1. Gregory M. Zarus Agency for Toxic Substances and Disease Registry
Data Quality
Gregory M. Zarus
Agency for Toxic Substances and Disease Registry

2. Are you in the fog when it comes to sampling data?

3. Does looking at lab data make you want to see an eye doctor?
ND j
B i N E
R D QL DL
SQL MDL IDL Q 91
Does looking at lab data make you want to see an eye doctor?

4. I want to fix that… I want to help you to become exposure detectives
I want to fix that… I want to help you to become exposure detectives. I want you to…
get the data quality you need
better interpret the data you get

5. What I’ll give you A decoder list for lab and field data
A list of things to look for when collecting or looking at field data
A list of things to look for when looking at lab data

6. Why EI is different
We are not the FDA: we’re not looking at the potential food supply and making judgment weather or not it should be eaten.
We are not the EPA: we’re not looking at the pollution in the environment and seeing if it poses a potential risk.
We are ATSDR: we’re looking for what the people are being exposed to.

7. Once upon a time..
I went to collect land crabs to evaluate exposures…..
using an efficient trap.

8. and caught a lot of crabs.

9. But this is the size that the locals ate…

10. Because the people used home-made traps that
limited the size of the crab caught.

11. Moral of the story Collect the right samples
to make the right exposure call.

12. So what did I do with the big crabs?

13. Visualize … A health call needs to be made about whether local crabs are safe to eat.
You get lab data on crabs and fish and you have to make the health call.
An exposure detective needs to review the sampling plan before collection (to ensure appropriate data is collected for the call).
An exposure detective who is given the data post facto needs to ask the right questions (to prevent false conclusions).

14. A good exposure detective has to check if the data is good enough for evaluating exposures.
You can be an exposure detective by doing 3 things.

15. Three things you need to do (The ARQ Method)
1. Check if there is adequate data to evaluate public health impact. (A)
2. Assess if the sampling plan represents the exposures. (R)
3. Understand standard quality assurance and control procedures. (Q)

16. Data Adequacy Is there enough data to make a health impact call?
Have samples been collected…
…in the right areas (on- and off-site)?
…with the right media?
…or analyzed for the right contaminants?
…in a manner to evaluate exposure?

17. Data Adequacy Have samples been collected in the appropriate areas on-site?
With likely contamination
Where occupational exposures are possible
With public access
Where contaminants may or may not be migrating

18. Data Adequacy Have samples been collected in the appropriate areas off-site?
…where the public can be exposed
…of current or past contaminant off-site migration
…with sensitive populations
…during the right times

19. Data Adequacy Have samples been collected in the appropriate media?
Ambient air
Indoor air
Biota
Indoor dust
Sediment
Outdoor PM
Surface water
Groundwater
Surface soil
Subsurface soil
Soil gas

20. Data Adequacy Have samples been collected/analyzed for appropriate contaminants?
Site processes and history
Contaminant break-down products
Sample preparation, collection, preservation, and analysis method(s)
Look for:

21. Representativeness of Sampling
SAMPLING PLANS should include:
Purpose and objectives
Type(s) of media to be sampled
Location and number of samples
Sampling equipment and method description
Quality Assurance and Quality Control (QA/QC)
Analytic method
Limitations

22. Representativeness of Sampling (continued)
SAMPLING REPORTS should include or reference
The sampling plan
Reason for changes in the field (if)
Times/dates of sampling and analysis
Analytic data summary and raw data
Evaluation of field and laboratory QA/QC

23. Representativeness of Sampling (continued)
The PURPOSE AND OBJECTIVES of the sampling plan will influence:
Sampling approach
Analytic approach
QA/QC requirements
If results represent “worst case,” typical, etc. levels of contaminants

24. Representativeness of Sampling (continued)
The sampling, analysis, and QA/QC will influence:
Detection limits
Specificity/interferences
Reproducibility

25. Representativeness of Sampling –Sampling Approach (continued)
Air sampling
Real-time
Time-weighted
Grab/instantaneous
Combination
Soil sampling
Discrete
Composite

26. Representativeness of Sampling (continued)
Biota
Whole fish
Edible portion only

27. Representativeness of Sampling Analysis
Air sampling
Screening
Field analysis
NIOSH & EPA methods
NOTE: Agency sampling method  agency analytical method!!!
Soil sampling
Screening
NIOSH, EPA
Biota
FDA, USDA, EPA
Contaminant vs. metabolite
Modified method……

28. Quality Assurance/Quality Control QA/QC
Were samples
-contaminated in the field?
-stored/preserved properly?
-analyzed within an appropriate time frame?
-analyzed appropriately by the lab?

