Relative Standard Deviation (RSD) Check

Overview

The Check for Relative Standard Deviation (RSD) evaluates if the precision between more than two values exceeds Quality Control (QC) limits.

Notes:

•For the %RSD between recovery values (e.g. between the recoveries of a matrix spike, matrix spike duplicate, and matrix spike triplicate) the QC_RSD column or the result_value columns for all samples must be populated in DT_RESULT_QC for this check to function.

•For the %RSD between sample results (e.g. between the parent sample, field duplicate, and field triplicate) the QC_RSD column or the QC_SPIKE_RECOVERY and/or QC_DUP_SPIKE_RECOVERY columns for all samples must be populated in DT_RESULT_QC for this check to function.

•Control limits for this check are determined as described in the RT_DQM_CONTROL_LIMITS section.

Check

Check name is EarthSoft.DQM.RelativeStandardDeviation= Relative Standard Deviation (RSD).

Parameters

This check uses four Global Parameters:

1.msd_parent_sample_ms – See Global Parameters for additional information.

2.reportable_result – DT_RESULT.REPORTABLE_RESULT values to include in DQM datasets.

3.result_type_code – DT_RESULT.RESULT_TYPE_CODE to include in DQM datasets.

4.test_type_batch – AT_TEST_BATCH_ASSIGN.TEST_BATCH_TYPE to include in DQM dataset.

Twenty-four parameters are available for this check. None of these parameters allow for input of analyte-specific or method-specific PARAM_VALUEs. The descriptions for the parameters below are written as if there are triplicate values to compare; however, the Relative Standard Deviation check can be run on any number of values >2.

1.Multiplier – Multiplier used with detection limit in the sample grouping for field-collected triplicate samples (e.g. field sample, field duplicate, and field triplicate) or laboratory-created triplicate samples (e.g. field sample, laboratory duplicate, laboratory triplicate) to test for results with concentrations > multiplier * detection limit. Defaults to 5.

2.flag_QC – TRUE/FALSE for flagging QC samples; QC sample types are configured in the sample_type_LCSD and sample_type_MSD parameters.

Result type:

3.Result_type_LCSD – The result type for the Laboratory Control Sample Duplicates and Laboratory Control Sample Triplicates. Defaults to "SC".

4.Result_type_MSD – The result type for the Matrix Spike Duplicates and Matrix Spike Triplicates. Defaults to "SC".

5.Result_type_lab_dup – The result type for the Laboratory Duplicate and Laboratory Triplicates. Defaults to "TRG".

6.Result_type_field_dup -The result type for the Field Duplicate and Field Triplicates. Defaults to "TRG".

Sample type:

7.Sample_type_LCSD – The sample types for Laboratory Control Sample Duplicates and Laboratory Control Sample Triplicates. Defaults to "BD"; additional SAMPLE_TYPE_CODEs must be defined in RT_SAMPLE_TYPE and added to the listing in the PARAM_VALUE for this parameter in the DQM QAPP (e.g. BT, BQ).

8.Sample_type_MSD – The sample type for Matrix Spike Duplicates and Matrix Spike Triplicates. Defaults to "MSD"; additional SAMPLE_TYPE_CODEs must be defined in RT_SAMPLE_TYPE and added to the listing in the PARAM_VALUE for this parameter in the DQM QAPP (e.g. MST, MSQ).

9.Sample_type_lab_dup – The sample type for the Laboratory Duplicates and Laboratory Triplicates. Defaults to "LR"; additional SAMPLE_TYPE_CODEs must be defined in RT_SAMPLE_TYPE and added to the listing in the PARAM_VALUE for this parameter in the DQM QAPP (e.g. LT, LQ).

10.Sample_type_field_dup – The sample type for the Field Duplicate and Field Triplicates. Defaults to "LR"; additional SAMPLE_TYPE_CODEs must be defined in RT_SAMPLE_TYPE and added to the listing in the PARAM_VALUE for this parameter in the DQM QAPP (e.g. FT, FQ).

Compare to:

11.Compare_to_calculated_LCSD – Y/N option to compare calculated RSD to reported RSD for grouping of Laboratory Control samples; defaults to “N”.

