Everything has limits. We like to tell new graduates that the sky’s the limit. For industrial hygiene and environmental laboratories, the limit of concern is not how high we can go, but how low. What’s the smallest speck of sand (or smallest amount of workplace contaminant) we can see with that spectrometer or chromatograph? And how reliably can we see that speck? The reliability needs to be sufficient for the field hygienist to have confidence in the decisions she is making. While the concept is simple, actually figuring out the threshold of reliability is a bit more challenging.
Analytical detection limits are developed by statisticians, applied by analytical laboratories, and used by policy makers, regulators, and lawyers. The varied purposes of these limits and the wide variety of terms used to describe them add complexity for the hygienist trying to protect workers. By going back to basic principles, we can cut through some of the complexity and make the concept of analytical limits more understandable.
Attempts to develop definitions related to analytical detection capability, and formulas for calculating various limits, go back as far as Bernard Altshuler and Bernard Pasternack’s paper “Statistical Measures of the Lower Limit of Detection of a Radioactivity Counter,” which was published in
Health Physics
in 1963. Perhaps the best-known author on the subject is Lloyd Currie, now retired from the National Institute of Standards and Technology, who published a seminal paper in
Pure and Analytical Chemistry
in 1968 and later led the development of the 1995 International Union of Pure and Applied Chemistry’s “Nomenclature in Evaluation of Analytical Methods Including Detection and Quantification Capabilities.” Currie’s work focused on the statistical derivations of these terms. The statistics are necessary to determine the level of confidence in the numerical value of the limit being calculated.

