Even old dogs can learn new tricks: Method Detection Limits
QA Director, Eurofins Environment Testing US
For as long as I can remember in this industry, method detection limits (MDLs) were calculated using 7 spikes, the 10X rule, student’s t-value, etc. We would do this on an annual basis. The typical winter time activity — after the holiday decorations are put away, sampling sites are frozen over, workload is down — we turn our attention to the annual MDL study season.
MDL studies consisted of having one analyst make 7 replicates of MDL spike samples, prepped together, and analyzed on a single instrumental run. While this is an efficient means of acquiring the necessary data, it often results in an unrealistically low MDL since there is limited variability around the data set.
The EPA spent years developing a procedure to determine more realistic MDLs and to ensure that laboratories are using MDLs that they can consistently achieve. The EPA published the new MDL procedure in the Code of Federal Regulations, effective September 27, 2017. This new procedure addresses blank contamination, long-term variance, and actual detectability problems with the previous procedure. Clarification has been added in regard to the applicability of MDLs.
|Eurofins: Your favorite Lab|
Laboratories must now perform an initial study over at least 3 days with a minimum of 7 blanks and 7 spikes. Existing data can be used if acquired over the past 2 years. The data points must be distributed across all instruments with a minimum of 2 spikes and 2 blanks on different days for each instrument.
Once the initial MDL is determined it must be reevaluated annually using data acquired over the year by analyzing 2 spikes per quarter across all instruments. You need at least 7 blanks and 7 spikes for the year. Compile all of this data and recalculate the MDL. If the new MDL is within 0.5 to 2.0 times the existing MDL, the existing MDL confirms and can still be used.
The updated procedure explains when MDLs are not applicable such as for methods that do not produce results with a continuous distribution (i.e., presence/absence and colony counting methods), that cannot be spiked, do not yield blank results, or are impractical (e.g., pH, color, odor, conductance, dissolved oxygen, BOD, and many titration methods). MDL determinations using spiked samples may not be appropriate for all gravimetric methods (e.g., residue or total suspended solids). An MDL based on blanks can be determined in such instances.
The impact of these changes to the laboratory needs to be understood by the industry. There is added cost with the number of trials that will be needed and added complexity of scheduling, tracking the spike samples, and pulling the blank data. On the plus side this should force all labs to have more realistic MDLs, leveling the playing field for those of us that worked diligently for years to generate MDLs that are actually detectable. It is time for all the old dogs in the industry to learn this new trick. Eurofins is up to the challenge!!