12.Compare_to_calculated_MSD – Y/N option to compare calculated RSD to reported RSD for grouping of Matrix Spike samples; defaults to “N”.

13.Compare_to_reported_lab_dup – Y/N option to compare calculated RSD to reported RSD for grouping of Laboratory Duplicate samples; defaults to “N”.

14.Compare_to_reported_field_dup – Y/N option to compare calculated RSD to reported RSD for group of Field Duplicate samples; defaults to “N”.

Acceptable variance:

15.Acceptable_variance_LCSD – acceptable variance when comparing calculated RSD to reported RSD for grouping of Laboratory Control Samples (%); defaults to 1.

16.Acceptable_variance_MSD – acceptable variance when comparing calculated RSD to reported RSD for the grouping of Matrix Spike Samples (%); defaults to 1.

17.Acceptable_variance_lab_dup – acceptable variance when comparing calculated RSD to reported RSD for the grouping of Laboratory Duplicate samples (%); defaults to 1.

18.Acceptable_variance_field_dup – acceptable variance when comparing calculated RSD to reported RSD for the group of Field Duplicate samples (%); defaults to 1.

Reported before calculated:

19.Reported_before_calculated_LCSD – Y/N option to try reported RSD before the calculated RSD when running RSD check for the grouping of Laboratory Control Samples; defaults to “N”.

20.Reported_before_calculated _MSD – Y/N option to try reported RSD before the calculated RSD when running RSD check for the grouping of Matrix Spike Samples; defaults to “N”.

21.Reported_before_calculated _lab_dup – Y/N option to try reported RSD before the calculated RSD when running RSD check for the grouping of Laboratory Duplicate samples; defaults to “N”.

22.Reported_before_calculated _field_dup – Y/N option to try reported RSD before the calculated RSD when running RSD check for the grouping of Field Duplicate Samples; defaults to “Y”.

Detection Limits:

23.Lab_dup_detection_limit – the detection limit column to use in the calculation of the %RSD between the grouping of Laboratory Duplicate sample results. Defaults to “first_non_null_value”.

24.Field_dup_detection_limit – the detection limit column to use in the calculation of the %RSD between the grouping of Field Duplicate sample results. Defaults to “first_non_null_value”.

Additional Notes for Parameters

For the grouping of Laboratory Control Samples and Matrix Spike Samples

1.If compare_to_calculated = "Y" or reported_before_calculated = "N" then

1.DQM will calculate the RSD for every result using qc_spike_recovery and qc_dup_spike_recovery columns.

2.DQM will also calculate the RSD if the qc_rsd column is null.

2.If compare_to_calculated = "Y" then

1.DQM will compare the calculated RSD to the reported value in the qc_rsd column.

2.DQM will create an Error in the Errors tab if the difference between the reported and calculated RSD values exceeds acceptable_variance.

3.If reported_before_calculated = "N" then

1.DQM will use the DQM-calculated value to compare to the Control Limit

2.If QC_SPIKE_RECOVERY or QC_DUP_SPIKE_RECOVERY is NULL or Not Numeric, but QC_RSD is populated (with a numeric value), DQM will use the reported value to compare to the Control Limit.

3.If the Control Limit is exceeded, an Exception will be added to the Exceptions table.

For the grouping of Laboratory Duplicate Samples and Field Duplicate Samples

1.If compare_to_reported = Y or reported_before_calculated = "N" then

1.DQM will calculate the RSD for every result using DT_RESULT.RESULT_NUMERIC for all samples in the grouping.

2.DQM will also calculate the RSD if the qc_rsd column is null.

2.If compare_to_reported = Y then

1.DQM will compare the calculated RSD to the reported value in the qc_rsd column.

2.DQM will create an Error in the Errors tab if the difference between the reported and calculated RSD values exceeds acceptable_variance.

3.If reported_before_calculated = "N" then

1.DQM will use the DQM-calculated value to compare to the Control Limit

2.If the Control Limit is exceeded, an Exception will be added to the Exceptions table

4.If reported_before_calculated = "Y" then

1.If one of the results in the laboratory duplicate grouping or field duplicate grouping is not-detected, DQM will create an Error in the Errors tab indicating that one of the results in the grouping is not-detected and DQM will not evaluate the %RSD for the grouping.

