Combined Interpretations of the 2003, 2009, and 2016 Standards that apply to Volume 1 of the 2016 TNI Standard

MODULE 4: CHEMISTRY TECHNICAL REQUIREMENTS
Section: 1.5.2

Question:  Our laboratory was recently cited with "The laboratory has not established procedures to relate LOD's with Limits of Quantitation (LOQ's) (EPA 200.8, 300.0, 6020, 9056)." I must be missing something here, or perhaps an expectation is being created by this standard, which is a favorite deficiency for auditors.  The LOD (which is not clearly defined in the NELAC standard) is interpreted by our company to be equivalent to the MDL.  The MDL is derived from an MDL study following the guidance in 40CFR136. The LOQ is interpreted by our company to be lowest concentration that can be reliably and accurately reported, and it is required that it be encompassed in the calibration curve.  Typically the low point of the calibration curve is equal to the LOQ when adjusted for sample preparation factors. In this context, the LOD and the LOQ are numbers that are derived independently, and the only requirement is that the LOD be equal to or less than the LOQ.  However, auditors are implying that there should be a fixed factor (or range, such as 1-10) separating the LOD and the LOQ.  In our laboratory, metals by ICP-MS have ratios of LOQ/LOD that range from 1.4 to 100.  In my 20+ years in environmental laboratories, I have seen LOQ/LOD ratios that are all over the place. Is the standard giving me permission to adjust MDLs upward, provided that they are supported by MDL studies, so that the LOD is within a fixed factor of the LOQ?  This would be convenient, because the LOD would not change from year to year.  Our clients would be very appreciative of invariable LODs.  The question is, is this something that NELAC supports? I would like to see some clarification on this standard, because the way it is written, it seems to create the perception that we are missing something.

TNI Response:  This clause was deleted in the 2009 standard.  The SIR is obsolete.

Question:  The standard allows a lab to forgo the MDL determination if they do not report outside the calibration range unless the method requires it. Since the determination of the MDL and the LDR are pointless exercises for labs that do not report outside their calibration range, I believe that the MUR of 3/12/07 allows a lab to delete these requirements from method requirements since it does not change the chemistry, it does not change the determinative step, and it does not change the performance of the method. Agree? Disagree?

TNI Response:  The Standard has been modified and this is no longer applicable. The 2009 standard allows laboratories to forego the LOD study if data are only reported to the LOQ.  The 2016 standard reversed this and all labs must now perform LOD studies.  Also, the 2012 Method Update Rule that added section 136.7 to 40 CFR also requires labs to perform MDL studies.

Question:  It is felt that the LOD validation procedure in the 2003 NELAC Standard is ambiguous and can result in two different interpretations. By using the relevant standards (C.3.1.b, D.1.2.1.a) as well as definitions in the glossary especially for terms such as a quality system matrix, you can construe two different procedures.  One interpretation is that the LOD must be determined only in the matrix of the sample. In other words, if a lab is analyzing wastewater effluent samples, the LOD must be validated only in a wastewater effluent matrix. Not only is this not practical but not possible for many analytes.
This is a challenge to the practical and second interpretation which allows for the LOD to be validated in a reagent water matrix.  As someone who is engaged in quality assurance work, whenever an alternative interpretation is brought to me, I must evaluate objectively all viewpoints and I feel there is merit to the alternative argument. With respect to the two choices, we like to hear from you as to which choice is right and as stated we like to alert you that they may be an ambiguity issue with the LOD procedure.

TNI Response:  C.3.1 b) only states that the LOD must be verified on each instrument used for reporting data. It doesn't place a limit on when that must happen. Therefore, if one of the replicates analyzed for the MDL determination meets the applicable criteria for LOD verification, that would meet the intent of the Standard. The value of the spike used to calculate the MDL must be at no more than 2-3x the calculated MDL. It is expected that all of the replicates in such a study would show a recovery to be used to calculate the MDL.  The language is virtually identical in the 2009 standard, but this section has been extensively revised in 2016.  The SIR will be obsolete when the 2016 standard is implemented.

Question:  The Standard states you must use the "established test method acceptance criteria" or "client data quality objectives for accuracy" when confirming the LOQ.  First of all, our clients virtually never set 'data quality objectives for accuracy', they rely on the lab to set that for the methods.  Second, only new methods set acceptance criteria at the LOQ.  What criteria should be used, for instance, for EPA 8000-series tests, which only designate an 85-115% CCV criteria at mid-level for most methods, but usually mandate no specific criteria for LCSs?  While using our lab generated LCS criteria may work, those numbers are usually developed using mid-level spiking levels.  Many new methods state CCVs must be within 60-140% at the LOQ, while LCSs may be within 50-150% of expected values.  Mid-level criteria are usually tighter.  Last, there's a problem with respect to the statement in TNI Standard (section 1.5.2.2), "The annual LOQ verification is not required if the LOD was determined or verified annually on that instrument."  I have seen several LOD/MDL studies that have good precision (produces a relatively low LOD), but the accuracy is terrible (average of 250% recovery).  In my opinion, being able to physically see something at extreme low levels doesn't mean you can accurately determine the concentrations at low levels.  Some range should be set like 50-150% at the LOQ for confirmation.

TNI Response:  It is up to the laboratory to determine these criteria if they are not specified in the method or by other rule or regulation. The language is virtually identical in the 2009 standard, but this section has been extensively revised in 2016.  The SIR will be obsolete when the 2016 standard is implemented.  Webcast training on LOD and LOQ is now available on the TNI website under the Educational Delivery System tab.

