Analytical Variance: The Lab
It’s not unique to cannabis — analytical variance is the name of the game across labs of all kinds, from pharmaceuticals to animal feed. Since this is such an important aspect of lab testing, we’re diving deeper into the topic…
First of all, what do we mean when we say analytical variance?
Analytical variance is the variation in results when testing cannabis samples. It acknowledges the differences in measurements on samples run by different labs with different testing methods.
Analytical variance can be a result of multiple factors, including the equipment, the extraction, the people doing the extraction, and the variability of the sample itself (see Part 1, Analytical Variance: The Farm).
Why does it exist, and why does it matter?
Allowable variance exists in every type of manufacturing production, which means there’s an acceptable range of deviation or variation from the true value of a particular process. It’s important because it allows for some flexibility in the production process, while still ensuring the final product meets the required quality and safety standards. Manufacturers need to adjust their processes and establish appropriate control mechanisms and monitoring systems to ensure the allowable variance is within the acceptable range.
For example, in Aspirin, every single tablet won’t have the exact same amount of acetylsalicylic acid, because of the natural variations in raw materials, equipment, and processing conditions. The United States Pharmacopeia (USP), a nonprofit organization, provides guidelines for allowable variances in pharmaceutical products like Aspirin. This helps manufacturers ensure that the final product meets the required quality standards.
In cannabis, on the other hand, production isn’t federally regulated and does not yet have a state-run reference lab establishing variance requirements, so labs are left to establish procedures on their own.
We rely on reference methods developed for food and apply them to our matrix (or the type of material we’re testing like flower, concentrates or extracts, or infused products).
Each matrix can have different physical and chemical properties that affect the extraction efficiency of the analytes (being tested). For example, infused products like edibles can have different matrix interferences from flower, such as high concentrations of fat or oil, which will impact how we extract the active compounds to then quantify the results.
How does it show up in our practices in the lab?
In order to earn accreditation with ORELAP, we applied reference methods for food, adjusted and applied it to our matrices, and then ran internal validation. We have four key steps:
- Accuracy — We’re looking at how close the measurement is to the true value. Our Shelf Testing series is an example of how we use the same method to re-test a sample and analyze the relative differences to the original value.
- Precision — Similarly, we’re performing extractions multiple times on the same sample to evaluate how close the results are to one another. Ideally, results are within a small margin to each other.
- Selectivity — We must ensure the extraction and analytical method being used can accurately identify the target analytes from non-target analytes. In terms of cannabinoids, we are only required to quantify THCA, d9-THC, CBDA, CBD and d8-THC, but we confirmed our selectivity by identifying these analytes from non-target analytes such as CBN, CBC and CBDV, which we are also able to quantify accurately.
- Sensitivity — This is where we look at the smallest quantity we can quantify to determine the lowest concentration of the target analyte that can be reliably detected and measured. If we have an action limit (a specific threshold concentration of an analyte, where any result above this limit causes a failure, and the product cannot be sold for public consumption) of 1 PPM, we need to be able to quantify at or below 0.5 PPM multiple times to test our sensitivity limits.
For example, we may spike a sample with a known concentration of THC, and take the same sample through our entire extraction process. We know the true result is supposed to be 100 PPM, and we get a result of 95 PPM. We calculate the percent RPD (Relative Percent Difference) to measure the accuracy of our testing method and determine our in-house variance.
We’re doing this with four to seven samples at the same concentration. For every sample, we’re also looking at percent recovery or the percentage of the target compound that’s measured in the sample during testing. Across multiple data points, we then look at the RSD (Relative Standard Deviation) to assess the precision of our method.
What does this mean for our customers’ results and their businesses?
While we are a unique testing industry with no established industry guidelines, every lab is driving towards the same destination. The state conducts blind studies twice per year, where they give labs a sample with the same concentration. Every lab reports their results, which helps the state confirm that different methods can arrive at the same results.
We have our way of getting there, and others have their own methods. Between instruments, the extraction process, methods for analysis, staffing, etc., there are too many variables. As we’ve discussed, we can only control what happens in our lab. There’s another whole set of variables related to the product (see Part 1, Analytical Variance: The Farm). Today, the important thing is that we all end up within the state’s limits of variance.
What does it mean for the industry?
The long-term solution is for the state is to create a reference lab with set parameters and methods for cannabis labs. This exists for environmental testing labs, which does not include cannabis. We try to meet requirements that aren’t designed specifically for us, doing the best with the rules we have.