2.If one of the results in the laboratory duplicate grouping or field duplicate grouping is <multiplier*detection_limit, DQM will create an Error in the Errors tab indicating that one of the results in the grouping is <multiplier*detection_limit and DQM will not evaluate the %RSD for the grouping.

3.If all the results in the laboratory duplicate grouping or field duplicate grouping is <multiplier*detection_limit, DQM will not evaluate the %RSD and will not create an error.

4.If qc_rsd is null or not numeric, DQM will use the DQM-calculated value to compare to the Control Limit.

5.If the Control Limit is exceeded, an Exception will be added to the Exceptions table.

Notes:

•If running a dataset for both the RPD check and the RSD check, it is suggested that different sample_type_codes be used for the RPD check than the RSD check as DQM will evaluate all sample_types listed for the check.

•Default parameter values listed above refer to the value provided in the DQM Starter QAPP.xlsx.

Rules

There are four rules:

1.LCS RSD >CL – The percent relative standard deviation between the recoveries for the grouping of Laboratory Control Samples exceeds the control limit.

2.MS RSD >CL – The percent relative standard deviation between the recoveries for the grouping of Matrix Spike Samples exceeds the control limit.

3.Lab Rep RSD > CL – The percent relative standard deviation between the results for the grouping of Laboratory Duplicate Samples exceeds the control limit.

4.Field Rep RSD > CL – The percent relative standard deviation between the results for the grouping of Field Duplicate Samples exceeds the control limit.

Exceptions (where table is not indicated, it is DT_RESULT_QC)

For Laboratory Control Sample grouping:

•Review results with DT_SAMPLE.SAMPLE_TYPE_CODE = sample_type_LCSD and DT_RESULT.RESULT_TYPE_CODE = result_type_LCSD.

•Qualify a result if rsd > rsd_cl.

•Qualify associated results related by TEST_BATCH_ID and result_type_code in Global Parameters.

•Use the LCSD qualifier for results with detected concentrations and the LCSD NON_DETECT qualifier for non-detect results from RT_DQM_RULE.

For Matrix Spike Sample grouping:

•Review results with DT_SAMPLE.SAMPLE_TYPE_CODE = sample_type_MSD and DT_RESULT.RESULT_TYPE_CODE = result_type_MSD.

•Qualify a result if rsd > rsd_cl.

•Qualify associated parent sample results. The parent sample is found based on the msd_parent_sample_ms global parameter:

oWhen msd_parent_sample_ms = “Y”: Qualify associated results related by SYS_SAMPLE_CODE in the PARENT_SAMPLE_CODE field of the MS's parent sample. For example:

▪MSD sample's SYS_SAMPLE_CODE = ABC_MSD and PARENT_SAMPLE_CODE = ABC_MS.

▪MSD's parent sample's SYS_SAMPLE_CODE = ABC_MS and PARENT_SAMPLE_CODE = ABC.

oWhen msd_parent_sample_ms = “N”: Qualify associated results related by SYS_SAMPLE_CODE in the PARENT_SAMPLE_CODE field. For example:

▪MSD sample's SYS_SAMPLE_CODE = ABC_MSD and PARENT_SAMPLE_CODE = ABC.

▪MS sample's SYS_SAMPLE_CODE = ABC_MS and PARENT_SAMPLE_CODE = ABC.

•Use the MSD qualifier for results with detected concentrations and the MSD NON DETECT qualifier for non-detect results from RT_DQM_RULE.

For Laboratory Duplicate Sample grouping:

•Review results with DT_SAMPLE.SAMPLE_TYPE_CODE = sample_type_lab_dup and DT_RESULT.RESULT_TYPE_CODE = result_type_lab_dup.

•Qualify a result if rsd > rsd_cl.

•Qualify associated results related by TEST_BATCH_ID and result_type_code in Global Parameters.

•Use the Lab Rep qualifier for results with detected concentrations and the Lab Rep NON DETECT qualifier for non-detect results from RT_DQM_RULE.

For Field Duplicate Sample grouping:

•Review results with DT_SAMPLE.SAMPLE_TYPE_CODE = sample_type_field_dup and DT_RESULT.RESULT_TYPE_CODE = result_type_field_dup.

