individual “doses” of powder. If you do this test with multiple labs I would not compare the labs between themselves for this. This test would solely be to determine variability in individual doses of your product and help you understand if you are a potential source of the varied results or if in fact the labs are all wildly different.