29. QA/QC QA/QC Sampling QA/QC Field, lot, or trip blank Chain of custody
Field duplicate or split samples (blind)
Field spike samples
Sample preservation, storage, and transport

30. QA/QC (continued) Laboratory QA/QC: Chain of custody Holding time
Matrix spikes, matrix spike duplicates
(recoveries are good)
Lab blank
Duplicate injections
Instrument calibration and sample dilution

31. QA/QC (continued) Field sampling meets requirements
No contaminants in field, trip, or lot blank
Chain of custody held
Samples received in good condition at lab

32. QA/QC (continued) Analyses meet requirements if
No contaminants in lab blank
Chain of custody held
Matrix spikes are within method-specified percent recovery range
Field duplicate samples show similar results (air is not the same)

33. QA/QC (continued) Analyses meet requirements if
Samples are extracted/analyzed within holding time
Contaminant concentrations within calibration range of instrument

34. Common QA/QC Terms (a gift to you for decoding lab data)
Detection Limits
Instrument (IDL)
Method (MDL)
Contract required (CRDL)
Sample quantitation limit (SQL)
Instrument – Lowest amount of substance that can be detected by the instrument.; does not consider any other effects such as sample handling, matrix, preparation.
Method – Takes into account the reagents, sample matrix, and preparation steps applied to a sample in specific analytical methods.
CRDL – Are not necessarily the lowest detectable levels achievable, but rather are levels that a CLP laboratory would routinely and reliably detect and quantitate in a variety of sample matrices.
SQL – Takes into account sample characteristics, sample preparation, and analytical preparation (dilution, use of a smaller aliquot) which would affect the quantitation limit.
Your key

35. Common QA/QC Terms (continued)
Data qualifiers
U- analyzed for but not detected (ND)
J - value is estimated (j)
B - found in blank as well as sample
E - concentration exceeds calibration range of instrument
Instrument – Lowest amount of substance that can be detected by the instrument.; does not consider any other effects such as sample handling, matrix, preparation.
Method – Takes into account the reagents, sample matrix, and preparation steps applied to a sample in specific analytical methods.
CRDL – Are not necessarily the lowest detectable levels achievable, but rather are levels that a CLP laboratory would routinely and reliably detect and quantitate in a variety of sample matrices.
SQL – Takes into account sample characteristics, sample preparation, and analytical preparation (dilution, use of a smaller aliquot) which would affect the quantitation limit.
Your key

36. Common QA/QC Terms (continued)
Data qualifiers
D – identified in an analysis at a secondary dilution factor
N - spiked sample recovery not within control limits
R – data not usable, rejected
i – interferant: conflicting results reported by instrument due to another compound
91 – if following a concentration is a “Q value” and indicates that the sample was a 91% match to the instrument’s library spectra
Instrument – Lowest amount of substance that can be detected by the instrument.; does not consider any other effects such as sample handling, matrix, preparation.
Method – Takes into account the reagents, sample matrix, and preparation steps applied to a sample in specific analytical methods.
CRDL – Are not necessarily the lowest detectable levels achievable, but rather are levels that a CLP laboratory would routinely and reliably detect and quantitate in a variety of sample matrices.
SQL – Takes into account sample characteristics, sample preparation, and analytical preparation (dilution, use of a smaller aliquot) which would affect the quantitation limit.
Your key