•Qualify a result if rsd > rsd_cl.

•Qualify associated results related by SYS_SAMPLE_CODE in the PARENT_SAMPLE_CODE field. For the RSD Check, the PARENT_SAMPLE_CODE for all samples in the grouping should be the unspiked “N” sample. All field-collected samples in the grouping will be qualified.

•Use the Field Rep qualifier for results with detected concentrations and the Field Rep NON DETECT qualifier for non-detect results from RT_DQM_RULE.

An example remark saved to dt_result.dqm_remark is: “[91.978 > 50] RSD(calc) > ld_rsd control limit.”

Calculations (where table is not indicated, it is DT_RESULT_QC)

RT_DQM_CONTROL_LIMITS is discussed in more detail here.

For the RSD between recoveries for the Laboratory Control Sample grouping and the Matrix Spike Sample grouping or between results for the Laboratory Duplicate Sample grouping and the Field Duplicate Sample grouping:

•If QC_RSD is Numeric,

•then rsd = QC_RSD

•else %rsd = 100 * (Square root of variance / mean of measurements)

For Laboratory Control Sample grouping:

•If RT_DQM_CONTROL_LIMITS.LCS_RSD is Numeric,

•then rsd_cl = RT_DQM_CONTROL_LIMITS.LCS_RSD

•else rsd_cl = QC_RSD_CL.

For Matrix Spike Sample grouping:

•If RT_DQM_CONTROL_LIMITS.MS_RSD is Numeric,

•then rsd_cl = RT_DQM_CONTROL_LIMITS.MS_RSD

•else rsd_cl = QC_RSD_CL.

For Lab Duplicate Sample grouping:

•If RT_DQM_CONTROL_LIMITS.LD_RSD is Numeric,

•then rsd_cl = RT_DQM_CONTROL_LIMITS.LD_RSD

•else rsd_cl = QC_RSD_CL.

For Field Duplicate Sample grouping:

•If RT_DQM_CONTROL_LIMITS.FIELD_RSD is Numeric,

•then rsd_cl = RT_DQM_CONTROL_LIMITS.FIELD_RSD

•else rsd_cl = QC_RSD_CL.

Errors (where table is not indicated, it is DT_RESULT_QC)

For Laboratory Control Sample groupings and Matrix Spike Sample groupings:

•If QC_RSD is NULL or Not Numeric, or qc_spike_recovery and qc_dup_spike_recovery are also NULL or Not Numeric, then add error and skip result.

•If the Control Limit is NULL or Not Numeric in both rt_dqm_control_limits and dt_result_qc, then add error and skip result.

•If compare_to_calculated = “Y”, DQM will create an Error in the Errors tab if the difference exceeds acceptable_variance.

An example error message is:

The Spike Recovery data used to calculate RSD is missing from DT_RESULT_QC at sys_sample_code = AQ_SAMPLE_2_FIELD_QUAD and analytic_method= SM2540D and cas_rn=TSS.

For Laboratory Duplicate Sample groupings and Field Duplicate Sample groupings:

•If DT_RESULT.RESULT_NUMERIC is NULL or Not Numeric when DETECT_FLAG = ‘Y’ for one of the samples in the laboratory duplicate pair, then add error and skip result.

•If one of the results in the grouping is not-detected, then add an error and skip result. (If all results in the grouping are non-detect, then skip the result without an error.)

•If one of the results in the grouping is <multiplier*detection_limit, then add an error and skip result. (if all results in the grouping are <multiplier*detection_limit, then skip the result without an error).

•If compare_to_reported = “Y” DQM will create an Error in the Errors tab if the difference exceeds acceptable_variance.

Example error messages are:

Some Replicates have detected results while other Replicate results are ND. for Field Replicate RSD at sys_sample_code='AQ_SAMPLE_2_PARENT' and analytic_method='SW8260' and cas_rn='108-88-3'

The variance between the reported RSD (1) and the calculated RSD (19.861) exceeded the Acceptable Variance (2). for Field Replicate RSD at sys_sample_code='AQ_SAMPLE_2_PARENT' and analytic_method='SW8260' and cas_rn='100-41-4'