Method Detection Limit - Frequent Questions
The MDL procedure is designed to be a straightforward technique for estimation of the detection limit for a broad variety of physical and chemical methods. The previously promulgated version of the MDL procedure (Revision 1.11) stated: “The method detection limit (MDL) is defined as the minimum concentration of a substance that can be measured and reported with 99% confidence that the analyte concentration is greater than zero and is determined from analysis of a sample in a given matrix containing the analyte.”
The 2016 revision of the MDL procedure (Revision 2) differs from Revision 1.11 of the MDL procedure in three significant ways.
- The MDL procedure now uses method blanks to calculate an MDL, in addition to the spiked samples that have always been used to calculate the MDL. As a result, the new definition of the MDL is: "The method detection limit (MDL) is defined as the minimum measured concentration of a substance that can be reported with 99% confidence that the measured concentration is distinguishable from method blank results." The value calculated from the spiked samplesSpiked samplesare prepared from a clean reference matrix, such as reagent water, spiked with a known and consistent quantity of the analyte. is called the MDLS. The MDLS calculation is the same as the MDL calculation in Revision 1.11. The method blank samples are used to calculate the MDLb, which is a very similar calculation that also calculates the 99% confidence level that the result is derived from the sample rather from contamination/noise. The MDL is the higher of the two values (either the MDLS calculated using spiked samples or the MDLb calculated using method blanks). EPA considers this change important because as detector sensitivity improves, the background contamination of the laboratory, consumable supplies, and equipment can be more important in determining the detection limit than the sensitivity of the instrument.
- The MDL now requires that the samples used to calculate the MDL are representative of laboratory performance throughout the year, rather than on a single date.
- A laboratory has the option to pool data from multiple instruments to calculate one MDL that represents multiple instruments.
Questions and comments submitted on the 2015 proposed Methods Update Rule, and EPA's responses.
- Will the laboratory need to analyze significantly more samples than before to calculate the MDL?
No, most commercial laboratories that currently participate in analyses supporting the National Pollutant Discharge Elimination System (NPDES) calculate the MDLs for each method/instrument combination once per year.
Comparison of Number of Samples Analyzed in MDL Procedures Samples Required Revision 1.11 Revision 2 Spiked samples 7/year 8/year (2/quarter) Methods blanks (MBs) 0 0 (use routine MBs)
Additional blanks would be required for determination of the MDL only if the laboratory is implementing a new method for the first time, or has just acquired a new instrument. Otherwise, no additional method blanks are required because Revision 2 of the MDL procedure uses the routine method blanks that are already being analyzed with every batch of samples.
- Is the laboratory required to recalculate the MDL every quarter?
No, the MDL is only calculated once a year. MDL spiked samples are now analyzed every quarter in which the method is used, but the calculation is only required to be performed once a year.
- Will laboratories have to analyze more samples for methods that are rarely used?
No, Revision 2 of the MDL procedure could potentially require fewer samples than Revision 1.11 of the MDL procedure for the rarely used methods. For example, if a laboratory analyzed 7 batches of samples spread out over a 2-year period (either because they have a new instrument or because not enough analyses are conducted in a one-year period) then the laboratory would have enough sample spikes and blanks to recalculate the MDL. This would be 7 samples in 2 years, which is half of what was normally done for Revision 1.11 of the MDL procedure, if performed once a year.
- If the laboratory does not use a method during a quarter, will the laboratory still need to analyze low-level spiked samples?
No, the laboratory needs to analyze at least seven low-level spiked samples and seven method blanks for one instrument in a two-year period (spread over 3 batches), but is also supposed to analyze two spiked samples per quarter in separate batches any quarter samples are analyzed. See Sections (2)(b) and (3)(a) of the MDL procedure. A practical way for a laboratory to stay in compliance is to analyze a low-level spiked sample with the first two analytical batches every quarter. If no samples are analyzed, then there is no need to analyze spiked samples or method blanks. If one batch of samples is analyzed during a quarter, then the laboratory should include one low-level spiked sample in that batch. If two or more batches of samples are analyzed, the laboratory should include one low-level spiked sample in at least two of those batches.
- What happens if the laboratory has less than 7 sample spikes when calculating the MDL?
The minimum number of samples is 7, but more samples are used if more are available. If the analysis is performed regularly, then there will likely be 16 spiked samples per instrument (2 per quarter over 2 years) and many more blanks. If the analysis is performed very rarely, then there may be less than 7 spiked samples. In this case, the laboratory needs to perform a new initial MDL procedure, but can use the samples that are available over the last 2 years to contribute to calculating the new initial MDL. For example, if a laboratory only had 4 available spiked samples and 4 method blanks over the last 2 years, then 3 more spiked samples and 3 more method blanks would be required for determining the initial MDL. These 3 additional spiked samples and method blanks could be analyzed in one batch (assuming the existing 4 pairs of samples were in at least two batches). See Section (2)(b) for the initial MDL procedure. In this scenario, the lab would analyze 7 spiked samples in 2 years, which is less than the laboratory would have analyzed doing an MDL study every year using Revision 1.11 of the MDL procedure.