37. Common QA/QC Terms (continued)
Data qualifiers
H – EPA data qualifier indicating high bias
L – EPA data qualifier indicating low bias
K – EPA data qualifier representing an unknown bias
Q – EPA data qualifier indicating that the result is estimated because the concentration is below the CRQLs
T – indicates that the value was summed from other constituents by software. Result was not present in original laboratory results
CC – sample identification number indicating that a duplicate sample was also collected at this location and this data was not present in the original lab results. Only conservative results were selected to be presented.
Instrument – Lowest amount of substance that can be detected by the instrument.; does not consider any other effects such as sample handling, matrix, preparation.
Method – Takes into account the reagents, sample matrix, and preparation steps applied to a sample in specific analytical methods.
CRDL – Are not necessarily the lowest detectable levels achievable, but rather are levels that a CLP laboratory would routinely and reliably detect and quantitate in a variety of sample matrices.
SQL – Takes into account sample characteristics, sample preparation, and analytical preparation (dilution, use of a smaller aliquot) which would affect the quantitation limit.
Your key

38. R U SQL i ND N E B D QL DL MDL IDL Q 91
Remember this? R
U SQL
i ND N E
B D QL DL
MDL IDL Q 91
Now you know what it means

39. So what does this crab data tell you?
Lead
240
Mercury U
Zinc
2400 B
Sodium
59000
Arsenic
400
Calcium
5800
Chromium
5.5 j
DL = 10 ppm

40. Where to look for the QA/QC data… the Data Summary Report
Compiled by the laboratory
Contains
Narrative
Glossary
Data summary
Data sheets
Support information

41. Failure to Meet QA/QC Criteria
Disclaimer
Acknowledge
Use of possibly unreliable data
Possibility of an inaccurate conclusion

42. QA/QC Sampling Terms Data qualifiers for field sampling
QA1 = screening
(monitoring instrument or sampling with little QA)
QA2 = screening + a little proof
(monitoring/sampling with blanks, spikes, and duplicates)
QA3 = screening + a lot of proof
(QA2 + splits, checks, and more duplicates)
Instrument – Lowest amount of substance that can be detected by the instrument.; does not consider any other effects such as sample handling, matrix, preparation.
Method – Takes into account the reagents, sample matrix, and preparation steps applied to a sample in specific analytical methods.
CRDL – Are not necessarily the lowest detectable levels achievable, but rather are levels that a CLP laboratory would routinely and reliably detect and quantitate in a variety of sample matrices.
SQL – Takes into account sample characteristics, sample preparation, and analytical preparation (dilution, use of a smaller aliquot) which would affect the quantitation limit.
Your key

43. Frequent Mistakes with Quality Issues
Misuse of field screening data
QA1, QA2, QA3
Concluding that a health concern does/does not exist on the basis of
Incomplete data (A)
Data that are not representative (R)
Poor quality data due to improper sample collection or analysis (Q)

44. What makes Adequate, Representative, and Quality Samples?
Do samples represent actual conditions? (R)
Are data available for critical sampling points? (A&R)
Are the quantity and quality of the data adequate to draw conclusions about those most likely to be exposed? (Q)

45. Critical Samples Source (ARQ) Environmental media (A&R)
Time periods (A&R) Exposure point (R)
Transport ID (R) Background (RQ)

46. Sources of Error Sampling design
Sampling methodology and sample handling procedures
Sample heterogeneity
Analytical procedures

47. A Detective Must Evaluate
Health data requirements
Field data quality
Laboratory data quality
Media-specific considerations

48. Do you feel qualified to be a detective? Did you to learn…
How to get the data quality you need?
How to interpret the data you get?
ARQ

49. That’s all folks! Credits Z ® Productions no rights reserved
Lynn Wilder’s HA QA course 2001
Brim Hall EI 2002
NYC 9-11 Response 2001
Vieques EI 2001
Indian Wells EI 2001
Salisbury EI 2001
Rubbertown EI 2000
WTI Strike Response 2000
Z’s X-mas 2000
Jared Burchette 2000
Clarkston EI 2000
Trinity NC DENR 1995
Warner Bros 1972
Z ® Productions no rights reserved