Background Document from Technical Work Group

November 30, 2005

During the course of discussions of the Technical Work Group (TWG) and
subgroups, questions were raised that could impact the work of the TWG,
but that the group considered to be fundamental policy issues.  These
policy issues could potentially impact how pilot studies are designed,
how reporting limits (LQ, LD, or LC) are set, and how compliance is
evaluated.  The TWG requested this document be sent to the federal
advisory committee as background information.  The TWG is providing
commentary on the following issues:

Issue 1: Descriptive vs. Prescriptive

There are two fundamentally different approaches to how reporting limits
are set and used.  The two approaches are described as
“Prescriptive” and “Descriptive.”  Depending on which approach
is used, the work of the TWG is fundamentally shifted.  

One proposed method is to examine both approaches in parallel and
present the results as such. However this would be a very large time and
cost intensive approach.  If the FACDQ were to decide on only one of
these approaches, the TWG would be able to focus its work more
effectively.

The Descriptive Approach – an approach where laboratory data is used
to describe that lab’s LQ, LD, and LC for that lab.  It is then up to
a client or agency to determine whether these values are acceptable.

Each laboratory determines their own specific LQ, LD, and/or LC by means
of a statistical procedure set in regulation for each analyte and
analytical method. 

These descriptive estimates of LQ, LD, and LC change over time for a
specific laboratory.  The determination of the LQ, LD, and/or LC may be
periodic (e.g. once per year) or on-going (like a running average).

If any of these estimate were selected to act as the Compliance
Evaluation Threshold (CET), the CET would change depending on which
laboratory is analyzing the samples and the CET are reassessed over time
at a particular laboratory and vary between laboratories.  

This is how the MDL is currently being used in CWA monitoring.  

Using a descriptive approach would favor conducting a single-laboratory
pilot study design.  Because it is up to a single laboratory to describe
its own LQ, LD, and LC and demonstrate these meet reporting limits.

The Prescriptive Approach -- an approach where LQ, LD, and / or LC is
set in regulation and is the same for a given regulated analyte for all
laboratories.  It is important to note, the laboratory may not be
required to perform a formal LQ, LD, LC study for this approach; simply
validating their procedures and equipment meet the requirements being
prescribed.

These values would not change over time or between laboratories.

If LQ, LD, and/or LC are used as a CET, the CET is the same for all
permit holders irrespective which laboratory or instrumentation they use
or changes over time.  

This is the approach used in the Information Collection Rule and the
Unregulated Chemical Monitoring Rule and the California Drinking Water
primacy program. 

As outlined in B) Issue 2, the laboratory would on an ongoing basis
analyze a validation sample to demonstrate compliance.  That sample will
require a range of limits.

Using a prescriptive approach would probably favor a multi-laboratory
pilot study design as a regulator would require data from many
laboratories to sufficiently define and prescribe reporting limits for
the laboratories under its jurisdiction.

Combined Approach – It is possible to combine both approaches.  For
example, The California CWA primacy program has a mixture of these two
approaches.  The LQ is set prescriptively and the LC is set
descriptively.  Other permutations are also possible.

Issue 2: Verification of LQ, LD, and LC.   

A second policy question that impacts the work of the TWG is: Do LQ, LD,
and LC need to be verified?  If a laboratory describes estimates of LQ,
LD and/or LC or if the laboratory has a prescribed LQ, LC, or LD; should
there be procedures in place to verify that the laboratory can actually
measure analytes at those levels in the way expected. If verification is
required then the TWG can recommend how to do this.  There are two
alternatives:

The individual laboratory does not verify the LC, LD, or LQ it is using
(whether descriptive or prescriptive) actually meets the definitions
used (e.g. a false positive rate of 1% for unspiked blanks and a 50%
false negative rate for spiked blanks at LC).  As with the current
Appendix B descriptive approach, the laboratory runs and keeps on file a
copy of the MDL studies it performed at a snapshot in time.  There is no
requirement in Appendix B to verify this value.

Each individual laboratory verifies LQ, LD, and/or LC by some
standardized procedure with performance characteristics. This may be
periodic or batch specific. 

Using a descriptive approach, a laboratory may do a LC or LD
determination (e.g. an MDL study).  Then the laboratory would have run a
series of unspiked and spiked blanks to determine the false positive and
false negative rates and see if they match the expected rates.

Using a descriptive approach, a commonly used technique is to analyze
check solutions with a concentration at or below the LQ which must be
recovered within certain accuracy limits.  Blanks must not have higher
concentrations than LD or LC.  LC or LD must be no more than a fixed
fraction of LQ.  Laboratories must pass on a batch by batch basis these
QC requirements or the data may not be submitted.  

Verification for a descriptive approach is generally very complex and
costly while verification under a prescriptive approach is a great deal
simpler and less expensive.  

Using a prescriptive approach, the laboratory simply analyzes a
validation sample at or below the LQ, which must meet recovery
requirements set out by method or procedure.

Verification procedures have more commonly used historically with the
prescriptive rather than descriptive approach.

C) Issue 3: Lowest possible LQ, LD, or LC

Do LQ, LD, and LC need to be the “absolute lowest possible” for all
analytes in all situations?  

Determining the lowest possible LQ, LD, or LC requires a great deal more
effort and cost.  This is true irrespective of whether this lowest
possible value is set descriptively or prescriptively.

Verification of LQ, LD, and LC is much more complicated the lower the
concentration is.  Also the verification techniques will be different
depending on whether one is seeking the lowest possible value in every
case.

The lowest possible LQ, LD, and LC would probably favor using a
descriptive approach and a single-laboratory study design. 

Study design would be much simpler if LD and/or LC were only needed for
analytes with very low WQBELs.

WORKING DRAFT FOR DISCUSSION 

Federal Advisory Committee on Detection and Quantitation Approaches and
Uses in Clean Water Act Programs

 PAGE   

 PAGE   1 

Technical Work Group

DRAFT FOR DISCUSSION

11/30/05