- Could one high blank result drastically elevate the MDL?
It depends. A high blank result could be eliminated in a number of ways. If there are 100 or more method blanks results, then the highest blank result is not considered (since the MDL procedure measures the 99% confidence limit). Also, a method blank result can be ignored if it is associated with a documented instance of gross failure (e.g., instrument malfunctions, mislabeled samples, cracked vials). Any data associated with rejected samples also are not used for the MDL calculation. Section (4)(d) of Revision 2 of the MDL procedure states “Include all routine data, with the exception of batches that are rejected and the associated samples reanalyzed.”
- What if a laboratory buys a new instrument and wants to include it in a multi-instrument MDL?
Provisions for a new instrument have been added to Section (3)(e) of Revision 2 of the MDL procedure. The laboratory needs to analyze a minimum of two spiked samples and two method blanks on the new instrument.
- If a laboratory uses a certain analysis daily, then they will have hundreds, potentially thousands, of method blanks to review for a two-year period.
In order to make the annual verification less labor intensive for laboratories that analyze many sample batches of certain methods, the laboratory will have the option to use only the last six months of method blank data or the fifty most recent method blanks, whichever criteria yields the greater number of method blanks.
- Cost of new MDL procedure.
The additional effort and expense is mainly in the computation, not additional sample preparation and analysis. Modern laboratories have laboratory information management systems (LIMS) that can query the data needed for the annual verification. Once the data are compiled, the MDLS calculation is identical to the calculation in Revision 1.11. The MDLb is a similar calculation, and is simplified if blanks are almost always "non-detect." The implied requirement that the calculated MDL must be no less than one-tenth the spike value in Revision 1.11 of the MDL procedure has been eliminated in Revision 2, so laboratories will not have to redo the MDL procedure if the calculated MDL is lower than expected. This will reduce workload, and is especially helpful for methods with long analyte lists. Also, the fact that the MDL samples are now analyzed throughout the year with routine samples eliminates the need to set aside a specific batch of samples and time to analyze the MDL samples.
- Why are acceptable calibrations and batch quality control (QC) not mentioned in the Initial MDL procedure?
If the laboratory is performing an initial MDL without client samples in the batch, most batch QC is not required. The spiked samples are essentially laboratory fortified blanks, and the MS/MSD samples are not required if there are no client samples in the batch. Ongoing MDL samples should be analyzed with client samples the laboratory receives, so all normal batch QC should be present. The methods already specify that calibrations must be completed before performing any analyses, so there is no need to add this requirement to the MDL procedure itself.
- Why is so much ongoing data collection necessary, and what additional quality is this practice of ongoing data collection providing?
Ongoing data collection captures instrument drift and the variation in equipment conditions throughout the year. Many laboratories currently analyze the MDL aliquots immediately after the instrument is serviced and all consumable instrument parts are new, thus yielding a best-case MDL value. Ongoing data collection leads to an MDL that represents what is actually practiced throughout the year.
- If the MDL and ML values change, permit limits may need to be reviewed.
EPA acknowledges that MDL values for some analytes may increase due to the revision of the MDL procedure. Some permits may contain limits that cannot be met by approved methods in 40 CFR Part 136. If this is the case, the permittee should use the most sensitive method allowed among the approved methods in Part 136. The NPDES rule, "Use of Sufficiently Sensitive Test Methods for Permit Applications and Reporting” (August 19, 2014) discusses this issue in detail. (Additionally, supporting documents are available in the docket at regulations.gov, docket no. EPA-HQ-OW-2009-1019.) This situation already existed with Revision 1.11 of the MDL procedure, and permittees and permitting authorities should have addressed this issue. The new MDL procedure may cause some additional contaminants to have MLs above the permit requirements for a specific analysis. The "Sufficiently Sensitive Method" rule is very clear about what to do in this case; see 40 CFR 122.21(e)(3).
- How long does a laboratory have to implement the current MDL procedure after promulgation of the Methods Update Rule?
EPA recognizes that it is not possible for any laboratory to make this change instantaneously. The laboratory should comply with the requirements of its control authority or permitting authority to implement Revision 2 of the MDL procedure.
- Definition and Procedure for the Determination of the Method Detection Limit, Revision 2 (2016)
- Methods Update Rule (Final rule - August 28, 2017)