Jump to main content.

Procedures for Detection and Quantitation

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

Meeting #10

Questions, Clarifications and Rationale

  1. What are key components of the DQFAC Single Lab DL-QL Procedure (Version 2.3) and what was learned from the Pilot Study or other potential procedures that provided input into this version of the procedure?
    1. If numerical blank results are available then the detection limit is calculated using this data. In the pilot study this type of detection limit determination was more effective for false positive protection than detection limits involving extrapolation from spikes.
    2. Spikes are used at or near the quantitation limit so that precision and accuracy data is generated at the quantitation limit. Again, it is considered more effective than extrapolating precision and accuracy at the Quant limit from spikes at other concentrations.
    3. Initial Detection and Quantitation limits determined are immediately verified and are subject to ongoing verification
  2. For sections 1.1.8 and 2.3 related to when 5% or more blanks are greater than the DL, what led to the decision to use 20 method blanks as the cut off for using the highest result?

This cutoff point was driven by a review of the data - it may be worth some additional review.

  1. For sections 1.1.8 and 2.3 related to when 5% or more blanks are greater than the DL, isn't the decision to use the next highest result if 20-100 method blanks are available an indirect way to designate a False Positive rate of 5%?
    1. No, rather it is a way to ensure that action is taken when false positive rates are above 5%. In the pilot study most false positive rates were less than 1%.
  2. For 1.2.6.1, what is the reason for including the sentence "precision and accuracy at the QL will be expected to be somewhat worse at the mid-level, so it is not appropriate to use criteria established for mid-level spikes at the QL"?
    1. To avoid regulatory agencies and others specifying that the existing precision and accuracy data in methods (generated at the mid level) be used for the QL.
  3. For ongoing verification, is there a reason for using four per year and likewise, is there a concern with the generation of enough data in the specified time period?
    1. The four per year is a minimum. It is a compromise to ensure that the cost of implementation of the new procedure is not unreasonable relative to the existing MDL procedure. Keep in mind that the verification has to be performed on all instruments.
  4. The ACIL procedure that was pilot tested used a statistical k factor for calculation. Why in the DQFAC Single Lab DL-QL Procedure (version 2.3), is a statistical t factor used for censored methods?
    1. Use of a k factor when determining the DL for censored methods would result in unnecessarily stringent criteria for choosing the appropriate QL. For example, when determining a startup DL and QL based on 7 spike results, the RSD of those replicates would need to be 12% or less, or the QL would need to be raised. Because blank data from censored data would rarely yield hits, and would never follow a normal distribution, the effect of k on false positive rates would likely be very minor for censored methods.
  5. What targets are used in the procedure for false negatives and false positives.
    1. Less than or equal to 1% false positives and less than or equal to 5% false negatives.
  6. What data from the pilot study and other data sets were used to assist the Procedure Strike Team and Technical Work Group in developing the DQFAC, Single Lab DL-QL Procedure (Version 2.3), and what did the data indicate that led to specific decisions in developing the procedure?
    1. The laboratories' evaluation of the ACIL procedure in the Pilot Study revealed some suggested changes, including the clearer distinction between censored and uncensored methods, and the greater need for ongoing verification. Existing blank and low-level spike data submitted by FLDEP, East Bay MUD and TestAmerica laboratories were used to evaluate the revised procedure, and informed the decision of k vs. t for both censored and uncensored methods.

Top of Page

Local Navigation

Jump to main content.