The IUPAC nomenclature was later incorporated into the ISO 11843 series of standards, which is widely followed in Europe. In the United States, many terms, definitions, and statistical derivations have been developed to meet different needs. Different methodologies for determining analytical limits can lead to different results. This article makes no attempt to define all of the terms related to these limits; it focuses instead on the underlying concepts, which are quite similar.
IH practitioners have a hierarchy of controls. Similarly, laboratories have what could be called a hierarchy of limits, which is summarized in Table 1. These limits are listed in increasing numerical order—that is, IDL < MDL < MQL < RL.
Analytical detection limits are developed by statisticians, applied by analytical laboratories, and used by policy makers, regulators, and lawyers.
Table 1.
Hierarchy of Analytical Limits
Tap on the table to open a larger version in your browser.
Detection and Quantitation Limits
Manufacturers of laboratory instruments need a figure of merit for the ability of their instruments to distinguish an analyte of interest from background. The Wisconsin Department of Natural Resources defines
instrument detection limi
t (IDL) as “the concentration equivalent to a signal, due to the analyte of interest, that can be distinguished from background noise by a particular instrument.” The primary purpose of the IDL is to compare various lab instruments; it does not consider other aspects of a measurement method such as sampling, sample preparation, matrix effects, and laboratory reagents. The IDL will always be lower than the
method detection limit
(MDL), which does take these things into account.
One of the best-known and most widely followed methodologies for determining the MDL is given in 40 CFR 136, Appendix B. EPA published a new definition and procedure for the MDL in December 2016. An update to 40 CFR 136, which the agency worked on for more than a decade, was expected to be published in October. (At the time this article was written, only the pre-publication version of the update to 40 CFR 136 was available.) Table 2 compares the revised EPA definition of MDL, the original EPA definition, and the definition used by AIHA-Laboratory Accreditation Programs, LLC (see
Policy Module 9
Table 2.
Definitions of Method Detection Limit
Tap on the table to open a larger version in your browser.
One of the key changes in EPA’s definition is that it describes the MDL as the value where there is 99 percent confidence that the concentration is “distinguishable from method blank results,” not “greater than zero.” This change answers one of the criticisms of the earlier definition: it did not consider that method blanks might have a nonzero value.
The MDL is used to determine the ability of an analytical method (not just the instrument as in the IDL)
to detect the presence
of an analyte of interest with a given level of confidence, which can be anywhere from 90 percent to 99 percent for a given laboratory and method (both EPA and AIHA-LAP, LLC define it at 99 percent). The MDL does not give 99 percent confidence in the quantification of the analyte. Typically, the MDL is calculated as 3.3 standard deviations above the mean blank signal, as described in Currie’s 1968 article. Procedures can vary as to how many blanks should be run to determine the MDL. For the initial MDL determination, the new EPA method calls for seven spiked samples and seven method blanks. Other procedures call for anywhere from five to 20 blanks. What is important is that the laboratory uses a defensible procedure, which is a requirement for laboratories that are accredited by AIHA-LAP, LLC and other ISO/IEC 17025 accrediting bodies.
What, then, is the difference between the MDL and the MQL? Basically, it is the difference between detecting the presence of an analyte of interest and quantifying the amount of that analyte—with a stated level of confidence. Typically, and again going back to Currie’s 1968 publication, the MQL is taken to be ten standard deviations above the mean blank signal (or a factor of three standard deviations above the MDL). Some scientists have questioned Currie’s factor of ten, but the key point is that the lowest value at which a method can reliably
an analyte is higher than the value at which it can reliably
that analyte.
It is important to note that MDLs and MQLs are not only specific to a laboratory, but may be specific to an instrument within a laboratory. Published MDLs and MQLs (or other similarly defined limits) can be found in governmental methods (for example, OSHA, NIOSH, and EPA) or in consensus standards such as ASTM test methods. These values are based on studies within a given agency or on interlaboratory studies. However, the MDLs at a given laboratory may be lower or higher than the MDLs published by government agencies.
MDLs and MQLs are not only specific to a laboratory, but may be specific to an instrument within a laboratory.
Laboratory Reporting Limit
Typically, it is the reporting limit (RL), rather than the MDL or MQL, that laboratories will make available to customers. AIHA-LAP, LLC defines “reporting limit” as
the lowest concentration of analyte in a sample that can be reported with a defined, reproducible level of certainty. This value is based on the low standard used for instrument calibration.
The reporting limit can be equal to the MQL, or a value above the MQL that takes into account variations over time or variations that may occur in instrument performance. AIHA-LAP, LLC requires that laboratories establish the reporting limit through the analysis of media spiked samples taken through the entire analytical process. The laboratory must also state the acceptance criteria for these samples and verify the reportable level annually, or whenever the method or instrumentation is changed, through the analysis of media spiked samples. The important point is that performance at the reporting limit has been established and verified according to a laboratory-developed procedure, and that this procedure is available to the IH. Laboratories are generally required to repeat MDL studies annually, and thus the MDL could fluctuate. Some labs may set the RL at a value that is less apt to fluctuate with time.
AIHA-LAP, LLC policy also requires that laboratories report measurements below the RL as “less than” or “not detected” and reference the RL. The reporting of zero concentration is not permitted. The IH Laboratory Accreditation Program (IHLAP) requires a documented process for “defining, establishing, verifying, and reporting of minimum reporting limits.” Some customers believe that when laboratories do not report numerical results below the RL, they are withholding valid information from the client. On the contrary, the laboratory is protecting its accreditation status and its reputation. Laboratories will naturally avoid setting RLs too high because doing so hurts their competitive position. (There may be some legitimate uses for data below the RL, but that is beyond the scope of this article.)
ASTM Committee D22 on Air Quality, recognizing the ongoing confusion over the terminology and statistical derivations of various detection limits, has organized a series of conferences on the subject. The first of these conferences was held in August 2016; the second will be held in Washington, D.C., in October 2018. Committee D22 is also developing a consensus standard for determining the detection limit of a well-behaved method. A proper understanding of the analytical hierarchy of limits can help the IH practitioner understand the information that the laboratory provides, have confidence in the results, and make the right decisions for protecting workers.
is a fellow technical advisor at Savannah River National Laboratory in Aiken, S.C. He is vice chairman of the AIHA Sampling and Laboratory Analysis Committee and a member of the AIHA-LAP, LLC Technical Advisory Panel. He can be reached at (803) 952-4400 or via
is quality control coordinator for Wisconsin Occupational Health Laboratory, University of Wisconsin- Madison. He is vice chairman of the AIHA-PAT, LLC Board and a member of the Sampling and Laboratory Analysis Committee. He can be reached via
. Acknowledgement: The authors gratefully acknowledge the support and assistance of Ryan LeBouf of NIOSH, past chair of the AIHA Sampling and Laboratory Analysis Committee.

AIHA Laboratory Accreditation Programs, LLC:
LAP Policies and Guidelines
Analytical Chemistry
: “Limits for Qualitative Detection and Quantitative Determination” (March 1968).
Chemometry Consultancy:
References on Limit of Detection
EPA: Definition and Procedure for the Determination of the Method Detection Limit, Revision 2 (
, December 2016).
European Federation of Clinical Chemistry and Laboratory Medicine: “Limit of Detection, Limit of Quantification and Limit of Blank,” presentation by Elvar Theodorrson (
, October 2015).
Pure and Applied Chemistry
: “Nomenclature in Evaluation of Analytical Methods Including Detection and Quantification Capabilities (IUPAC Recommendations 1995)” (
, October 1995).
Wisconsin Department of Natural Resources:
Laboratory Certification and Registration Program

Defining Laboratory Analytical Limits