Long Term-Method Detection Levels (LT-MDL)
for the US Geological Survey, National Water Quality Laboratory

 

Background of the LT-MDL Process

The long-term method detection level (LT-MDL) is a modification and enhancement of the EPA 40CFR Part 136 definition of a method detection level (MDL) used at the National Water Quality Laboratory of the U.S. Geological Survey (USGS). The EPA MDL does not capture the total variability that occurs in a laboratory during the year with multiple analytsts, sample-handling staff, instuments, calibration curves, and conditions. The USGS extended the EPA MDL to include variability over an entire year, and therefore is commonly slightly higher than an EPA MDL.

The sample types used to calculate the LT-MDL are varied. Instead of using spiked samples to represent the blank population, analysts were asked to prevent their instruments from censoring any signal (particularly for inorganic analyses). This way, even negative concentrations could be captured and the entire distribution portrayed. This full distribution of blank results, is centered close to zero in the absence of bias. Comparing the blind blank results to the spiked blank results showed very comparable results, and it was determined that the spiked samples were unnecessary for LT-MDL determination for analyses that always provide blank signal (ICP, ICPMS, and nutrients for example).

Further, although the blind blank samples could provide a full distribution of results, both less-than and greater-than zero, using the 99th percentile of the blank population seems a more direct route than s x t to finding the value representing the 1% false positive rate. So, both the s x t and the 99th percentile are calculated on the blind blanks, to assure that we are selecting a representative LT-MDL.

In 2008, the internal set blank data were included in the assessment of LT-MDLs. In cases where blind blanks returned mostly non-detects, or the blind blank or blind spike data were suspect in some way, the inclusion of a third set of samples for comparison of LT-MDLs proved to be very useful. Since there are many more set blanks (n>>100) than blind blanks (n<=50), the occasional "zinger" within the set blank results will generally be eliminated when calculating the 99th percentile, and will not negatively affect the LT-MDL. This feature will continue to be tested during 2009 and 2010.

REFERENCES:

USGS Series	Open-File Report
Report Number	99-193
Title	New reporting procedures based on long-term method detection levels and some considerations for interpretations of water-quality data provided by the U.S. Geological Survey National Water Quality Laboratory
Edition	-
Language	ENGLISH
Author(s)	Childress, Carolyn J. Oblinger; Foreman, William T.; Connor, Brooke F.; Maloney, Thomas J.
Year	1999
Originating office
USGS Library Call Number	(200) R29o no.99-193
Physical description	iv, 19 p. ill. ;28 cm.
ISBN