Question:  The description of the LOQ does not specify a requirement that the standard used for verification of the LOQ include all sample processing steps (like extractions and digestions). The language presented for the LOD verification clearly states that all sample processing steps must be included in the LOD verification process.  Should all sample processing steps be included in the analysis of the LOQ standard verification?

TNI Response:  C.3.2 states 'The laboratory shall determine the LOQ for each analyte of concern according to a defined, documented procedure. This implies that the procedure used for samples and LOQ is by the same defined, documented procedure.  The language is virtually identical in the 2009 standard, but this section has been extensively revised in 2016.  The SIR will be obsolete when the 2016 standard is implemented.

Question:  Are there parameters/analytes for which TNI does not require MDLs/LOD?  For example, residue(TSS, TDS, TS), BOD.....

TNI Response:  C.3.1 c) states in part 'An LOD study is not required for any component for which spiking solutions or quality control samples are not available such as temperature, or, when test results are not to be reported to the LOD (versus the limit of quantitation or working range of instrument calibration)...' The NELAC Standard doesn't require any LOD determinations unless data are reported below the LOQ, AND a spiking solution or quality control sample is available. The proposed revisions to the TNI standard provide additional clarity to this subject.  The language is virtually identical in the 2009 standard, but this section has been extensively revised in 2016.  The SIR will be obsolete when the 2016 standard is implemented.

Question:  Routine samples for TKN are analyzed daily with an automated distillation titration apparatus.  Typical results are 15 to 30 ppm based on a 100 mL digestion.  MDL studies of LFB samples @ 2.5 ppm yield results too little variation for valid LOD/LOQ values.  However, this lab never sees samples at this low in the range.  Do the sections cited above relieve the lab from performing any further MDL's?  Can the laboratory state it will not report any values lower than say 5 ppm or some other conservative value?

TNI Response:  C.3.1 c) states in part 'An LOD study is not required for any component for which spiking solutions or quality control samples are not available such as temperature, or, when test results are not to be reported to the LOD (versus the limit of quantitation or working range of instrument calibration)...' The NELAC Standard doesn't require any LOD determinations unless data are reported below the LOQ, AND a spiking solution or quality control sample is available. If the laboratory SOP or QA Manual states that this analysis is not reported below the LOQ, then an LOD is not required by the Standard. If more stringent standards or requirements are included in a mandated test method or by regulation, the laboratory shall demonstrate that such requirements are met.  The language is virtually identical in the 2009 standard, but this section has been extensively revised in 2016.  The SIR will be obsolete when the 2016 standard is implemented.

Question:  For our lab, MDL is the LOD. Once an acceptable LOD is established via MDL, if the spike used meets the LOD spike concentration criteria (2-3X the LOD of the single analyte), is it necessary to prep and analyze another sample, or can one of the replicates analyzed for the MDL determination itself can be considered as a verification of the LOD?

TNI Response:  C.3.1 b) only states that the LOD must be verified on each instrument used for reporting data. It doesn't place a limit on when that must happen. Therefore, if one of the replicates analyzed for the MDL determination meets the applicable criteria for LOD verification, that would meet the intent of the Standard. The value of the spike used to calculate the MDL must be at no more than 2-3x the calculated MDL. It is expected that all of the replicates in such a study would show a recovery to be used to calculate the MDL. The language is virtually identical in the 2009 standard, but this section has been extensively revised in 2016.  The SIR will be obsolete when the 2016 standard is implemented.

Question:  In reviewing an onsite report for metals analyses, the auditor noted that the lab did perform LOD analyses annually but that the LODs for vanadium and arsenic were lower than the blank analysis and that new LODs and LOQs must be reestablished.  I see nothing in D1.2.1.b that makes any comparison of an LOD to the blank results.  Does an LOD need to be higher than the blank?

TNI Response:  There is no requirement to compare the LOD with the blank. Theoretically, a blank should have no detectable contaminants. Any detection in a blank must be evaluated according to D.1.1.1d).  The language is virtually identical in the 2009 standard, but this section has been extensively revised in 2016.  The SIR will be obsolete when the 2016 standard is implemented.

Question:  Is it acceptable to demonstrate the accuracy and precision at the Limit of Quantitation (LOQ) using a concentration that is less than the LOQ, including down to ½ the LOQ?

TNI Response:  "The validity of the LOQ shall be confirmed by successful analysis of a QC sample containing the analytes of concern in each quality system matrix 1-2 times the claimed LOQ." The standard requires concentrations that are 1-2 times the claimed LOQ. A concentration below the LOQ may not have the same accuracy and precision and therefore could not be used to verify the LOQ. The language is virtually identical in the 2009 standard, but this section has been extensively revised in 2016.  However, this SIR may still be relevant.

Question:  Section 1.5.2.1 Limit of Detection (LOD) in its opening sentence states: "If the laboratory is not reporting a value below the Limit of Quantitation, a Limit of Detection study is not required."  Does this opening statement supersede, nullify, lessen and/or somehow impact on the requirements found in 1.5.2.1 d), e) and f)? To paraphrase, if you do not report a value below the Limit of Quantitation does a laboratory still have to initially determine the LOD, perform an LOD each time there is a change in the method or the instrument and/or verify the LOD annually.  There is a need for clarity the standard and obtain an standards interpretation of this section of the standard as it pertains to the above question.

TNI Response:  If the laboratory is not reporting a value below the Limit of Quantitation, a Limit of Detection study is not required, unless specified by the reference method. This statement supersedes any requirements that follow in Section 1.5.2.1.  The 2009 standard allows laboratories to forego the LOD study if data are only reported to the LOQ.  The 2016 standard reversed this and all labs must now perform LOD studies.  Also, the 2012 Method Update Rule that added section 136.7 to 40 CFR also requires labs to perform MDL studies.